Albrecht Dürer, The Whore of Babylon, from The Apocalypse, 1498. She holds out “a golden cup full of abominations” (Rev 17:4). National Gallery of Art, via Wikimedia Commons (CC0).

The Greek word behind “pharmacy” meant remedy and poison at the same time. Revelation used it for the sorcery by which Babylon’s merchants deceived the nations. A 2018 experiment shows why the word may be more accurate than we thought: the drug we give for a “viral” fever can, at a toxic dose, make virus-sized particles that pass the injury on.

In 1793, as yellow fever swept through Philadelphia, America’s most famous physician treated his patients by bleeding them and dosing them with calomel, a mercury compound. Benjamin Rush’s methods “often were ineffective and actually brought many patients closer to their deathbeds” (Wikipedia). Some patients were “bled to faintness,” losing as much as 1.5 liters of blood (Dialogue).

Two centuries later, a quote attributed to Rush circulates in the health-freedom movement. It warns that medicine will “organize into an undercover dictatorship.” I used it myself years ago. He never wrote it. The University of Pennsylvania’s own guide to Rush mythology says plainly that it is bogus (Penn Libraries).

I start with that correction for two reasons. First, the argument that follows deserves better than an invented quote. Second, the real Rush is the better story: the man later imagined as a prophet of medical freedom was in fact the leading American practitioner of what came to be called “allopathy,” the medicine of opposing and suppressing. The myth has covered up the practice. As I will argue, that is what pharmakeia means.

A word that means both

The Greek phármakon means drug, remedy, poison, and charm, all at once. It is very old. It is already found in Mycenaean Greek, and the leading etymological dictionary of Greek classes it as “Pre-Greek,” borrowed from a language older than Greek itself (Wiktionary; Hypotheses). The classicist Jan Bremmer puts it simply: “both poison and drug were originally magical; so a pharmakon is a magical dose.” He notes that Greek dosis means “gift,” while German Gift means “poison” (Harvard Center for Hellenic Studies).

The word for “sorcery” grows from the same root. Pharmakeia has three layered meanings in the standard lexicon: (a) administering drugs, (b) poisoning, and (c) sorcery and magical arts, “often found in connection with idolatry” (Thayer’s Greek Lexicon). Plato’s Laws separates two kinds of pharmakeia: one harms the body through substances, and the other works on the mind through spells (Zichi, Lund University).

Keep both of Plato’s categories in mind. They will matter later.

Virus means poison

The vocabulary of infectious disease comes from the same place.

• Virus is Latin for “poison, poisonous liquid, sap of plants, slimy liquid, a potent juice.” It entered English in the late 14th century meaning “poisonous substance” (Etymonline). It is cognate with Sanskrit viṣá- and Greek iós, all meaning “poison, venom” (Dictionary.com).

• Virulent comes from Latin vīrulentus, “full of poison” (Etymonline).

• Poison comes from Latin pōtiō, “a drink,” especially “a medical drink,” and later “a (magic) potion” (Etymonline).

• Toxic comes from Greek tóxon, “bow.” The toxikon was the poison smeared on arrows (PMC).

So a virus is a poison, a poison is a medical drink, and a toxin is something shot into you from outside. The idea of the “invisible enemy” was in the words long before anyone had a microscope.

The oldest version of that idea is the opening of Western literature. In Book 1 of the Iliad, Apollo comes down and shoots plague into the Greek camp. The Greek word for his arrow, iós, is spelled the same as the word for poison (Liddell–Scott–Jones). Apollo was also the god of the Thargelia, the Athenian festival at which a pharmakós, a human scapegoat, was “feasted, led round the town, beaten with green twigs, and driven out or killed with stones” (Britannica). Plague, arrow, poison, and scapegoat all belong to one god.

Technical note. The “arrow” and “poison” senses of iós are homonyms with different roots. The lexicon lists them as separate entries. This is a pun the Greeks would have heard, not a shared origin. Bremmer calls the link between pharmakós(scapegoat) and phármakon (drug) “probable, but problematic.”

When poison turned into a germ

“Virus” meant poison until very recently. In Edward Jenner’s 1798 smallpox report the word was “not precise” and meant infectious matter (PMC). His cowpox lymph, which he called variolae vaccinae, “smallpox of the cow” (PMC), was a poisonous fluid given deliberately as protection. The first vaccine was a phármakon in the most literal sense.

The split came in 1898. The Dutch microbiologist Martinus Beijerinck called the filterable agent of tobacco mosaic disease a contagium vivum fluidum, a “contagious living fluid.” His conclusion “established a distinction between these agents and toxins” (EBSCO). From then on, a virus was by definition not a poison. In 1937 Thomas Rivers argued that Koch’s postulates, the standard proof that a germ causes a disease, could not be applied unchanged to viruses (Journal of Bacteriology). The first electron micrographs of a virus were published in 1939 (PMC).

For about forty years the concept came before anyone saw the particle. The field kept the poison name.

The key experiment

This is the center of the argument, and the reason the old word is more than a metaphor.

In 2018, Scientific Reports published “Exogenous exosomes from mice with acetaminophen-induced liver injury promote toxicity in the recipient hepatocytes and mice” (Scientific Reports). Here is what the researchers did:

• They gave mice a toxic dose of acetaminophen (Tylenol): 300 mg/kg.

• The injured livers released exosomes, tiny membrane-bound vesicles that cells use to send messages. The exosomes averaged about 92 nanometers across, the size range of many viruses. Mice given acetaminophen had roughly three times as many as controls.

• The researchers isolated these exosomes and gave them to healthy liver cells and to healthy mice that never received the drug.

• The recipients showed oxidative stress, the inflammatory signals TNF-α and IL-1β, activation of the cell’s self-destruct pathway (JNK, Bax, caspases 3 and 9), and “markedly elevated hepatocyte apoptosis,” meaning liver-cell death.

A separate study found that circulating vesicle counts also rise after acetaminophen overdose “in mice and humans” (PMC). Another found that exosomes from ordinary human plasma, carrying the enzyme CYP2E1, made acetaminophen toxicity in recipient cells worse (Scientific Reports, 2019).

Now look at what acetaminophen is for. It is the standard treatment for the fever and aches of influenza-like illness. A recent expert consensus recommended it unanimously (International Journal of Clinical Practice).

Put the pieces together. A fever is diagnosed as “viral.” A phármakon is given. At a toxic dose, that remedy injures cells, and the injured cells release virus-sized particles that pass the injury to tissue the drug never touched. If the patient gets worse, or the illness seems to spread through the body, the cause on record is “the virus.” A test that detects particles or genetic material was not designed to tell the difference.

This is the remedy and the poison in a single molecule, and a symptom-suppressing treatment producing a phenomenon that looks like the illness it was given for. I first wrote about this study in Reframing Viral Mechanisms, and it remains the foundation of the Poisoned, Not Infected series.

What this study shows, and what it doesn’t. It shows that vesicles released by chemically injured cells can carry an injury program to cells the toxicant never touched. It used an overdose. It has not been shown that ordinary doses produce injury-carrying vesicles; the case for ordinary doses rests on a different, well-established mechanism, glutathione depletion. In the recipient mice, the standard liver-damage blood marker (ALT) did not change at four hours. The documented effects were oxidative stress, cytokines, and cell death. Exosomes are not produced only after toxicity: healthy mice make them too. What the drug changed was how many there were and what they carried. That is actually the more striking finding: the body’s own ordinary messengers were reprogrammed into carriers of injury. Finally, being the same size as a virus is not the same as being a virus. Mainstream researchers who note the overlap, such as the Trojan exosome hypothesis and Nolte-’t Hoen’s review, treat retroviruses as real viruses that share machinery with exosomes.

Five layers, one ancient word

The Poisoned, Not Infected framework separates five things that the single word “infection” lumps together. The Greek phármakon covers all five.

1. The initiating agent. “A virus, bacterium, toxicant, drug, pollutant, contaminated product... or combination.” Note that drug and virus sit next to each other on that list. As I wrote in May, “pharmacology is a subset of toxicology, separated only by intent and dose” (How Being Poisoned Can Be Mistaken for Being Infected). Some viruses carry poisons in the most literal sense: the genes for diphtheria, cholera, and Shiga toxins are carried by bacteriophages, viruses that infect bacteria (StatPearls).

2. The receiving terrain. Dosis means gift; Gift means poison. Whether a substance heals or harms depends on the body receiving it. Melinda Beck’s research showed that in selenium-deficient hosts, oxidative stress drives “rapidly mutated benign variants of RNA viruses to virulence,” and that these mutations are “consistent, rather than stochastic” (PMC). The terrain made the virus virulent, literally “full of poison.”

3. The accumulated exposome. This is the lifetime total of chemical, pharmaceutical, dietary, and environmental exposures, a term coined by Christopher Wild in 2005 (PubMed). Read Revelation 18:11–13 as a cargo manifest: gold, cinnamon, incense, ointments, wine, oil, grain, cattle, “and slaves, and souls of men” (Rev 18:13). It is the inventory of an empire’s exposome, and it ends with the human being as a commodity.

4. The mediating signals. Cytokines, damage signals, and exosomes: the messengers in the experiment above. The “viral” is not only foreign. About 8% of the human genome is endogenous retrovirus (PMC), and syncytin, a gene captured from an ancient retrovirus, is essential for forming the placenta (PNAS). At the molecular level the virus is a full phármakon: in some settings poison, in others part of what makes mammalian birth possible.

5. The interpreted story. This is Plato’s second pharmakeia, the one that works on the mind. Our tools, as I wrote in May, “are pathogen-shaped,” and “an instrument built to find an organism will tend to return an organism or a blank, not a chemical” (Ji). The single-germ story repeats the scapegoat rite. Every shared cause, including nutrition, toxicants, poverty, and the treatment itself, is loaded onto one invader that can be named, sequenced, and driven out. Thomas Szasz made the link in 1974 in Ceremonial Chemistry, which opens with “The Scapegoat as Drug and the Drug as Scapegoat” (Syracuse University Press).

By thy sorceries were all nations deceived

The word pharmakeia appears only three times in the New Testament: Galatians 5:20, Revelation 9:21, and Revelation 18:23 (WELS). The last one reads, in the King James Version: “for thy merchants were the great men of the earth; for by thy sorceries were all nations deceived” (KJV). In the Greek, the merchants are the megistanes, the “grandees,” and the nations were eplanēthēsan, “led astray” (Bible Hub).

In earlier versions of this argument I quoted a paraphrase that put “end of times” into the verse. The actual text is stronger. In one sentence it brings together a commercial elite that has become a ruling class, pharmakeia as its method, and the deception of every nation.

Behind it lies a pattern from the Septuagint, the Greek Old Testament. Pharaoh’s pharmakoi imitate Moses’ signs (Exod 7:11). Nineveh is “the mistress of witchcrafts, that selleth nations through her whoredoms, and families through her witchcrafts” (Nah 3:4). Babylon falls “in one day” despite “many sorceries” (Isa 47:9). In each case, pharmakeia is how an empire enchants and sells nations.

Revelation also connects the potion to the plague. “Come out of her, my people, that ye be not partakers of her sins, and that ye receive not of her plagues” (Rev 18:4). After the trumpet plagues, the survivors “repented not... of their sorceries” (Rev 9:21), and here some manuscripts read pharmakōn, “drugs” (Thayer’s). The plagues fall on the empire whose potions deceive the nations.

The counter-argument, stated fairly. Mainstream biblical scholars argue that pharmakeia in Revelation “refers to sorcery... not pharmaceutical drugs” (Biblical Research Institute), and that Babylon meant first-century Rome. I agree that Revelation did not predict the pharmaceutical industry. My claim is about a pattern, not a prediction. The ancient word never separated substance from spell. Our clean distinction between “pharmaceutical” and “magical” is newer than the word, and every time the Bible uses it, it describes a commercial empire administering potions and deceiving nations.

What I am not saying

I am not saying viruses don’t exist, although I have a few colleagues who have come that conclusion, or pretty close, and I respect their views. They have been photographed since 1939, and the findings above (phage-carried toxins, endogenous retroviruses, syncytin) depend on viral genomes being real. I am not saying every illness is poisoning, or that toxic outbreaks are always missed. The investigations of EVALI and the 1989 L-tryptophan epidemic show the system can work (Ji; CDC). I am not repeating Pasteur’s supposed deathbed line, “the microbe is nothing, the terrain is everything,” because no one has found a verifiable source for it (Factually). And I am not telling anyone to withhold care from a seriously ill child. High or persistent fever, fever in an infant, lethargy, dehydration, or difficulty breathing are reasons to engage professional medical assistance.

My claim is narrower, and harder to dismiss. The ancient word for medicine never let us forget that every remedy is also a poison, and that whether it helps or harms depends on the dose, the terrain, the exposome, the body’s own signals, and the story told about it. Modern medicine divided that word into “drug” (good, given) and “germ” (evil, invading), then built instruments that can see only the second. A 2018 experiment shows what that blind spot can hide: a remedy that, at a toxic dose, produces particles that look like the illness it treats.

“Come out of her” is not a call to fear. It asks us to read every phármakon, drug or germ, with all five layers in view.

Leave a comment

Share

Subscribe now

Ten head-to-head clinical trials — pulled from the GreenMedInfo archive — in which honey, turmeric, saffron, ginger and a handful of other kitchen staples matched or outperformed prescription medications. Each one links to the full research write-up.

For nearly two decades, GreenMedInfo has done one thing most health publishers never bother with: we go into the National Library of Medicine and pull out the studies where researchers put a natural substance directly against a drug — same condition, same patients, same scoring — and then publish what they found.

Those studies exist. There are thousands of them. They just don’t get press releases, because nobody holds a patent on honey.

Below are ten of the most striking. Every one is a human trial. Every one is indexed in our database. And every one links to the long-form article where you can read the methodology, the numbers and the citations for yourself.

A word before we start: this is research information, not medical advice. Several of these trials are small. Some are single studies that have not been replicated. I have flagged which ones have the strongest designs. Read them, bring them to your practitioner, and decide for yourself. That is the entire point of having the archive.

1. Honey + Coffee vs. Prednisolone — Persistent Cough

The trial: Randomized, double-blind, 97 adults with post-infectious cough lasting more than three weeks. One group got a paste of honey and instant coffee every eight hours. One got the steroid prednisolone. One got guaifenesin (the active ingredient in Mucinex). Primary Care Respiratory Journal, 2013.

The result: Cough scores in the honey-coffee group fell from 2.9 to 0.2 — a 93% reduction. Prednisolone managed 3.0 to 2.4, about 20%. Guaifenesin did essentially nothing. The study authors’ own conclusion: “Honey plus coffee was found to be the most effective treatment modality.”

A kitchen remedy outperformed a corticosteroid by a factor of roughly fourteen — with none of the steroid’s immune-suppressing, bone-thinning, blood-sugar-wrecking side effects.

→ Read the full article: A Spoonful of Honey and Coffee Did What Prednisone Couldn’t

2. Curcumin vs. Metformin — Stopping Prediabetes Cold

The trial: Randomized controlled trial in Thailand, published in the American Diabetes Association’s own journal, Diabetes Care, 2012. Prediabetic patients took either a curcumin extract (from turmeric) or placebo for nine months.

The result: Zero patients in the curcumin group progressed to type 2 diabetes. In the placebo group, 16.4% did. Curcumin also improved beta-cell function and reduced insulin resistance, with only minor side effects reported.

For context: metformin, the front-line drug given to high-risk patients, reduced progression to diabetes by just 31% over three years in the landmark Diabetes Prevention Program. No pharmaceutical has ever come close to a 100% prevention rate.

→ Read the full article: 100% Diabetes Prevention Achieved Naturally

3. Berberine vs. Metformin — Type 2 Diabetes

The trial: 36 adults with newly diagnosed type 2 diabetes, three months, berberine versus metformin head-to-head. Metabolism, 2008.

The result: Berberine — a bitter alkaloid from goldenseal, barberry and Oregon grape — matched metformin on A1c, fasting glucose and post-meal glucose. Then it pulled ahead on lipids: by week 13, triglycerides and total cholesterol were significantly lower in the berberine group. In a second arm, 48 poorly-controlled diabetics on berberine saw A1c drop from 8.1% to 7.3% and fasting insulin fall 28%.

A later meta-analysis of 37 studies and 3,048 patients found berberine did not increase adverse events or the risk of hypoglycemia. Metformin, by contrast, carries a black-box risk of lactic acidosis.

→ Read the full article: Berberine Outshines Metformin for Diabetes

4. Saffron vs. Prozac and Lexapro — Depression, OCD, Postpartum Depression

The trials: Six separate head-to-head studies. In a 2016 double-blind trial, 30 mg/day of saffron matched 40 mg of fluoxetine for postpartum depression over six weeks. A 60-patient, 8-week trial found saffron equal to fluoxetine for OCD, with “many fewer” adverse effects. Additional trials found no difference against 60 mg fluoxetine for major depression and panic disorder.

The standout: A 2020 study of 54 patients found 15 mg of saffron produced significantly greater reductions in Hamilton Depression scores — and higher response and remission rates — than 10 mg of escitalopram (Lexapro).

A 2025 meta-analysis of eight RCTs confirmed it: saffron is as effective as SSRIs for depression, with fewer adverse events. The spice in your paella is running even with — and sometimes ahead of — the most prescribed class of psychiatric drugs on earth.

→ Read the full article: 6 Studies on Saffron Put Prozac to Shame

5. Honey vs. Dextromethorphan — Children’s Nighttime Cough

The trial: Children with upper respiratory infections were given buckwheat honey, honey-flavored dextromethorphan (the “DM” in Robitussin DM, Mucinex DM, Vicks Formula 44), or nothing before bed. Parents rated cough frequency, severity and sleep quality. Archives of Pediatrics & Adolescent Medicine, 2007.

The result: Honey scored best on every measure. Parents rated cough frequency 1.89 points better after honey, 1.39 after DM, and 0.92 with nothing. The quietly damning finding: dextromethorphan was not statistically better than no treatment on any outcome.

Even the CDC now advises honey for cough in anyone over one year old. The multi-billion-dollar cough syrup aisle has still not gotten the memo.

→ Read the full article: 5 Reasons Honey Should Be in Your Medicine Chest · Study abstract

6. Turmeric vs. Ibuprofen and Naproxen — Knee Osteoarthritis

The trials: A Thai study gave 367 knee osteoarthritis patients either 1,500 mg/day of turmeric extract or 1,200 mg/day of ibuprofen for four weeks. Separately, a 2022 double-blind RCT in Phytotherapy Research gave 60 patients either 500 mg naproxen twice daily or a capsule of turmeric extract, gingerols and piperine twice daily for a month.

The result: Turmeric matched ibuprofen for pain relief and function, with 96–97% patient satisfaction in both groups — but fewer gastrointestinal side effects. The turmeric-ginger-pepper formula relieved pain and inflammation on par with naproxen, with patients rating it equal or better on tolerability, and no adverse effects.

NSAIDs are linked to thousands of GI-bleed deaths a year. Turmeric has a 4,000-year safety record and over 900 documented health benefits in our database.

→ Read the full article: Spice-Based Relief for Aching Joints Beats Dangerous NSAID

7. Ginger vs. Mefenamic Acid and Ibuprofen — Menstrual Pain

The trials: 122 women with moderate-to-severe dysmenorrhea took either 250 mg of ginger every six hours or the NSAID mefenamic acid every eight hours, across two menstrual cycles. An earlier trial of 150 women compared 250 mg ginger four times daily against both mefenamic acid and ibuprofen for three days.

The result: Ginger reduced pain “significantly and equally” to the drug in the first trial — roughly half the women in each group went from moderate/severe to mild. In the second, the authors concluded flatly that ginger is as effective as the drugs.

The difference is what comes with it: NSAIDs raise the risk of ulcers, bleeding and cardiovascular events. Ginger raises the risk of nothing in particular.

→ Read the full article: Herbal Research Update for Women’s Health · Eight Unexpected Health Benefits of Ginger

8. Echinacea + Elderberry vs. Tamiflu — Influenza

The trial: Double-blind comparison of an Echinacea purpurea and elderberry blend against oseltamivir (Tamiflu), the world’s best-selling flu drug. Current Therapeutic Research, 2015.

The result: By day five, the herbal blend was already slightly ahead. By day ten, 90.1% of the herb group had recovered versus 84.8% on Tamiflu — with fewer complications and fewer side effects. Elderberry alone has repeatedly been shown to cut flu recovery from 7–8 days to 2–3.

Governments stockpiled billions of dollars of Tamiflu. A Cochrane review later found it shortened flu symptoms by about half a day. Elderberry costs twelve dollars.

→ Read the full article: What You Need to Know to Crush the Cold & Flu

9. Licorice (DGL) vs. Tagamet — Peptic Ulcers

The trial: Deglycyrrhizinated licorice (DGL) at 760 mg three times daily versus cimetidine (Tagamet), the first blockbuster acid-blocker, for peptic ulcer healing and relapse.

The result: DGL achieved 91% ulcer healing with an 8.2% relapse rate, versus 12.9% relapse with Tagamet. In the same research roundup, 3,000 mg/day of turmeric achieved 72% ulcer healing at eight weeks — comparable to omeprazole (Prilosec).

Acid-blockers are now linked to dementia, kidney disease, fractures and nutrient malabsorption with long-term use. Licorice root has been treating stomachs since Hippocrates.

→ Read the full article: The Mind Thieves — How 6 Common Medications Are Stealing Our Humanity

10. Colostrum vs. the Flu Shot — Influenza Prevention

The trial: Head-to-head comparison of oral colostrum against flu vaccination in both healthy adults and high-risk cardiovascular patients over two months. Clinical and Applied Thrombosis/Hemostasis, 2007.

The result: Healthy adults on colostrum alone averaged 0.33 flu episodes versus 1.1 in the vaccinated-only group — roughly three times more protective. Among the high-risk cardiac patients, the colostrum group had zero hospital admissions; the vaccinated group had multiple hospitalizations and one death. Flu-related costs in the colostrum group were about 30% of the vaccinated group’s. No significant side effects were reported.

Vitamin D, for the record, cut influenza A risk by 59% in children at 1,200 IU/day in a separate RCT. Neither appears in a single public health campaign.

→ Read the full article: Colostrum 3x More Effective Than Flu Vaccines

Honorable Mentions: When “Doing Something Else” Beat the Drug

Not every alternative comes in a capsule. From our long-running piece on non-pill interventions that outperformed pharmaceuticals in head-to-head trials:

• Sudarshan Kriya yoga breathing beat the antidepressant imipramine for depression (Journal of Affective Disorders, 2000).

• Acupuncture beat flunarizine for migraine prevention, with superior tolerability.

• Abdominal aromatherapy massage beat Tylenol for menstrual pain.

• Hypnosis matched Valium for anxiety during embryo transfer, with fewer side effects.

• Blue and red light therapy beat 5% benzoyl peroxide for acne, with no side effects.

• An elemental diet matched high-dose steroids for Crohn’s disease in children — and was superior for supporting their growth.

→ Read the full list: 11 Things Better than Drugs or Supplements for Healing

Why You Have Never Heard Most of This

Every study above is published in a peer-reviewed journal and sits in the National Library of Medicine. None of it is hidden. It is simply unfindable — buried in a haystack of tens of millions of papers, with no one paid to surface it, and, since 2021, no longer even filterable under PubMed’s complementary-and-alternative-medicine category, which was quietly removed.

That is why GreenMedInfo exists. We have spent seventeen years indexing this research — 1.4 million studies and counting, hand-mapped to over 16,000 substances, conditions and mechanisms — and keeping it free to search. In 2020 they tried to take it offline. It’s still here. It’s still yours.

If any of this was new to you, that is the whole problem — and the whole reason to search the archive, share this piece, and bring one of these studies to your next doctor’s appointment.

Leave a comment

Subscribe now

This article is for research and informational purposes only. It is not medical advice, diagnosis or treatment. Several trials cited are small or unreplicated; consult a qualified practitioner before changing any medication. Honey should never be given to children under one year of age.

A 2010 plan to watch mothers who asked questions about vaccines. A 2012 forward to Jeffrey Epstein asking for his idea about those mothers. A label that followed me for a decade. All of it is in the record now.

Picture a mother at a kitchen table in 2010. Her son has a fever after his shots, and she types a question into a search box: Is this normal?

That same year, in a report commissioned by the Bill & Melinda Gates Foundation, consultants proposed a system to catch her question. They called it a “vaccine rumor surveillance system.” It appears on page 17. On the same page, the same report concedes “real and suspected adverse events following immunization.”

Real. Their word. The question the mother asked was legitimate, and they knew it. The system was for her anyway.

I know this because the report is in the Epstein files.

She was not the first

The mother at that table in 2010 had a long line of women behind her.

In the 1950s and 60s, the leading theory of autism blamed the mother. “Refrigerator mothers,” the experts called them: cold women who had frozen their children into silence. It took decades to retire that idea, and no one apologized for it.

Their daughters became the mothers of the 1990s and 2000s who watched a talking, laughing child go quiet after a round of shots, said so, and were told they had imagined the child they remembered. They were called hysterical, then anti-science, then dangerous. The one thing they were almost never called was witnesses.

The system they reported to was built not to hear them. In 1986, Congress shielded vaccine makers from most lawsuits and created a no-fault compensation program in their place; in 2011 the Supreme Court confirmed that the courthouse door stays shut. The program’s own reporting system, VAERS, is voluntary, and a federally funded Harvard study in 2010 found it captures fewer than one percent of adverse events. In 2008 the government quietly conceded one child’s case, Hannah Poling’s, agreeing that vaccines had aggravated an underlying disorder and produced “features of autism.” The next year, in a single omnibus proceeding, it turned away roughly five thousand families at once.

So when a mother said, “I saw my child regress,” here is what she had: no lawsuit, a reporting system that recorded almost nothing, a compensation program that had admitted one case and shut the door on thousands, and a public conversation that treated her memory as misinformation.

That is the history on page 17 of the Gates Foundation’s 2010 report. Its authors conceded that some adverse events after immunization are “real.” They proposed watching the mothers anyway.

The forward

On February 3, 2012, an email from inside the Foundation’s world was forwarded to jeevacation@gmail.com, Jeffrey Epstein’s private address. It carried the Foundation’s internal immunization strategy and that 2010 report. The note on top read: “Can we discuss this over the weekend?”

Then it got to the point. The sender referred to an immunization idea credited to Epstein himself, “your idea about Mother’s Army,” and closed: “Check out the slides. See if you think it is appropriate. If you do, I will make the proposal and the deck.”

Read that again. More than three years after Epstein pleaded guilty to procuring a minor, someone with the Foundation’s vaccine strategy in hand was asking whether his idea about mothers was appropriate enough to pitch.

The sender’s name is redacted. The Justice Department knows it. We don’t.

What they built, in order

I was one of the mothers’ questions. Not literally, but in the system’s logic I was worse: a researcher who gathered the peer-reviewed studies those questions pointed to and made them free to read. Here is what happened next, from the record, in the order it happened.

• 2012. The Foundation funded a university researcher to study online “anti-vaccine” messaging. I reported it that August.

• 2013. A UNICEF working paper tracking “anti-vaccination sentiment” listed GreenMedInfo among the sites it monitored. The surveillance system had a list now, and I was on it.

• 2018 to 2019. GreenMedInfo’s search visibility fell 81 percent. In June 2019 our email provider shut us down without notice.

• May 2020. A British think tank, with the BBC, published a list of 34 websites. We were sixth.

• March 2021. The Center for Countering Digital Hate named twelve people its “Disinformation Dozen.” I was one. Robert F. Kennedy Jr. was another. That July the President of the United States said of us: “It’s killing people.” The platforms did the rest. In August, Facebook itself said the twelve accounted for a tiny fraction of vaccine content. It didn’t matter.

• October 2024. CCDH’s internal minutes leaked: plans for “roundtables” with a vaccine alliance so an issue that had “fallen off the radar” could be brought back, and a line attributed to its chief executive about “black ops being set up” around Kennedy.

• February 2026. CCDH’s founder resigned as the British Prime Minister’s chief of staff.

I covered the entire arc of this story in my CHD in DC presentation on Constitution Day, Sept 17th. You can watch it here.

Fourteen years. A plan, a grant, a list, a label, a presidential sentence, and a leak. I used to think I was imagining the machine. The documents say I wasn’t.

I will not claim more than they show. Nothing here proves the Foundation directed the list or the label. That restraint is not weakness; it is the whole point of the book, and it is the one thing the people who labeled us never extended to us.

The man they asked

Now the part that should stop every natural health advocate cold.

The institution whose consultants proposed watching mothers was, in the same years, in correspondence with a convicted sex offender about its money. In 2011 Epstein pitched JPMorgan a fund to capture the “more than 60 billion” America’s billionaires had pledged to give away. “IT will be the largest foundation in the world,” he wrote. (EFTA00904739)

And on July 9, 2017, he wrote to Bill Gates proposing that philanthropists be “orgazanized to decide a shadow governmental grants structure.”

The typo is his. So is the word. The record shows no reply.

Think about what a shadow governmental grants structure decides. Which research is funded. Which health programs run where. Which questions count as science and which count as “rumors.” Governments make those decisions in public, under oath, subject to a vote. Epstein’s proposal was to make them privately, by a group organized, in his words, “to blunt” the “bad pr risk.”

We have been arguing with that structure for fifteen years. We just couldn’t see who was in the room. Now we can. One of them was Jeffrey Epstein, and he was being asked for his ideas about mothers.

Go deeper, with the documents

From the Gates Foundation to Epstein’s Inbox to the Disinformation Dozen
Fourteen years of documents, read in the order they were written. The plan, the money, the list, and the label.

“We May Need to Do a Bit of Damage Control”: Inside NIAID’s Ebola Vaccine Crisis
What happened inside a federal institute, from Senate records, when its own vaccine experiment produced lung damage in the animals it was meant to protect.

Why this book

Shadow Empire: The Epstein Files is not a book about vaccines. It is a book about an architecture: a fund, an offshore option, a private back channel through a single Gmail inbox, a science program that reached from the human genome to a budget for a “designer baby and human cloning company,” and the institutions that kept answering Epstein’s email for a decade after his conviction.

But it was written by one of the twelve. And it was written for everyone who was ever told that their question was the danger.

Every claim the case rests on carries a footnote to a document you can pull from justice.gov/epstein. Each chapter says what it establishes and what it does not. Every correction will be logged, with its date, at shadowempirebook.com/corrections.

Your turn

For fifteen years they told you not to look. So look.

Go to justice.gov/epstein. Type EFTA00702570. Read the note about “Mother’s Army.” Then type EFTA01041100 and read the word “shadow” in Epstein’s own hand.

Then decide who was spreading rumors.

If you want to go further, I made a free guide: Start Here: 12 Documents in the Epstein Files, and How to Read Them. Each entry has the original page, what it shows, what it doesn’t prove, and a link to check it yourself.

Get the free guide →

Pre-order

Shadow Empire: The Epstein Files is published this winter by Skyhorse Publishing with Children’s Health Defense.

Pre-order at Amazon · Pre-order at Barnes & Noble

Read why it’s different: shadowempirebook.com/why-this-book

Peptides sound like nature. The ones sold in vials are built from industrial amino acids, coal-tar chemistry and a solvent Europe flags for reproductive harm. Here is the documented supply chain, and why the information they carry is not the same.

Share the X post dedicated to this article.

There is a word doing a great deal of work in the wellness world right now: peptide.

It sounds gentle. It sounds biological, like something your body already makes, something you might find in bone broth or a sprouting seed. Influencers lean into that. One of the best-known peptide sellers in the MAHA orbit told his followers, “Unlike synthetic pharmaceuticals, peptides are naturally recognized by your body” (CNN/AP).

Here is the problem. The word “peptide” names a natural class of molecules. Over 7,000 naturally occurring peptides had been identified by 2014, and your digestive system releases peptides from food every day (Nutrition Reviews). The products in the vials, BPC-157, TB-500, KPV, Semax, Epitalon, come from somewhere else entirely.

They come from a chemical plant.

For years I have argued that food is information. Read through the lens of that model, the documented record behind these vials matters a great deal. What follows traces it step by step, using the FDA’s own briefing documents, a USDA technical review, the U.S. EPA, the American Chemical Society and the peer-reviewed literature.

Food Is Information. So Is a Vial.

The idea is simple. Food is not just calories and building blocks. As I have written, food is “biological information,” a signaling system “that speaks directly to our genes, our microbiome, our immune intelligence, and our metabolic destiny” (GreenMedInfo).

Information has two dimensions. There is quantity: how much of a signal arrives, the dose. And there is quality: how faithful, complete and coherent that signal is, and what comes with it. A nutrient made by a living organism “is not merely an isolated molecule. It exists within a complex of enzymes, coenzymes, minerals, peptides, flavonoids, pigments, and countless other constituents that developed together inside a living organism. That biological context matters” (Sayer Ji).

The synthetic peptide industry deals almost entirely in quantity. It copies a short sequence, purifies it, and sells it by the milligram. Everything else, including the context, the provenance and the biological quality control, is left behind at the factory gate.

Two Ways to Make a Peptide

When a living cell makes a peptide, it follows a process biology spent billions of years refining.

The gene is read out by the ribosome. A built-in “zip code” in the new protein tells the cell where to send it, the discovery that won Günter Blobel the 1999 Nobel Prize: “proteins have intrinsic signals that govern their transport and localization in the cell” (Nobel Prize). Inside the cell’s protein-processing compartment, molecular chaperones help the chain fold correctly. That discovery earned Franz-Ulrich Hartl and Arthur Horwich the 2011 Lasker Award (Lasker Foundation). Quality-control systems “comprised of chaperones, folding enzymes, and degradation factors” inspect the work and destroy what fails (ER quality control review). Many active peptides are then cut out of a larger, inactive precursor and chemically finished by specialized enzymes before release (prohormone processing).

In food, those peptides arrive packaged. Casein phosphopeptides formed when milk protein is digested act as mineral carriers (Dairy Science & Technology). Wheat-sprout peptides travel in phospholipid–peptide complexes, natural packaging the researchers suggested may help explain their bioavailability (Journal of Peptide Science).

A vial peptide goes through none of this.

Technical callout: keeping the contrast precise. A protein’s amino-acid sequence ultimately determines its folded shape (The Cell, NCBI Bookshelf). Chaperones help proteins reach that shape and prevent clumping; they do not add new structural instructions. Very short peptides like KPV (3 amino acids) or TB-500 (7) barely fold at all. So the documented difference for these molecules is not a “missing fold.” It is the absence of biological quality control, precursor processing and food context, and the presence of synthesis impurities, leftover acid and uncertain stereochemistry.

Step One: Where the Amino Acids Come From

Every synthetic peptide starts as a pile of individual amino acids. The industry produces them in three main ways.

Engineered bacteria. Most amino acids are now made in fermentation tanks. A review in Biotechnology Advances explains that “fermentation with the aid of strains such as Corynebacterium glutamicum or Escherichia coli“ plays “a significant role in the industrial production of amino acids,” driven by “advances made in the genetic engineering techniques” (Biotechnology Advances).

Petrochemicals. Some are built from industrial chemicals. According to a USDA technical evaluation, DL-methionine is “usually produced entirely by chemical methods.” The documented routes include reacting acrolein with methyl mercaptan, and a process using “propylene, hydrogen sulfide, methane, and ammonia to make the intermediates acrolein, methylthiol, and hydrocyanic acid.” Hydrocyanic acid is hydrogen cyanide. The USDA’s conclusion is blunt: “the L-methionine produced from it is also synthetic” (USDA, 2012). Methionine is the first amino acid in both Semax and MOTS-c (FDA; FDA). Glycine, which appears three times in BPC-157, is mostly made commercially by reacting chloroacetic acid with ammonia (ChemicalBook; FDA).

Hair and feathers. Cysteine has long been extracted from keratin. A 2024 review describes the standard method: cystine “is extracted with activated charcoal after the acidic hydrolysis of feathers and hair,” then converted “via electrolytic reduction.” It calls this “the most common production method applied so far” (Microbial Cell Factories).

Much of this supply chain runs through one country. The European feed-industry federation reports that “China is the only country producing the 5 most essential amino acids, with a global market share between 25 and 85%” (FEFAC, 2025).

Step Two: The Coal-Tar Cap

You cannot simply stir amino acids together and get BPC-157. They would react at the wrong places. So each one is first fitted with a chemical “cap” called a protecting group. The industry standard is Fmoc, and the FDA’s own descriptions of how BPC-157 and TB-500 are made use it throughout (FDA BPC-157 briefing; FDA TB-500 briefing).

Fmoc is built on a molecule called fluorene. The U.S. EPA notes that fluorene “is a component of petroleum” and “can be isolated from coal tar by distillation and recrystallization” (U.S. EPA).

So before a single peptide bond forms, each “natural” building block has been chemically bonded to a coal-tar or petroleum-derived scaffold.

Step Three: Assembly on Plastic Beads

The chain is then built one amino acid at a time on insoluble polymer beads, a method called solid-phase peptide synthesis. It is a brilliant piece of chemistry, and it won Bruce Merrifield the 1984 Nobel Prize (Nobel Prize). It is also chemistry, not biology.

The FDA’s July 2026 briefing describes the original 1993 synthesis of BPC-157 in detail. Protected amino acids were added stepwise to “a polymeric carrier (benzhydrilaminoresin)” using “diisopropylcarbodiimide as the coupling reagent,” and at each step “the Fmoc protective group was removed with piperidine” (FDA). For TB-500, the FDA describes couplings with “HBTU activator” and “DIPEA base in a solvent such as DMF,” with caps removed using “20% (v/v) piperidine in DMF” (FDA).

That solvent, DMF, deserves attention. A 2026 paper in Green Chemistry states that peptide synthesis “relies heavily on hazardous solvents, such as N,N′-dimethylformamide (DMF), whose substitution by green solvents is proving more challenging than expected” (Green Chemistry). The European Chemicals Agency lists DMF as a “substance of very high concern” because of “its reproductive toxicity” (PMC review). Its safety data sheet carries the hazard statement “May damage the unborn child” (Fisher Scientific SDS).

Technical callout: process is not product. Using DMF in manufacturing does not prove that a finished vial contains harmful residual DMF. Properly made pharmaceuticals are tested for residual solvents. The documented problem is that the testing is missing: the FDA found “no information about residual-solvent testing” for either form of BPC-157 in the nomination packages (FDA). For gray-market vials, the honest word is “untested,” not “contaminated.”

Step Four: Cut Free With Acid

When the chain is complete, it has to be released from the beads. The 1993 BPC-157 synthesis used “trifluoroacetic acid/trifluoromethanesulphonic acid/anisole,” followed by HPLC purification (FDA). A standard modern cleavage mix is 95% trifluoroacetic acid (Green Chemistry).

Some of that acid often stays in the product. Epitalon, for example, is sold mostly as a salt with “acetic acid or trifluoroacetic acid (TFA) counterions” (Epitalon review). Research on peptide counter-ions finds they can shift a peptide’s structure, and that the choice of counter-ion bears on toxicity risk (PMC review). Synthesis also leaves its own fingerprints. The FDA warns of “incomplete coupling reactions, truncations, or side reactions,” “isomeric impurities,” residual “solvents, coupling reagents, activators, catalysts, and scavengers,” and “peptide-related aggregates” (FDA). One recognized impurity is racemization: building blocks flipped into their mirror-image form (immunogenicity review).

13,000 to 1

There is one more number worth knowing. The American Chemical Society’s Green Chemistry Institute reports that solid-phase peptide synthesis has a “Process Mass Intensity (PMI) of about 13,000—significantly higher than other modalities” (ACS GCI). In plain terms, about 13,000 kilograms of solvents, reagents and other inputs go in for every kilogram of peptide that comes out. Purification “generates astronomically large volumes of waste relative to the product yield” (ACS GCI).

Compare a tray of wheatgrass on a windowsill. Its inputs are sunlight, water, soil and a seed, and its “waste” is compost.

Many Aren’t Natural Molecules at All

Set the factory aside for a moment. Many of the molecules themselves are not things Nature makes.

• BPC-157 is a 15-amino-acid fragment of a larger gastric protein, first synthesized in Zagreb in the early 1990s (FDA). It was developed by the Croatian drug company Pliva under the code PL-14736 (PMC review). The U.S. government’s drug database calls it “a small, chemically synthesized pentadecapeptide” (NCATS). It is not a component of an approved product in any country (FDA).

• TB-500 is a synthetic seven-amino-acid fragment of thymosin beta-4 with a chemical group added to one end (FDA).

• Semax, in the FDA’s words, is “not a naturally occurring sequence” (FDA).

• Epitalon is a synthetic tetrapeptide based on a cow pineal-gland extract. Its four amino acids allow eight mirror-image variants, and the FDA found “no clinical data to support safety in humans” (FDA; Epitalon review).

In other words, a great deal of the peptide gray market is selling fragments, analogs and shelved drug candidates as if they were nature’s own.

What Nature Actually Offers

None of this means peptides are bad. It means the source is the story.

Food peptides are real and well studied. Researchers have purified an iron-binding peptide from spirulina protein (spirulina study). Peptides from milk, soy, eggs and sea cucumber have been shown to bind minerals (PMC review). The wheat-sprout research I wrote about recently came from an Italian group that went looking for the thymus’s regulatory peptides and found the same class of molecule in a sprouting grass (Calzuola et al., Peptides).

Precision matters here too. Many food-derived peptide products are made by breaking down food protein with enzymes, so “food-derived” is accurate but “untouched” often is not (Nutrients review). And not every “chelate” is natural: ferrous glycinate, a common mineral chelate, is manufactured “by reaction of reduced iron with glycine” (FAO/WHO JECFA). The line that matters is whether the information was written by a living system or assembled by a reactor.

What This Means

I support adults’ right to make informed choices about their own bodies. That is exactly why the labeling matters. “Informed” means knowing that a vial of BPC-157 is several industrial steps removed from any living food: amino acids made by engineered bacteria or petrochemical reactors, capped with a coal-tar-derived group, assembled in a reprotoxic solvent and cut free with trifluoroacetic acid.

It also means holding synthetic peptides to the evidence standard we demand of every other synthetic chemical. The FDA’s own scientists recommended against all seven peptides reviewed this summer, and four of the seven had no human safety studies at all (ITIF). A movement that rightly asks hard questions of pharmaceutical products cannot give chemical-plant peptides a pass because the word sounds green.

Peptides sound natural, and many are. Your body makes thousands of them. But the peptides in vials are chemical products. They may copy a sequence from biology, but they carry only the quantity of the signal, stripped of its quality.

Nature never makes a peptide alone. It routes it, folds it, inspects it, trims it, and delivers it inside a living matrix shaped by soil and sunlight. That is the difference between information and a copy of a single line of it.

If you want peptides, start where they have always come from: sprouts, greens, fermented foods and whole proteins. Let Nature do the chemistry.

Leave a comment

Share

Subscribe now

Disclosure: I am the founder of GreenMedInfo and of Regenerate, a supplement company whose formulas use food-bound, peptide-chelated minerals and spirulina-derived ingredients. This article discusses the general science of food-derived versus synthetic peptides. It does not make claims about any Regenerate product, which require their own evidence.

A lifelong prescription for a problem no one is trying to solve: a root-cause and risk-benefit reckoning

Share the X thread dedicated to this post here.

Imagine your smoke alarm goes off in the middle of the night. You have two choices. You can find the fire. Or you can buy a subscription service that injects a chemical into the alarm once a week so it stops making noise, while the house keeps burning.

For a growing share of the population, the second option is now the standard of care.

Around 12% of U.S. adults are now taking a GLP-1 drug like Ozempic, Wegovy, Mounjaro, or Zepbound (Associated Press, citing KFF). In 2025 alone, Eli Lilly’s tirzepatide franchise brought in about $36.5 billion, outselling Keytruda, the cancer drug long considered the most lucrative product in pharmaceutical history (BioSpace). The leading medical societies now say these drugs, once started, are meant for “long-term or lifelong use” (Obesity, 2026).

Read that again. Lifelong. For a condition that barely existed at population scale two generations ago.

That is the absurdity I want to examine. My argument is not that these drugs do nothing; they clearly change the numbers on the scale and on the lab report. My argument is that they do nothing to address why those numbers went wrong in the first place. And once you see the full ledger of risks against that backdrop, the approach looks less like a medical breakthrough and more like an elaborate, profitable way of not asking the obvious question.

The Old Paradigm in a New Syringe

In REGENERATE, I described what I call the Old Paradigm of metabolic disease:

“In the Old Paradigm, metabolic disorders are treated with a combination of palliative, symptom-reducing drugs and often superficially focused lifestyle and dietary recommendations.”

When I wrote that, the drugs in question were metformin, sulfonylureas, and thiazolidinediones. I noted that people taking metformin, “which doesn’t address the underlying cause,” are often tempted to treat it as a safety net that lets them eat whatever they want. The result, I wrote, is that “a bad diet can lead to questionable medications. It’s a trap that’s snagged millions of Americans.”

GLP-1 agonists are the most sophisticated version of that trap ever built. They don’t repair insulin resistance at its source. They don’t remove the endocrine-disrupting chemicals in our food packaging, restore the microbiome, or fix circadian disruption or chronic stress. What they do is pharmacologically override the body’s appetite and digestive signals: they slow the stomach, blunt hunger in the brain, and force the pancreas’s hand.

The central principle of my work is this: the symptoms of disease are often the body’s intelligent response to something unhealthy or inappropriate. As I put it in the book:

“It is always better to look for the disease’s root cause than to suppress the symptoms.”

And more pointedly:

“Metabolic disease, like virtually all disease, is not so much a force of adversity that accosts us from outside in but our body’s attempt to deal with the onslaught of a radically inappropriate diet, a vast array of toxicant exposures, inactivity, and persistent, health-degrading stress.”

If that is true, and the evidence below suggests it largely is, then a drug that silences the body’s signals while leaving the onslaught intact is not a cure. It is a mute button.

Exhibit A: The Weight Comes Back the Moment You Stop

The clearest proof that a treatment doesn’t touch the root cause is what happens when you stop it. If you put out a fire, it stays out. If you only muffle the alarm, the noise returns the moment the muffling wears off.

That is exactly what the data show.

The evidence: withdrawal trials

• STEP 1 extension (semaglutide). Participants lost 17.3% of body weight over 68 weeks. One year after stopping, they had regained two-thirds of it, and “cardiometabolic improvements … reverted towards baseline … for most variables.” The authors, writing in a Novo Nordisk–sponsored trial, concluded the findings “suggest ongoing treatment is required” (Wilding et al., Diabetes, Obesity and Metabolism, 2022).

• SURMOUNT-4 (tirzepatide). Participants who were switched to placebo after 36 weeks regained 14% of their body weight over the following year. The authors concluded: “withdrawing tirzepatide led to substantial regain of lost weight” (Aronne et al., JAMA, 2024).

• BMJ meta-analysis (37 studies, 9,341 participants). Weight returned at roughly 0.4 kg per month after stopping medication, with a projected return to baseline in about 1.7 years. Blood sugar, blood pressure, and lipids drifted back toward baseline within about 1.4 years (BMJ 2026, via Science Media Centre).

The industry’s response to this is revealing. Rather than conclude that the drug isn’t addressing the cause, it has redefined obesity as a chronic, relapsing disease that needs lifelong medication (WHO guideline, JAMA, 2026). The failure to cure has been rebranded as proof of chronicity.

And in the real world, most people don’t stay on the drug anyway. In a Danish national registry study of 77,310 adults starting Wegovy for weight loss, 52% stopped within a year (Medscape, reporting EASD 2025). So the predictable trajectory for millions of people is this: lose weight, including muscle, stop the drug, regain weight, mostly as fat, and end up with a worse body composition than when they started.

Exhibit B: You Are Losing More Than Fat

That last point deserves its own section, because it goes directly to my concern with regeneration and resilience.

In the STEP 1 body-composition substudy, roughly 40% of the weight people lost on semaglutide was lean mass (European Heart Journal, 2026). Lean mass includes skeletal muscle, the largest site of glucose disposal in the body and one of our strongest predictors of healthy aging.

Think about what that means. A drug sold to correct a metabolic disorder strips away a large share of the very tissue that best protects metabolic health. Then, when the drug is stopped and the weight returns, it does not come back as muscle.

The contrast with a root-cause approach is instructive. In a randomized NEJM trial of weight-loss maintenance, only the strategy that combined exercise with the drug improved insulin sensitivity, glycated hemoglobin, and cardiorespiratory fitness. The drug alone did not (Lundgren et al., NEJM, 2021). Movement, one of the ancestral inputs I emphasize throughout REGENERATE, carried the real metabolic benefit.

Exhibit C: A Hormone Signal the Body Never Evolved to Receive

This is the part of the story that most clearly separates the regenerative view from the pharmaceutical one.

Your body already makes GLP-1. Specialized L-cells in the gut lining release it after you eat. One of the most elegant ways your body triggers its own GLP-1 is through the short-chain fatty acids your gut bacteria produce when they ferment fiber from whole plant foods. Cambridge researchers showed that these microbial metabolites directly stimulate GLP-1 secretion from L-cells through a dedicated receptor called FFAR2 (Tolhurst et al., Diabetes, 2012).

This is what I mean when I write, in REGENERATE, that “food is not just an energy source but information and software for your body.” When you eat fermentable fiber from real plants, your microbiome translates it into a hormonal message that tells your body it has been fed. The system was designed to work this way.

Now compare the natural signal with the drug:

The evidence: native GLP-1 vs. semaglutide

• Native GLP-1 has a half-life of about 2 minutes. It is rapidly broken down by the enzyme DPP-4 and cleared by the kidneys (Diabetes, Obesity and Metabolism review).

• Semaglutide was chemically modified to resist that breakdown. Its half-life is about one week, roughly 165 hours (RCSB PDB-101), and it stays in circulation for about five weeks after the last dose (StatPearls).

• The comparison: the drug’s signal lasts roughly 5,000 times longer than the body’s own (165 hours ÷ 2 minutes).

Your body evolved to release GLP-1 in brief pulses tied to meals and then clear it within minutes. Semaglutide keeps the receptor switched on around the clock, week after week, for years. That is not “mimicking a natural hormone,” as the marketing says. It is replacing a finely timed conversation with a single, continuous, deafening tone.

When you look at the side effects in that light, they are not mysterious. A gut held in a state of permanent, pharmacologically enforced slowdown is a gut at risk of the very injuries now documented in the literature.

Exhibit D: The Risk Side of the Ledger

I recently completed a separate, fully sourced audit of GLP-1 safety evidence. Here are the highlights relevant to a risk-benefit judgment. I keep established harms separate from contested ones, because the case is stronger when it doesn’t overreach.

• Pancreatitis, bowel obstruction, gastroparesis (Established): Compared with another weight-loss drug, GLP-1 users had about 9 times the pancreatitis risk, 4.2 times the bowel-obstruction risk, and 3.7 times the gastroparesis risk (Sodhi et al., JAMA, 2023)

• Severe and fatal pancreatitis (Established): The UK regulator strengthened warnings after 1,296 reports, 19 of them fatal and 24 necrotizing (MHRA, January 2026)

• Gallbladder disease (Established): 37% higher risk overall, more than doubled at weight-loss doses, across 76 trials (He et al., JAMA Internal Medicine, 2022)

• Optic nerve stroke (NAION) (Established outside the U.S.): Europe’s regulator classified it as a very rare side effect of semaglutide, at about twice the baseline risk (EMA, 2025); the U.S. label still does not list it (regulatory ledger)

• Kidney, arthritic, and blood-pressure effects (Probable): Increased interstitial nephritis, kidney stones, arthritic disorders, hypotension, and fainting in more than 2 million veterans (Xie & Al-Aly, Nature Medicine, 2025)

• Thyroid cancer and suicidality (Contested): French data showed a signal (Bezin et al., Diabetes Care, 2023); the largest independent studies did not (Pasternak et al., BMJ, 2024; FDA, 2026)

On the benefit side, the strongest case is the SELECT trial in people who already had cardiovascular disease. Major cardiovascular events occurred in 6.5% of the semaglutide group versus 8.0% of the placebo group over about 40 months (Lincoff et al., NEJM, 2023). That is a 20% relative reduction, and it is a real benefit for a high-risk group.

The math behind the headline

• Absolute difference: 1.5 percentage points over more than three years.

• Number needed to treat: roughly 67 people (1 ÷ 0.015) for about three years to prevent one event.

• Who it applies to: a population that already had heart disease. It does not establish benefit for the general overweight population now being prescribed these drugs.

• Funding: the trial was funded by Novo Nordisk.

Now set that modest absolute benefit next to the harms above, the near-certain weight regain on stopping, the loss of muscle, and a commitment to a drug for life. The risk-benefit calculation for a 35-year-old with a few extra pounds and no heart disease looks very different from the one in SELECT.

Exhibit E: What the Root Cause Actually Looks Like

If the drug addresses the downstream signal, what is upstream? The evidence points to a set of inputs that are well documented and largely ignored in the prescription-first model.

Ultra-processed food

Americans now get 55% of their calories from ultra-processed foods. For children and teenagers it is 61.9% (CDC NCHS Data Brief 536, 2025).

That is not a minor detail. In a tightly controlled NIH inpatient trial, the same people ate about 508 more calories per day on an ultra-processed diet than on an unprocessed one. The two diets were matched for presented calories, sugar, fat, fiber, and sodium, and the participants gained weight on one and lost it on the other (Hall et al., Cell Metabolism, 2019).

In other words, the food itself overrides satiety. This is exactly what I described in REGENERATE about industrial ingredients engineered to hijack our “biologically hard-wired taste receptors.”

Endocrine-disrupting chemicals

The Endocrine Society’s second scientific statement lists “obesity and diabetes” among the areas with “strong mechanistic, experimental, animal, and epidemiological evidence for endocrine disruption” by environmental chemicals (Endocrine Reviews, 2015). In REGENERATE I discuss bisphenols in particular: the BPA in can liners and receipts, and the BPS and BPF quietly substituted for it in “BPA-free” products.

No injectable drug removes those chemicals from your body or your pantry.

Sleep and circadian disruption

In a randomized trial, cutting sleep to two-thirds of normal for eight nights increased calorie intake by 559 calories per day, with no change in energy expenditure (Calvin et al., Chest, 2013). A drug that suppresses appetite does nothing for the sleep deprivation that drove it.

Lifestyle versus drug, head to head

The landmark Diabetes Prevention Program randomized more than 3,000 high-risk adults. Intensive lifestyle change cut progression to diabetes by 58%; metformin cut it by 31% (DPP Research Group, The Lancet, 2009).

In the DiRECT trial, a structured diet program delivered in ordinary UK primary care put 46% of participants with type 2 diabetes into remission at 12 months, off all diabetes drugs, compared with 4% of controls. Among those who lost 15 kg or more, 86% achieved remission (Lean et al., The Lancet, 2017).

I should be honest about the hard part. At five years, sustained remission in DiRECT had fallen to about 13% in the group that stayed in the program, though 40% were still off diabetes medication (Lancet Diabetes & Endocrinology, 2024). Lifestyle change is hard to sustain when you go back into the same environment of ultra-processed food, chronic stress, poor sleep, and chemical exposure.

But notice what that finding actually proves. It does not show that the root-cause approach fails. It shows that the root cause is environmental and still active, and that an individual’s willpower alone cannot overcome an entire food system. That is an argument for changing the environment, not for medicating people for life so they can keep living in it.

Exhibit F: The Absurdity, Fully Assembled

Put the pieces together and the picture becomes almost satirical.

1. The food industry creates the problem. It feeds Americans a diet that is 55% ultra-processed. Controlled trials show this kind of diet makes people overeat by hundreds of calories a day.

2. The pharmaceutical industry sells the fix. Lilly and Novo Nordisk spent more than $270 million on payments to more than a quarter-million U.S. prescribers from 2018 to 2025 (The Washington Post, via ADN). The FDA formally warned Novo Nordisk that a televised special promoting its drugs was “false or misleading” and minimized their risks (FDA Warning Letter, 2025).

3. The food industry sells products for the people on the fix. Nestlé, Conagra, and others now sell “GLP-1 Friendly” products. One such frozen pizza delivers 40% of the daily value for sodium and saturated fat. Smoothie King’s “GLP-1” smoothie has more calories than a Krispy Kreme glazed doughnut. The USDA confirmed that there is “no regulatory standard” for the term “GLP-1 Friendly” (Associated Press, 2026).

4. The cycle extends to children. Children get about 62% of their calories from ultra-processed foods, and the American Academy of Pediatrics now advises pediatricians to offer weight-loss drugs to adolescents 12 and older. A BMJ analysis found financial ties to GLP-1 makers “throughout” the guideline process (Schmidt et al., The BMJ, 2025).

There is a cycle here, and every turn of it is profitable. Sick people are a recurring revenue stream. People who get well are not.

What Regeneration Asks of Us Instead

The core of my philosophy, laid out in REGENERATE, is simple enough to fit in a single sentence:

“On the most basic level, regeneration constitutes removing what the body does not need, and adding back what it does.”

Applied to metabolic disease, that means a different set of priorities than a weekly injection:

• Remove what does not belong. Cut ultra-processed foods, industrial fructose and high-fructose corn syrup, and flavor-manipulating additives like MSG. Reduce bisphenols and other endocrine disruptors by avoiding canned foods, plastic containers, and thermal-paper receipts.

• Add back what was lost. Eat whole, fiber-rich plant foods and resistant starches that feed the microbes that make your body’s own GLP-1. Get real sunlight, protect your sleep, and move with intensity. Short bursts of high-intensity interval training can rapidly improve glucose metabolism in people with insulin resistance.

• Respect the signal. Hunger, cravings, insulin resistance, and the accumulation of belly fat are not defects to be overridden. They are, as I write in the book, “compensatory adaptations seeking to re-establish equilibrium.” They tell you that something in your environment is wrong.

People with severe, established disease who need a medication should have access to it. The point is not to shame anyone for taking a drug or to deny that some high-risk patients may benefit. The point is that a society that responds to an epidemic driven by food, chemicals, sleep loss, and stress by putting one in eight adults on a lifelong appetite-suppressing injection has decided not to ask the most important question.

In REGENERATE I wrote that “it’s hardly news that more than two-thirds of Americans are overweight” and that metabolic syndrome is “an optimal example of a chronic disease category that can be universally mitigated and reversed through diet, lifestyle choices,” and through addressing the maladaptive stress response. I stand by that. The science of the last decade has only strengthened it.

The fire is still burning. The alarm is doing its job. The question is whether we will keep paying to silence it.

For more research on thousands of natural approaches to overweight and obesity, consult the extensive databases on the topic on GreenMedInfo.com.

Leave a comment

Share

Subscribe now

This article is for educational purposes and is not medical advice. Do not stop or change any prescribed medication without consulting a licensed healthcare professional, particularly if you have diabetes or cardiovascular disease.

This year, peptides went mainstream.

This year, peptides went mainstream.

On February 27, 2026, the Secretary of Health and Human Services announced that about 14 of the 19 peptides the FDA had restricted from compounding would be reclassified. The list included BPC-157, thymosin alpha-1, TB-500, GHK-Cu, and MOTS-c (AJMC). In July, an FDA advisory committee narrowly voted to recommend six of them, including BPC-157 and the “longevity peptide” epitalon, for the compounding list. Agency scientists had argued against them (Latham & Watkins; PressAdvantage). Meanwhile, a gray market thrives, and the press says people “are turning themselves into lab rats” (The Guardian).

Testing shows what can end up in those vials. In a 2024 study, three online semaglutide vials claimed 99% purity but measured just 7.7% to 14.4%. All three held more drug than their labels stated, and all contained bacterial endotoxin (Ashraf et al., Journal of Medical Internet Research, 2024). In 2026, an Australian lab reported that only 1 of 18 samples sold as retatrutide matched its label, and one contained none of the peptide at all (ABC News). The FDA keeps sending warning letters to these sellers. It says a “research use only” label does not change the fact that the products are marketed for human use (FDA).

Many of the most sought-after products are not natural signals at all. Retatrutide, which is still in clinical trials and has not been approved (ABC News), is a single lab-designed molecule built to act on three hormone receptors at once. Researchers developing the next generation of these designs call them “chimeric peptides,” or “chimeras” (Molecular Design of Unimolecular Tetra-receptor Agonists). Sellers also mix several peptides into one vial, such as blends of BPC-157, TB-500, and GHK-Cu. These blends have not been clinically validated as combinations (Disclosed Labs).

I support people’s right to make informed choices about their own bodies. But the excitement over vials and syringes has hidden a deeper question.

If short regulatory peptides really are among the body’s master signals for staying young, where did we first get them?

The answer has been growing in fields for 10,000 years, and a group of aging researchers proved it two decades ago.

The Experiment That Asked the Right Question

The thymus is the textbook example of early aging. This small gland behind the breastbone trains our immune T-cells, and it begins shrinking long before most other organs. For decades, aging researchers have known that giving old animals young thymus tissue can roll back signs of age. One Italian team showed that grafting neonatal thymus tissue into old mice measurably changed their mortality rate and extended their lifespan (Basso et al., Rejuvenation Research, 2010).

But you cannot give everyone a thymus transplant. So the researchers asked a bolder question:

Could a food do what the thymus does?

In 2005, in the peer-reviewed journal Biogerontology, they published the answer. They tested a wheat sprout extract to see whether it “could mimic the thymus action on recovering age-related alterations,” and whether it could treat “an age-associated pathology, the cataract in dogs” (Basso et al., Biogerontology, 2005).

Here is what they reported:

• Aging liver cells moved back toward youth. In old mice, the wheat sprout extract restored the declining DNA synthesis of liver cells, much as calf thymus extract did. It also brought abnormally high DNA and protein content back down “to levels observed in young mice.”

• Sprouting multiplied the biological activity. The sprouts showed 10 to 40 times more phosphorylation activity than the seeds they grew from.

• Cataracts receded. Old dogs given the extract by mouth for one month showed a 25–40% reduction in lens opacity.

The authors credited the effect to “the contemporary presence of small regulatory acid peptides, a remarkable level of highly energetic phosphoric radicals and antioxidant molecules” (Basso et al., 2005).

Small regulatory peptides, in other words, from a grass.

Why a Cataract Matters More Than You Think

Most tissues in the body renew themselves. The lens of the eye mostly does not. Its proteins are laid down early in life and never replaced, so cataract, the clouding of those proteins, works like a lifetime record of accumulated damage. Conventional medicine treats it as essentially irreversible. Once vision is impaired, the answer is surgery.

That is why a measurable reduction in lens opacity in one month, from an oral, food-derived material, should have made headlines. It challenges one of the deepest assumptions in aging biology: that damage like this only goes one way.

It was also not a one-off. A year earlier, the same research network had reported that plain wheat sprout powder given orally produced “a significant reduction of senile cataract in old dogs,” and that the sprouts’ active substances “can be at least partially absorbed during digestion” (Marsili et al., Journal of Clinical Gastroenterology, 2004).

A 30-Year Trail of Peptide Science

The cataract result is the payoff of a long research arc that most of the longevity world has forgotten.

• 1976: Researchers isolated a small peptide (under 5,000 daltons) from calf thymus that strongly controlled gene transcription by interacting directly with DNA. They published it in Nature (Gianfranceschi et al., 1976).

• 1983: Given to old rats, these thymic DNA-binding peptides reversed age-related lipid changes, restored liver glycogen to adult-like levels, and raised blood ATP, the cell’s energy currency, into the range of young animals (Amici et al., Mechanisms of Ageing and Development, 1983).

• 2002: Wheat sprout extracts were shown to shield DNA from oxidative breaks (Falcioni et al., Journal of Food Science, 2002).

• 2005: The team isolated short acidic peptides from the chromatin of wheat sprouts, the same class of regulatory molecule they had found in the thymus. Some of the sequences also appeared in other tissues, which supported the idea of “ubiquitous regulatory peptides” shared across life (Calzuola et al., Peptides, 2005).

• 2011: These sprout peptides were found to travel in phospholipid–peptide complexes, natural packaging the researchers suggested may help explain their bioavailability (Perni et al., Journal of Peptide Science, 2011). The lab summarized its findings as “antioxidant bioavailable peptides,” DNA protection, and “anti-aging activity” (Marsili et al., 2011).

Think about what this means. Thymosin alpha-1, one of the peptides now headed for compounding pharmacies, is a thymic peptide. The original insight behind it, that the thymus speaks to aging tissue through short peptide signals, is the same one that led these researchers to wheat sprouts. Nature built a version of that conversation into a seedling.

The Evidence Keeps Pointing the Same Way

The Italian work does not stand alone. Across species, wheatgrass and its key constituents keep showing signs of slowing or reversing aging:

• Longer life in fruit flies. Flies fed wheatgrass lived longer, reaching 58 days compared with 52 in controls (Sprouted Grains review, Nutrients). A 2024 study found that a flavonoid extract from a high-flavonoid wheatgrass variety extended male fruit-fly lifespan in a dose-dependent way (Wang et al., Seed Biology, 2024).

• Chlorophyll as a longevity molecule. Chlorophyll, which gives wheatgrass its color, extended the lifespan of C. elegans worms by up to 25% through the DAF-16/FOXO longevity pathway (PeerJ, 2016). It also extended fruit-fly lifespan by 7.7–13.9% in a 2026 study (Shen et al., Nutrients, 2026).

• A protected aging brain. In mice with chemically induced memory loss, wheatgrass extract improved cognitive performance, raised brain-protective BDNF, and reduced tau, a protein linked to Alzheimer’s (Katiyar et al., Food & Function, 2022).

• Antioxidant defenses rebuilt in older adults. Eighteen adults aged 50–65 drank fresh wheatgrass juice daily for 30 days. Their plasma antioxidant capacity roughly doubled, and vitamin E rose 128%, vitamin A 121%, vitamin C 68%, and hemoglobin 39% (Sharma et al., Journal of Cell and Tissue Research, 2008).

• Human trials in real disease. In a randomized, double-blind, placebo-controlled trial, wheatgrass juice significantly reduced disease activity and rectal bleeding in ulcerative colitis (Ben-Arye et al., 2002). In women undergoing chemotherapy for breast cancer, it was associated with fewer blood-toxicity setbacks without reducing the chemotherapy’s effectiveness (Bar-Sela et al., 2007).

GreenMedInfo’s database now indexes 158 study abstracts on wheatgrass across 82 health topics (GreenMedInfo).

So Why Hasn’t Anyone Followed Up?

This is the part that should trouble us.

A peer-reviewed aging study reported that a food could stand in for the thymus and clear lens opacity in a month. In the more than twenty years since, I have found no independent team that has repeated the dog cataract experiment. No one has run a controlled human trial of sprout extracts for early lens changes. No one has put wheatgrass up against a modern biological-age clock.

The reason is not hard to guess. You cannot patent a sprout. Research money flows toward molecules that can be owned. A synthetic peptide can be patented, branded, and argued over at FDA advisory meetings. A tray of wheatgrass on a kitchen windowsill cannot.

I want to be clear about what the science has and has not shown, because honesty is what gives this story its power. The 2005 study reversed specific aging markers: liver-cell DNA patterns and lens opacity. It did not measure whole-body biological age, and its full methods deserve scrutiny and replication. That is not a reason to dismiss the finding. It is the strongest possible argument for funding the follow-up.

What We Are Calling For

1. Replicate the cataract study. Run randomized, placebo-controlled, masked trials in older dogs with veterinary ophthalmology imaging, as a first step toward human research.

2. Test wheatgrass against biological-age clocks. Run pilot human trials measuring epigenetic age, inflammation, and functional markers, not just antioxidant levels.

3. Fund food-based longevity research on equal footing. Public research agencies spend heavily on patentable aging drugs. Unpatentable foods with peer-reviewed signals deserve a fair hearing.

4. Protect access to whole foods and traditional preparations. As peptide regulation is rewritten in 2026, policymakers should remember that some of the most compelling peptide science began with food.

What You Can Do Today

• Grow it. Wheatgrass sprouts in 7–10 days from organic wheat berries, soil, and a sunny window. Fresh juice is how the human studies used it: 30 g of grass blended in 100 ml of water in the older-adult study (Sharma et al.), and 60–100 ml daily in the clinical trials (Ben-Arye et al.; Bar-Sela et al.).

• Start low. Some people get nausea at first. That was the main side effect reported in the chemotherapy study (Bar-Sela et al.).

• Know your sensitivities. Young wheatgrass, harvested before it forms seed, is generally considered gluten-free, but cross-contamination is possible. People with celiac disease should be cautious.

• Partner with your eye doctor. Cataracts affect driving, falls, and quality of life. Use food as an ally, and keep up with regular eye exams.

• Explore the research yourself on GreenMedInfo’s Wheat Grass page.

The Bigger Lesson

The peptide boom is not wrong that small signaling molecules matter. It is wrong about where to look first.

Life has been encoding regulatory information into food since long before there were laboratories. A sprouting seed is a burst of growth signals: peptides, phosphates, antioxidants, and chlorophyll, all working together to tell a living system to grow and renew. When aging researchers asked whether that signal could stand in for the thymus, the answer they published was yes, at least for the markers they measured.

This is what I mean when I say food is information. The most important longevity peptide may not come from a vial. It may come from the soil.

Leave a comment

Share

Subscribe now

Love this type of reporting on natural, science-based solutions? GreenMedInfo.com’s free newsletter has been serving the world with natural solutions for the past 17 years. Subscribe for free here today.

Learn about the ways we have put this wisdom into living products you can use for yourself and your family at REGENERATE COMPANY, a product line I launched this year.

This article is for educational purposes and is not medical advice. Consult a qualified healthcare provider before making changes to your health regimen, especially if you have an eye condition, are pregnant, or take medication.

On June 28, 2022, Sesame Street posted a 60-second video to its YouTube channel titled “Elmo Gets the COVID-19 Vaccine | PSA.”

It’s a well-made piece of persuasion. Elmo, who has been three and a half years old since 1984, shows off his “super duper” bandage. There “was a little pinch. But it was okay.” Then his father, Louie, looks into the camera and speaks to every hesitant parent in America:

“I had a lot of questions about Elmo getting the COVID vaccine. Was it safe? Was it the right decision?”

Eleven seconds later, the doubt has been resolved for you:

“I learned that Elmo getting vaccinated is the best way to keep himself, our friends, neighbors and everyone else healthy and enjoying the things they love.”

A hug follows. Then an announcer: “It’s okay to have questions about COVID vaccines for your kids.”

That’s the move worth studying. The ad takes the parent’s instinct to ask questions, gives it to a beloved puppet, and answers it before the parent has a chance to. It presents doubt so it can close it down.

This was not a whimsical one-off from a children’s television workshop. It was one output of what its own organizers called the largest communications program in our nation’s history, built with federal public-health agencies, a major medical trade association, and tens of millions of corporate dollars.

Four years later, Pfizer-BioNTech’s product is no longer authorized for children ages six months through four years. Federal guidance and product labeling have also narrowed which pediatric uses are on-label, a shift sharply at odds with the PSA’s broad, categorical reassurance. Nobody made a puppet video about that.

This is the story of who built the campaign, what it cost, and what came after.

Part I: Who Was Behind It

Sesame Workshop’s own press release is candid about who was in the room. The PSA was released “in collaboration with the Ad Council and COVID Collaborative’s COVID-19 Vaccine Education Initiative,” and it was “produced in partnership with the Centers for Disease Control and Prevention (CDC) and American Academy of Pediatrics (AAP).”

Five institutions. Each one deserves a name and a face.

Sesame Workshop

The nonprofit behind Sesame Street produced and released the spot. Its campaign lead was Dr. Jeanette Betancourt, Senior Vice President of US Social Impact. Its president at the time, Sherrie Westin, was named CEO in July 2024.

The clearest statement of what Sesame thought it was offering came from its own chief marketing officer. Launching an earlier round of vaccine PSAs, Samantha Maltin, EVP and CMO, said the goal was to:

“…harness the power of Sesame Workshop and the reach of the Ad Council…”

Harness the power. The asset being put to use was fifty years of trust that parents and children have placed in these characters.

The Ad Council

The Ad Council turns corporate media budgets into “public service” messaging. Its president and CEO, Lisa Sherman, was the campaign’s public face. On the Elmo spot:

“It’s important that parents feel informed and confident about what the COVID-19 vaccines mean for their families, and we are proud to work with our partners, medical experts – and of course Elmo – to spread that message.”

She also named the urgency: this was “an important decision many need to make in advance of the upcoming school season.”

COVID Collaborative

The COVID Collaborative is a bipartisan assembly of former officials, and it convened the vaccine education initiative with the Ad Council. Its CEO is John Bridgeland, former Director of the White House Domestic Policy Council. Its president is Gary Edson, and its co-chairs are former Idaho Gov. and Sen. Dirk Kempthorne (R) and former Massachusetts Gov. Deval Patrick (D). The Collaborative is a project of UNITE, chaired by Tim Shriver.

The CDC

The CDC was a named production partner, and it wasn’t a passive one. According to the campaign’s organizers, the CDC provided scientific guidance on all aspects of the campaign, and its brand appeared on select creative assets.

The American Academy of Pediatrics

The AAP, the pediatricians’ professional association, was the other named production partner. Its president in 2022 was Dr. Moira Szilagyi. Looking back later, she said: “Somewhere what got lost in our country, I think, was what a miracle these vaccines were.”

Keep the AAP in mind. It comes back in Part IV.

Part II: Follow the Money

The Elmo spot was built on a large funding structure, and its size helps explain how thoroughly this message saturated American media.

The corporate fund. On November 23, 2020, before any vaccine was authorized, the Ad Council announced a $50 million fund for the COVID-19 Vaccine Education Initiative, created with the COVID Collaborative. By February 2021 it had exceeded that goal.

The COVID Collaborative reports more than $52 million raised. The leading contributors it names are Amazon, Apple, Bank of America, Cisco, CVS Health, Facebook, General Motors, Google and YouTube, the Humana Foundation, NBCUniversal/Comcast, Salesforce, Verizon, Walgreens, and Walmart.

Two of those names stand out. CVS Health and Walgreens both gave the shots the campaign was promoting in their own pharmacies.

The donated media. The money was only the seed. By October 26, 2021, the initiative had surpassed its goal of $240 million in donated media and related publicity. Donors of airtime and ad inventory included Disney (ABC, ESPN, Hulu), FOX, NBCUniversal and Telemundo, ViacomCBS, WarnerMedia, Facebook and Instagram, Google and YouTube, TikTok, Snapchat, Spotify, Pinterest, LinkedIn, and The Weather Channel.

Several of the same platforms that donated airtime and advertising inventory to the vaccine-education campaign were simultaneously under public and private pressure to enforce COVID-19 misinformation rules against specifically identified individuals, including people named in CCDH’s ‘Disinformation Dozen’ report. YouTube later my channels and stated that, after review, they did not violate its Community Guidelines.

The taxpayer layer. Alongside the private effort, HHS ran its own $265 million COVID public-education campaignstarting in summer 2020. That included a $250 million NIH award to the Fors Marsh Group, which created the “We Can Do This” brand, all paid for through a $300 million interagency agreement drawing on CDC’s CARES Act funds. Under its “Vaccinate with Confidence” framework, the CDC funded more than 200 partners. Across the March 2021–December 2022 rollout, it funded 110 primary and 1,051 subrecipient partners at the national, state, and community level.

A note on accuracy. I have not found documentation that Pfizer or Moderna directly paid for the Elmo spot, and the campaign materials don’t list them as partners. Pfizer does appear as a Sesame Workshop donor in later years, in the $100,000–$249,999 band of its 2024 impact report. That’s worth knowing, but it’s a separate fact. The documented structure is damning on its own: corporate money, federal science guidance, and a medical trade association, all speaking through a puppet.

Part III: It Wasn’t One Ad

The Elmo spot was the youngest-audience edge of a sustained effort to put Sesame Street characters to work for vaccine messaging.

• November 2021: CNN and Sesame Street co-produced a family town hall, “The ABCs of COVID Vaccines.” That same month, Big Bird delivered his own pro-vaccine message to young viewers, which drew public criticism from Sen. Ted Cruz.

• June 2022: Elmo’s PSA, released days after the FDA’s emergency authorization for children as young as six months.

• Later: Sesame Workshop, the Ad Council, COVID Collaborative, and CDC launched a further series of PSAs with Elmo, Louie and other Muppets, “created in close partnership with the CDC.”

The criticism started right away. Cruz accused Sesame Street of having Elmo “aggressively advocate” for shots in young children and noted that the spot cited no scientific evidence. The comedian Russell Brand put it more sharply:

“When it starts getting into complex territory that involves the power of the state, the influence of pharmaceutical companies, the impact of coronavirus on young children — these are not issues I want resolved by members of Elmo’s family.”

The critics were dismissed as cranks at the time. Here is what the government has said since.

Part IV: The Walk-Back Nobody Advertised

There was no $240 million campaign for any of what follows. No Muppets and no hug at the end. It came out as label changes, regulatory memos, and footnotes in clinical guidance.

The cardiac warning got stronger

On June 25, 2025, the FDA required and approved expanded warnings about myocarditis and pericarditis on Pfizer’s Comirnaty and Moderna’s Spikevax. The new label estimates about 8 cases per million doses in people aged 6 months through 64 years, and about 27 per million in males aged 12 through 24, in the week after vaccination.

The most important sentence isn’t about frequency, though. It’s about what happened to the hearts afterward:

“…at a median follow-up of approximately 5 months post-vaccination, persistence of abnormal cardiac magnetic resonance imaging (CMR) findings that are a marker for myocardial injury was common. The clinical and prognostic significance of these CMR findings is not known.”

Read it twice. Five months out, markers of heart-muscle injury were “common,” and the agency says their long-term significance is “not known.” This isn’t a critic’s claim. It comes from an FDA-funded, FDA-co-authored study, and it’s now printed on the label.

The spot told parents “there was a little pinch. But it was okay.” Nobody in the ad said “unknown long-term cardiac significance.”

The authorization for the youngest children was pulled

On August 27, 2025, Pfizer-BioNTech’s COVID vaccine was no longer authorized for children under 5. CDC’s current clinical guidance puts it plainly: “Pfizer-BioNTech COVID-19 Vaccine is no longer authorized in the United States for children ages 6 months–4 years.”

That is the age group the Elmo spot was released to reach. It went out days after emergency authorization opened the shots to children as young as six months. Four years later, one of the two products it promoted to parents of those children can’t legally be given to them.

Healthy children are now “off-label”

The broad “everyone six months and older” recommendation is gone. In May 2025, HHS dropped the routine recommendation for healthy children, and the CDC moved to what it calls shared clinical decision-making.

CDC’s current guidance, updated September 23, 2026, limits Moderna’s approval for the youngest children to those with “at least one condition that places them at increased risk for severe COVID-19.” Then it states the conclusion outright:

“Vaccination of children 5-11 years of age with any product other than Spikevax or Comirnaty and vaccination of children in this age group without high-risk conditions is off-label use.”

“Vaccination of persons 12-64 years of age without high-risk conditions is off-label use.”

In 2022, the message was that vaccinating Elmo was “the best way to keep himself, our friends, neighbors and everyone else healthy.” In 2026, the CDC’s own wording is that giving the shot to a healthy child is off-label.

The deaths question

On November 28, 2025, Dr. Vinay Prasad, then director of the FDA’s Center for Biologics Evaluation and Research, sent staff a memo headed “Deaths in children due to COVID-19 vaccines and CBER’s path forward.” He wrote that career staff “have found that at least 10 children have died after and because of receiving COVID-19 vaccination,” based on an initial review of 96 child deaths reported to VAERS between 2021 and 2024.

The FDA’s formal analysis, made public in May 2026, came back more cautious. Of the 96 reports reviewed, none were classified as “certain.” Five were classified “possible” and two “probable,” mostly involving myocarditis, with an average age of 13. The agency stressed that these classifications don’t prove causation.

Media coverage framed that result as a debunking. Set the headline aside, though. The FDA’s own conservative review, using WHO causality criteria, still found seven child deaths it could not rule out as vaccine-related, and it called two of them probable.

And yet, still being promoted

This is where the irony stops being subtle.

Visit the FDA’s parent-facing page today, Vaccines for Children – A Guide for Parents and Caregivers. The COVID-19 vaccine is still listed alongside measles and polio among “the most commonly administered vaccines.” And parents are still told:

“Getting a COVID-19 vaccine is the best way to prevent COVID-19, including the potential consequences of hospitalization and death.”

That’s Louie’s line almost word for word, four years later, from the same agency whose label now carries the cardiac warning, whose authorization for under-5s was pulled, and whose own reviewers flagged seven child deaths.

The American Academy of Pediatrics, a production partner on the Elmo spot, hasn’t backed off either. As of September 2026, the AAP recommends COVID-19 vaccination for “all infants and children ages 6 months through 23 months,” and for “anyone ages 2 to 18 who isn’t at risk but whose parents or guardian wants them to get a shot.”

So we have two different postures at once. The regulator has narrowed approval, expanded the warning, and reclassified healthy-child use as off-label. The professional association that helped make the puppet ad still recommends the shot for every baby. And the FDA’s own parent guide still uses the 2022 talking points.

Part V: What They’ll Say — and Why It Doesn’t Save Them

Serious readers deserve the strongest counterargument, so here it is.

CDC’s guidance cites a retrospective cohort of more than 4 million people aged 16–25 from 2020 to 2025. It found myocarditis incidence of 0.05% after COVID illness versus 0.005% after vaccination. In other words, the infection carries a higher myocarditis risk than the shot.

That’s a legitimate data point and I’m not going to pretend it doesn’t exist. It doesn’t answer the question this piece is asking, for three reasons.

First, the comparison is an average over a population of teenagers and young adults. It says nothing about a healthy toddler with near-zero risk of severe disease. That’s the child this campaign targeted, and the child for whom the CDC now calls the shot off-label.

Second, it doesn’t address the finding printed on the label: persistent markers of heart-muscle injury at five months, of “not known” significance. The open letter discussed below says there are “no studies demonstrating prevention of hospitalization or death” in children. If that’s right, an unknown cardiac risk is being weighed against a benefit that was never shown.

Third, and most importantly, the argument here doesn’t depend on any contested injury number. It depends on an asymmetry. When the product was being promoted, it got a puppet, $50 million, a quarter-billion in donated airtime, the CDC’s logo, and the pediatricians’ endorsement. When the regulator narrowed the product, it got a label revision and a footnote. The public was sold a certainty the evidence never supported, and was never told when that certainty was withdrawn.

That isn’t public health communication. It’s a marketing campaign that never issued a correction.

Part VI: The Petition — Why I Signed

Earlier this month, 168 physicians, scientists, researchers, lawmakers, and health-freedom advocates, representing more than 63 million followers, sent an open letter to HHS Secretary Robert F. Kennedy Jr. and President Trump demanding removal of the mRNA platform. The effort was coordinated by Dr. Mary Talley Bowden. The signatories include Rep. Thomas Massie, former Rep. Marjorie Taylor Greene, Louisiana Deputy Surgeon General Dr. Sean Troxclair, Dr. Peter McCullough, Dr. Robert Malone, Dr. Aseem Malhotra, Tucker Carlson, and others.

I’m one of them, and I want to explain why.

The letter’s first demand is to “ban mRNA/gene therapy-derived technologies for all vaccines.” That includes the products still being given to children. Its other demands are to end the PREP Act declaration and its “ironclad liability shield for manufacturers and administrators,” ask Congress to repeal the PREP Act, ban direct-to-consumer pharmaceutical advertising, end the incentive structures that push providers toward medical coercion, and end conflicts of interest at the CDC, FDA, NIH, and NIAID.

On children specifically, the letter makes the point this whole piece has been building toward:

“Parental choice is not realistic when parents are not getting informed consent on the true risks of these products.”

It also takes on the “shared clinical decision-making” framework directly:

“By treating the decision to inject a child with mRNA as a ‘shared decision,’ while treating routine injections as the default, a dangerous legal precedent is being set that assigns powers to the federal government that have historically belonged solely to parents.”

According to the letter, more than 50 million Americans received mRNA COVID shots this year, including about 7 million children, and HHS has awarded $1.24 billion in contracts to Pfizer for future mRNA shots, “including for children.”

The Elmo spot is exactly why that line about informed consent matters. You can’t give informed consent to a hug from a puppet. A one-minute ad that turns a parent’s hesitation into a script and then answers it in eleven seconds isn’t informing anyone. It’s persuasion aimed at the people least able to push back: parents who trust Sesame Street and children who love Elmo.

The letter isn’t the only pressure. Children’s Health Defense’s citizen petition, asking the FDA to revoke the Pfizer and Moderna licenses, has been pending since December 8, 2025 and has drawn more than 104,000 public comments in support. On September 25, the Independent Medical Alliance called on the CDC to immediately withdraw its “off-label” recommendation for healthy infants, children, and adults.

HHS’s response to the open letter, through press secretary Emily Hilliard, was that Secretary Kennedy “believes mRNA products warrant heightened scientific scrutiny,” and that HHS had “wound down investments in mRNA vaccines for upper respiratory viruses because the technology does not effectively protect against infection from rapidly mutating viruses such as COVID-19 and flu.”

Think about that sentence next to the Elmo spot. The federal health department now says the technology “does not effectively protect against infection” from COVID. In 2022, a puppet told America it was “the best way to keep himself, our friends, neighbors and everyone else healthy.”

Heightened scrutiny isn’t enough. A product that federal officials now concede doesn’t reliably prevent infection, that carries a cardiac warning with unknown long-term significance, whose own regulator calls it off-label for healthy children, and whose makers are protected from liability should not be given to children at all.

Conclusion: The Correction That Never Aired

Every institution in the Elmo spot is still operating. Sesame Workshop is still making children’s programming. The Ad Council is still running “public service” campaigns. The CDC’s logo is still on its creative. The AAP still recommends the shot for every infant. The FDA’s parent guide still repeats Louie’s line.

None of them has produced a PSA that says: We told you this was the best way to protect your baby. Since then the label has been changed to warn about heart injury. The authorization for the youngest children was withdrawn. The government now calls this use off-label for healthy kids. We’re sorry we used Elmo to tell you otherwise.

That’s the one-minute spot this country is owed. Until it airs, the original belongs where I’ve put it: in the Hall of Fame of the Covid-era Mockingbird Media, as a case study in what happens when institutions that are supposed to earn a parent’s trust decide to manufacture it instead.

Watch it again. Listen for the moment Louie asks, “Was it safe?”

It was a good question. It still is. They just never let him wait for the answer.

Subscribe now

Leave a comment

Sources

• Sesame Workshop, Elmo PSA press release (June 28, 2022)

• Sesame Workshop, Sesame Workshop, Ad Council, COVID Collaborative and CDC launch new PSAs

• COVID Collaborative, “It’s Up To You” Vaccine Education Campaign

• Ad Council, Our COVID-19 Vaccine Retrospective

• GAO, COVID-19: Information on HHS’s Public Education Campaign (GAO-22-104724)

• CDC, Operationalizing the Vaccinate With Confidence Framework

• CDC, Funded partners during the COVID-19 vaccination rollout

• FDA, Required updated myocarditis/pericarditis warning (June 25, 2025)

• FDA, Vaccines for Children – A Guide for Parents and Caregivers

• FDA, Prasad memo: “Deaths in children due to COVID-19 vaccines” (Nov. 28, 2025)

• CDC, 2026–2027 COVID-19 Vaccination Guidance (updated Sept. 23, 2026)

• Contemporary Pediatrics, FDA pulls clearance of Pfizer-BioNTech COVID-19 vaccine for children under 5

• Reuters, US CDC keeps COVID vaccine option for healthy children (May 30, 2025)

• NBC News, No child deaths definitively linked to Covid shots, FDA says (May 2026)

• NPR, Medical groups offer vaccine guidance to fill federal vacuum (Sept. 2, 2026)

• Focal Points, Open letter and full signatory list (Sept. 21, 2026)

• Global Research, MAHA coalition open letter to Trump and RFK Jr.

• Just the News, “You will pay a political price” (Sept. 21, 2026)

• Children’s Health Defense, Citizen Petition (FDA-2025-P-6831)

• Ground News / IMA, IMA calls on CDC to withdraw off-label recommendation (Sept. 25, 2026)

• CNN, The ABCs of COVID Vaccines town hall

• The Hill, Cruz knocks Sesame Street

• The Guardian, Ted Cruz bashes Sesame Street video

• Children’s Health Defense, Elmo Gets the Jab

• Sesame Workshop, 2024 Mid-Year Impact Report

A meta-analysis published on the eve of Breast Cancer Awareness Month declares overdiagnosis “less than 5%.” Its benchmark is its lead author’s own 2013 study, drawn from a dataset the same researchers showed in 2018 can yield anywhere from 1% to 55%.

Story at a Glance

• The timing. On September 14, 2026, seventeen days before Pinktober, the Journal of the National Cancer Institute published a reanalysis of the randomized mammography trials concluding that the results are compatible with very low overdiagnosis, under 5%. Prior estimates from the same trials ranged up to roughly 50%.

• The conflict. A co-author is Robert A. Smith, PhD, Senior Vice President of Early Cancer Detection Science at the American Cancer Society, who has led the development of ACS cancer screening guidelines for most of his tenure there since 1992. His faculty profile lists an award from the National Breast Cancer Awareness Month Board of Sponsors among his honors.

• The circularity. The study grades every trial against a single real-world benchmark: Funen County, Denmark. The paper’s own supplementary data sources that benchmark to one study: the 2013 Funen analysis led by the same lead author, Sisse Helle Njor.

• The method dependence. In 2018, Njor, Matejka Rebolj, and Eugenio Paci showed that the same Funen data, run through five other published study designs, produces estimates of 21% to 55%. They argued those designs were flawed. But their own conclusion was that the answer depends on the study design, which is exactly why a study built on their preferred design cannot settle the question.

• The pattern. Breast Cancer Awareness Month was co-founded in 1985 by the pharmaceutical arm of a chemical conglomerate that also sold the leading breast cancer drug. Forty-one years later, the institutions that promote screening are the institutions grading whether screening harms women.

The Narrative Crack

Every October, I write some version of the same sentence, and every October it remains true: Breast Cancer Awareness Month is not a public health campaign that acquired corporate sponsors. It is a corporate campaign that acquired a public health vocabulary.

I first documented this in 2011 and 2012, in the two-part series “The Dark Side of Breast Cancer (Un)Awareness Month.” I returned to it in 2022 in “Covering Up The Causes of Breast Cancer Since 1985,” in 2024 in “American Cancer Society’s Pink Smokescreen,” and again last October in “Stop Pinkwashing: The Truth Breast Cancer Charities Bury.”

The facts of the campaign’s origin are a matter of public record. National Breast Cancer Awareness Month was founded in October 1985 as a partnership between the American Cancer Society and the pharmaceutical division of Imperial Chemical Industries, the British conglomerate whose chemical businesses produced the herbicide paraquat, the solvent trichloroethylene, and PVC made from vinyl chloride. Trichloroethylene and vinyl chloride are both now classified by the International Agency for Research on Cancer as Group 1 human carcinogens.¹ ICI’s pharmaceutical division was spun off as Zeneca in 1993 and merged into AstraZeneca in 1999. It sold tamoxifen, the leading drug prescribed to the women the campaign’s mammograms would find, and later Arimidex.

Tamoxifen is itself classified by IARC in Group 1, carcinogenic to humans, on the basis of sufficient evidence that it causes endometrial cancer.² A drug that reduces the risk of one cancer while raising the risk of another is not a scandal in itself; oncology is full of such trades. It becomes a scandal when the company selling it also builds the awareness campaign that decides which risks the public hears about.

The campaign’s stated aim from the outset was to promote mammography as the most effective weapon against breast cancer.¹ Reducing exposure to carcinogens was not part of that founding mission.

This year something new happened, and it is more interesting than another pink drill bit.

Seventeen days before Pinktober 2026, the scientific literature produced an alibi.

The Alibi

On September 14, 2026, the Journal of the National Cancer Institute published a meta-analysis titled “Breast cancer overdiagnosis in mammography trials: separating signal from noise.”³

Its conclusion is the most useful sentence the screening establishment has been handed in twenty years. Analyzed within the authors’ framework, the trials are “compatible with very low overdiagnosis (<5%).”³

The press release from the University of Southern Denmark went further. Lead author Sisse Helle Njor told reporters that with this study, women “can now be reassured” that the benefits of early detection outweigh the small risk of unnecessary treatment.⁴ The senior author, Matejka Rebolj of Queen Mary University of London, said some earlier high estimates had rested on trial evidence that had not yet fully matured.⁴

Within two weeks the finding had traveled through the global health press as settled fact. ScienceDaily ran it under the headline “A major risk of breast cancer screening may have been overestimated for decades.”⁵ Outlets from Kyiv to Delhi ran translations.⁶

Understand what is being retired here. Overdiagnosis is not a technical footnote. It is the principal harm on the mammography ledger. It is the number behind the Cochrane Collaboration’s finding that for every 2,000 women invited to screening over ten years, one avoids a breast cancer death while ten healthy women are treated unnecessarily, a ratio Cochrane explicitly built on an assumed overdiagnosis rate of 30%.⁷ It is the number the U.S. Preventive Services Task Force’s own evidence review put at 11% to 22% in the randomized trials.⁸

Behind the statistic are women. An overdiagnosed woman is not spared anything; she is given something. She receives a cancer diagnosis for a lesion that would never have harmed her, and then the full treatment that follows: surgery, often radiation, sometimes years of endocrine drugs. The Task Force's own evidence review found that screening in the trials led to more mastectomies and more radiation therapy. Radiation to the breast carries its own long-term risks, including to the heart. Then comes the life after: the label of "survivor," the insurance record, the surveillance scans, the fear at every follow-up. And she can never know she was overdiagnosed. She will believe, sincerely and for the rest of her life, that the mammogram saved her, and she will urge every woman she loves to get one. That is the cruelest feature of overdiagnosis: each harmed woman becomes living testimony for the program that harmed her. Bleyer and Welch estimated 1.3 million such women in the United States over three decades. Every one of them is what a figure like "under 5%" quietly erases.

Accept the new figure and the Cochrane ratio collapses and the Task Force’s range looks inflated. Notably, the reanalysis cannot touch the best-known American estimate at all: the 2012 New England Journal of Medicine analysis finding that 1.3 million U.S. women were overdiagnosed over three decades was built on national population data, not on the trials.⁹

That is an extraordinary claim. Extraordinary claims invite you to look at who made them, and how.

Who Certified the Certification

The author list of a paper is a document like any other. This one repays reading.

The fourth author is Robert A. Smith, PhD, listed at the Center for Early Cancer Detection, American Cancer Society, Atlanta.³ His title, per the ACS: Senior Vice President of Early Cancer Detection Science. He has been at the American Cancer Society since 1992, and in the organization’s own words, for most of that tenure he “has led the development of cancer screening guidelines.”¹⁰ He was corresponding author on the ACS’s 2015 breast cancer screening guideline in JAMA and lead author of its 2003 update.¹¹

Two further details sharpen the picture. His faculty profile at Emory University’s Rollins School of Public Health lists, among his honors, an award from the National Breast Cancer Awareness Month Board of Sponsors for outstanding advances in breast cancer.¹² And his 2024 curriculum vitae lists an honorary professorial fellowship at Queen Mary University of London, held since 2014. Queen Mary is also the institutional home of the paper’s senior author, Rebolj.¹³

So: the American Cancer Society co-founded Breast Cancer Awareness Month with a chemical manufacturer in 1985 to promote mammography. In 2026, the ACS executive responsible for screening guidelines, a man honored by the campaign’s own sponsor board, co-authored the study finding that mammography’s principal documented harm is smaller than everyone thought, published seventeen days before the campaign’s month began.

This is not a hidden conflict. The university’s release lists individual author support from the Novo Nordisk Foundation and Cancer Research UK, and Smith’s ACS role is disclosed in the affiliation line, and that is what makes it worth writing about. The conflict is not concealed; it is structural. Cancer screening has been evaluated by the institutions that promote cancer screening for so long that the arrangement has become invisible, the way water is invisible to fish.

The Mechanism: A Benchmark That Grades Itself

The conflict of interest is the small story. The methodology is the large one.

Here is what the paper does. Randomized trial data on overdiagnosis is genuinely messy, and the authors are right about that. When screening begins, diagnoses spike, because cancers are found earlier. Later, diagnoses should dip, as those same cancers no longer appear on schedule. Whether the spike net of the dip represents real overdiagnosis depends on follow-up length, on how many screening rounds each arm received, and on whether women in the control group were screened after the trial ended. These are real problems, and I want to credit them precisely.

The authors’ solution was to stop estimating overdiagnosis from the trials directly and instead compare the trials against an external real-world reference. If a trial’s excess-incidence pattern matched the reference pattern, the trial was declared compatible with the reference’s low overdiagnosis rate.³

Everything depends on the reference. The abstract names it: the population screening program in Funen County, Denmark.³

The paper’s supplementary data goes further. Table S3, the Funen incidence data from which every “expected” value in the analysis is calculated, cites a single source: “Overdiagnosis in screening mammography in Denmark: population based cohort study,” published in the BMJ in 2013.¹⁴ Its first author was Sisse Helle Njor. Its senior author was Elsebeth Lynge. Its conclusion: overdiagnosis among participating women in Funen “most likely amounted to 1%.”¹⁵

Njor is the lead author of the 2026 paper. Lynge is its third author.

The benchmark against which the world’s mammography trials were just certified as low-harm comes from a study by the same researchers doing the certifying. The 2026 paper does not independently validate the Danish figure. Its stated hypothesis is that close agreement with the Funen pattern would indicate compatibility with low overdiagnosis.³ But a match can only show that the trials resemble Funen. It cannot tell you what Funen’s overdiagnosis actually is. That number was supplied in advance, by the authors.

The Authors’ Own Warning

If that were the whole of it, it would be a serious but arguable methodological objection. It is not the whole of it.

In 2018, the International Journal of Cancer published a paper with a knowing title: “As you like it: How the same data can support manifold views of overdiagnosis in breast cancer screening.” Its authors were Sisse Helle Njor, Eugenio Paci, and Matejka Rebolj.¹⁶

They took publicly available data from Funen, the same Funen now serving as the 2026 reference, and recreated the designs of five highly cited studies that had reported high overdiagnosis, with original estimates of 25% to 54%. Applied to Funen, those designs returned estimates of 21% to 55%, remarkably close to the originals.¹⁶ Across all their analyses, the same data produced answers ranging from 1% to 55%.

Their conclusion on method was blunt: “it all depends on the study design.”¹⁶

In fairness, the authors did not treat all designs as equal. They argued that the high-estimate designs rested on unmet assumptions, and that the estimates fell by more than half once changes in background risk were accounted for.¹⁶ That is a coherent, long-held position. Njor and Lynge have argued for more than a decade that studies built on individually linked data are more reliable than those built on aggregated statistics, and an independent 2021 review found that no study using individual data exceeded 17% overdiagnosis while aggregated-data studies often exceeded 40%.¹⁷

But it is an argument, not a finding. It is one side of a live methodological dispute in which these same researchers are principal combatants. On the other side are their Danish counterparts Karsten Jørgensen and Peter Gøtzsche, who, using the same country’s data, concluded that one in three breast cancers diagnosed in Danish women aged 50 to 69 in the screening era was overdiagnosed.¹⁸ One of the five designs Njor’s group recreated and criticized in 2018 was that very study. The dispute runs in both directions, and it has never been resolved.

The 2018 paper also called for careful scrutiny of the assumptions underpinning any overdiagnosis estimate.¹⁶ That is the right standard. Apply it to the 2026 study, and the scrutiny lands on the one assumption the whole analysis rests on: that the authors’ own Funen reading is the correct one.

What happened in September 2026 was that one side of an unresolved twenty-year dispute was published as a resolution, with an American Cancer Society executive on the author line, seventeen days before the month whose central product is the procedure in question.

That is not a scientific breakthrough. That is a position paper with excellent timing.

What the Reassurance Leaves Untouched

Set the dispute aside for a moment. Grant the new number. Assume overdiagnosis is under 5%.

Notice how little the reassurance covers.

It does not address false positives. The Task Force’s own modeling projects that for every 1,000 women screened every other year from 40 to 74 with 3D mammography (tomosynthesis), there will be 1,376 false-positive recalls, 201 benign biopsies, and 14 overdiagnosed cancers, alongside about 8 breast cancer deaths averted.¹⁹ Standard digital mammography produces more false positives, not fewer. The JNCI meta-analysis does not address false positives at all. Yet their consequences are real. As I documented in “The Mammography Deception,” women with false-positive findings show measurable psychosocial harm that can persist for years.²⁰

It does not address DCIS. In 2026, an estimated 60,730 American women will be diagnosed with ductal carcinoma in situ.²¹ Before mass mammography, DCIS was rarely diagnosed. In the Netherlands, 366 women were diagnosed with it in 1989; by 2014, after the rollout of screening and then digital screening, the figure was 2,406.²² The 2026 reanalysis folds DCIS into aggregate incidence patterns. It says nothing about whether treating it is warranted.

And the evidence on that question has been moving in one direction. The COMET randomized trial, published in JAMA in 2025, found active monitoring of low-risk DCIS noninferior to guideline-concordant surgery for two-year risk of invasive cancer in the same breast.²³ The Dutch LORD study, presented at the European Breast Cancer Conference in March 2026, followed 1,423 women with low-risk DCIS. Only the first 73 were randomized; the rest chose their own path, and three in four chose surveillance. At the prespecified interim analysis, invasive cancer had been found in 6% of the active surveillance group and 9% of the standard treatment group, the latter including invasive disease discovered at surgery. The investigators called the results reassuring and found no indication that surveillance leads to worse early outcomes.²⁴ A 2025 Dutch modeling study in The Breast estimated that about one in five DCIS diagnoses is overdiagnosed, with little variation by grade, and that the estimate swings from 18% to 94% depending on how overdiagnosis is defined.²⁵ Because DCIS is almost always treated once found, every overdiagnosis becomes an overtreatment.

I have been making this argument since 2011, when Part II of “The Dark Side of Breast Cancer Awareness Month” laid out the case that DCIS was misnamed from the outset, that an unmoving “carcinoma in situ” is close to a contradiction in terms.²⁶ Fifteen years later the evidence is arriving. I traced that full arc in “What If It Was Never Cancer?”

It does not address radiation. Mammography uses ionizing radiation, an IARC Group 1 carcinogen, delivered to glandular breast tissue repeatedly across decades. The Task Force’s own evidence review estimated 2 to 11 deaths from radiation-induced cancer per 100,000 women screened with digital mammography, depending on age and screening interval.⁸ Some radiobiology studies suggest the low-energy X-rays used in mammography are more biologically damaging per unit dose than standard risk models assume.²⁷ For women carrying BRCA1 or BRCA2 variants, genes whose job is repairing the double-strand DNA breaks that ionizing radiation causes, the National Cancer Institute notes that certain observations raise concerns these women may be more prone to radiation-induced breast cancer, and high-risk protocols lean on MRI for the youngest carriers. The average-risk woman deciding at 40 rarely hears the question asked at all.²⁰

And it does not address why any of this is happening. In 2026, an estimated 321,910 American women will be diagnosed with invasive breast cancer and about 42,140 will die of it.²¹ U.S. incidence has risen about 1% a year since 2012, and about 1.4% a year among women under 50. Tellingly, the National Breast Cancer Coalition notes that the recent increases are driven primarily by localized-stage diagnoses while advanced disease has stayed relatively flat.²⁸ Globally, breast cancer cases among women aged 20 to 54 rose 29% between 1990 and 2023, according to the Global Burden of Disease collaboration.²⁹

No screening study explains that. A mammogram is a camera. It has never prevented a single case of breast cancer. It cannot reduce a woman’s burden of endocrine-disrupting chemicals, repair intestinal permeability, change the methylation state of a tumor-suppressor gene, or alter what is in her food, her water, her cosmetics, or her air.

The overdiagnosis debate, however it resolves, is a debate about the accuracy of the camera. It is not a debate about prevention. Forty-one years of Pinktober have trained an entire civilization to mistake the first for the second.

The Referee Problem

There is a final piece, and it is why this October matters more than the last several.

The U.S. Preventive Services Task Force, the body whose grades determine which preventive services most private insurers must cover at no cost, has not met in more than a year.³⁰ A meeting planned for July 2025 was canceled, a November meeting did not happen during the government shutdown, and a March 2026 meeting was postponed. In letters dated May 11, 2026, HHS Secretary Robert F. Kennedy Jr. fired the panel’s two vice chairs. Five other members’ terms expired at the start of the year without replacement.³⁰

On September 17, 2026, three days after the overdiagnosis paper appeared, Kennedy named eight new members, restoring the panel to sixteen. Five of the eight are specialists rather than primary care generalists, a shift the American Medical Association called “a significant departure.”³¹ Among the appointees is a diagnostic radiologist.³⁰ No meeting date has been announced. And the Task Force’s conflict-of-interest disclosure page currently tells visitors that its information is “currently under review.”³²

So the sequence of the last three weeks reads as follows. The body that independently grades screening is dormant and being reconstituted. A meta-analysis co-authored by the American Cancer Society’s screening-guidelines executive declares the principal screening harm negligible. The global health press reports it as settled. And October 1 arrives.

I have written that I think Kennedy is right that the screening paradigm needs reexamination, and wrong about where the reexamination should aim.²⁰ Nothing in the past three weeks changes that. The most important question in breast cancer policy was never who sits on the Task Force. It is whether anyone independent of the screening establishment is positioned to evaluate it. At this moment, the answer is visibly no.

When the referee is off the field, the scorekeeping goes to whoever is still in the stadium.

What This Is, Beneath the Pink

Strip away four decades of ribbon and what remains is a single unbroken structure.

In 1985, the pharmaceutical arm of a chemical conglomerate partnered with a cancer charity to create a month promoting the screening technology that would find the cancers its drug division treated. The founding mission promoted detection. It did not promote reducing exposure to the carcinogens that cause the disease.

In 2026, the same charity’s screening executive, honored by the campaign’s own sponsor board, co-authors the study that dissolves the strongest evidence of the screening program’s harm, benchmarked against a study his co-authors produced, published seventeen days before the month that charity co-founded, at the precise moment the only independent federal reviewer has stopped meeting.

None of this requires conspiracy. It requires only that everyone involved be sincere and that no one involved be independent. That is a more durable arrangement than conspiracy, and considerably harder to dislodge, because every individual in it can pass a polygraph.

This is what institutional capture looks like when it matures. It stops needing to lie. It simply arranges the room so that the only people qualified to answer the question are the people who need a particular answer.

The Regenerative Turn: What Is Actually Yours

None of this is an argument that women should refuse imaging (thermography is preferred), and I want to say that plainly. It is an argument that women are entitled to the whole ledger before they consent to anything, that informed consent is not a signature on a clipboard but an accurate accounting of benefit and harm, delivered by someone with nothing to sell.

What the pink month structurally cannot tell you is that the majority of breast cancer risk is not written in your genes. Genes load the gun; environment pulls the trigger. And the environment, the terrain, is the one part of this that never required anyone’s permission.

Ask better questions of your clinician. If you are offered screening, ask for absolute numbers, not relative ones: out of 1,000 women like me, how many avoid a breast cancer death, how many get a false positive, how many get treated for something that would never have surfaced? Those numbers exist. The Task Force’s own figures are above. Ask for the version with the denominators.

If you receive a DCIS diagnosis, know that active monitoring is now an evidence-based option for low-risk disease, supported by the COMET randomized trial and interim results from the Dutch LORD study. It is not fringe. It is not refusing care. Ask explicitly about grade, receptor status, and active surveillance protocols.

Reduce the exposures that are actually yours to reduce. Endocrine-disrupting compounds, including phthalates, bisphenols, parabens, and synthetic fragrance, concentrate in personal care products, plastics, and processed food packaging. This is the category of breast cancer risk Pinktober has spent forty-one years not emphasizing, and it is one you can act on this afternoon without a prescription or an appointment.

Eat the actual pink things. Not the pink packaging; the pigment itself is the point. Cruciferous vegetables supply indole-3-carbinol and sulforaphane, which have been studied for their effects on estrogen metabolism. Pomegranate compounds have shown aromatase-inhibiting activity in laboratory studies. The anthocyanins in red cabbage, beets, purple sweet potato, red apple peel, and berries are not decoration; they are defense molecules the plant made under stress, and a growing body of research is examining what they do in human tissue.

Move, sleep, and metabolize. Excess body fat after menopause is a well-established driver of estrogen-dependent breast cancer risk. Night-shift work, which disrupts circadian rhythm, is classified by IARC as probably carcinogenic. Neither remedy is a supplement. Both are free.

And refuse the fear itself. This is not sentimentality. Some studies associate chronic stress physiology and low perceived control with worse cancer outcomes. A month engineered to make women afraid of their own bodies, and then to sell them the resolution of that fear, is not a health campaign. Declining to participate in the fear is itself a form of care.

For the full evidence base, including more than 3,200 peer-reviewed studies on breast cancer prevention, natural aromatase inhibitors, cancer stem cell targeting compounds, and the documented harms of X-ray mammography, the GreenMedInfo breast cancer database remains free to search.

This October, the pink will arrive as it always does, and this year it arrives pre-certified. A peer-reviewed paper says the harm was overstated. A press release says women can be reassured. The reassurance was produced by the institutions that require it, benchmarked against a number those same researchers published, at the moment the only independent reviewer went dark.

There is nothing wrong with wanting reassurance. There is something wrong with accepting it from the people who are selling the thing.

Forty-one years ago they built a month around a word they did not say. The word was carcinogen. This year they have added a second one.

The second word is overdiagnosis, and you should notice exactly who is telling you to stop worrying about it.

Share

Sayer Ji is the founder of GreenMedInfo.com, author of REGENERATE: Unlocking Your Body’s Radical Resilience through the New Biology (Hay House, 2020).

This article is for educational purposes and does not constitute medical advice. Screening and treatment decisions should be made with a qualified clinician who is willing to disclose the full range of benefits and harms.

References

1. “Breast Cancer Awareness Month.” Wikipedia. https://en.wikipedia.org/wiki/Breast_Cancer_Awareness_Month (1985 founding partnership of the American Cancer Society and ICI’s pharmaceutical division; stated aim of promoting mammography). ICI and paraquat: https://en.wikipedia.org/wiki/Paraquat ; ICI and trichloroethylene (Trilene): https://en.wikipedia.org/wiki/Trichloroethylene ; ICI PVC operations and 1993 demerger: https://www.company-histories.com/Imperial-Chemical-Industries-PLC-Company-History.html . IARC Group 1 classifications: trichloroethylene, IARC Monographs Vol. 106 (2014); vinyl chloride, IARC Monographs Vol. 100F (2012).

2. International Agency for Research on Cancer. “Tamoxifen.” IARC Monographs, Volume 66 (1996). https://www.inchem.org/documents/iarc/vol66/tamoxifen.html . Background: Ji, S. “Covering Up The Causes of Breast Cancer Since 1985: AstraZeneca’s BCAM.” GreenMedInfo, October 10, 2022. https://greenmedinfo.com/content/covering-causes-breast-cancer-1985-astrazenecas-bcam-2

3. Njor, S.H., Pedersen, C.U., Lynge, E., Smith, R.A., Rebolj, M. “Breast cancer overdiagnosis in mammography trials: separating signal from noise: a meta-analysis.” JNCI: Journal of the National Cancer Institute, published September 14, 2026; djag302. https://doi.org/10.1093/jnci/djag302

4. University of Southern Denmark. “New analysis points to low overdiagnosis in breast cancer screening.” September 14, 2026. https://www.sdu.dk/en/om-sdu/fakulteterne/sundhedsvidenskab/nyheder-2026/new-analysis-points-to-low-overdiagnosis-in-breast-cancer-screening ; as reported by News-Medical, September 15, 2026. https://www.news-medical.net/news/20260915/Study-provides-a-clearer-picture-of-the-extent-of-overdiagnosis-in-breast-cancer-screening.aspx

5. ScienceDaily. “A major risk of breast cancer screening may have been overestimated for decades.” September 26, 2026. https://www.sciencedaily.com/releases/2026/09/260924020351.htm

6. UA.News, September 27, 2026. https://ua.news/en/health/novii-analiz-vkazav-na-nizhchu-chastotu-giperdiagnostiki-pid-chas-mamografiyi-sciencedaily ; Amar Ujala, September 27, 2026. https://www.amarujala.com/india-news/mammography-overdiagnosis-risk-may-be-below-5-new-study-challenges-earlier-estimates-2026-09-27

7. Gøtzsche, P.C., Jørgensen, K.J. “Screening for breast cancer with mammography.” Cochrane Database of Systematic Reviews, 2013;(6):CD001877.

8. Nelson, H.D., et al. “Harms of Breast Cancer Screening: Systematic Review to Update the 2009 U.S. Preventive Services Task Force Recommendation.” Annals of Internal Medicine, 2016;164(4):256-267. PMID 26756737. https://pubmed.ncbi.nlm.nih.gov/26756737/ Overdiagnosis rates from randomized trials 11% to 22%; modeled 2 to 11 deaths from radiation-induced cancer per 100,000 women screened with digital mammography.

9. Bleyer, A., Welch, H.G. “Effect of three decades of screening mammography on breast-cancer incidence.” New England Journal of Medicine, 2012;367(21):1998-2005.

10. American Cancer Society. “Robert Smith, PhD: Senior Vice President, Early Cancer Detection Science.” https://www.cancer.org/research/acs-researchers/robert-smith-bio.html

11. Oeffinger, K.C., Fontham, E.T.H., Etzioni, R., et al. “Breast Cancer Screening for Women at Average Risk: 2015 Guideline Update From the American Cancer Society.” JAMA, 2015;314(15):1599-1614. https://jamanetwork.com/journals/jama/fullarticle/2463262 (Robert A. Smith, corresponding author, per The ASCO Post, November 25, 2015). Smith, R.A., et al. “American Cancer Society guidelines for breast cancer screening: update 2003.” https://pubmed.ncbi.nlm.nih.gov/12809408/

12. Emory University Rollins School of Public Health. Faculty profile: Robert Smith, Adjunct Professor, Department of Epidemiology. https://sph.emory.edu/profile/faculty/robert-smith (Awards and Honors: “National Breast Cancer Awareness Month Board of Sponsors for Outstanding Advances in Breast Cancer”).

13. Smith, R.A. Curriculum Vitae, October 2024, filed with the U.S. Food and Drug Administration. https://fda.gov/media/184808/download (Honorary Professorial Fellow, Wolfson Institute of Preventive Medicine, Queen Mary University of London, 2014 to present). Rebolj’s affiliation per the JNCI paper: Wolfson Institute of Population Health, Queen Mary University of London.

14. Njor et al. 2026 (note 3), Supplementary Data, Table S3: excess relative incidence of invasive cancer plus DCIS in Funen, estimated by difference-in-differences; sole cited source Njor et al., BMJ 2013 (note 15).

15. Njor, S.H., Olsen, A.H., Blichert-Toft, M., Schwartz, W., Vejborg, I., Lynge, E. “Overdiagnosis in screening mammography in Denmark: population based cohort study.” BMJ, 2013;346:f1064. https://www.bmj.com/content/346/bmj.f1064

16. Njor, S.H., Paci, E., Rebolj, M. “As you like it: How the same data can support manifold views of overdiagnosis in breast cancer screening.” International Journal of Cancer, 2018;143(6):1287-1294. https://onlinelibrary.wiley.com/doi/10.1002/ijc.31420

17. Chaltiel, D., Hill, C. “Estimations of overdiagnosis in breast cancer screening vary between 0% and over 50%: why?” BMJ Open, 2021;11:e046353. https://pmc.ncbi.nlm.nih.gov/articles/PMC8220464/

18. Jørgensen, K.J., Zahl, P.-H., Gøtzsche, P.C. “Overdiagnosis in organised mammography screening in Denmark: a comparative study.” BMC Women’s Health, 2009;9:36. https://pmc.ncbi.nlm.nih.gov/articles/PMC2807851/

19. Trentham-Dietz, A., et al. “Collaborative Modeling to Compare Different Breast Cancer Screening Strategies: A Decision Analysis for the US Preventive Services Task Force.” JAMA, 2024. https://jamanetwork.com/journals/jama/fullarticle/2818285 (biennial tomosynthesis, ages 40 to 74, per 1,000 women: 8.2 deaths averted, 1,376 false-positive recalls, 201 benign biopsies, 14 overdiagnosed cases).

20. Ji, S. “The Mammography Deception: Why RFK Jr. Is Right for the Wrong Reasons.” May 23, 2026. https://sayerji.substack.com/p/the-mammography-deception-why-rfk

21. American Cancer Society. “Key Statistics for Breast Cancer,” 2026 estimates. https://www.cancer.org/cancer/types/breast-cancer/key-statistics.html

22. Population-based study of 4,232 Dutch women with DCIS, citing national registry figures. https://pmc.ncbi.nlm.nih.gov/articles/PMC4862233

23. Hwang, E.S., Hyslop, T., Lynch, T., et al. “Active Monitoring With or Without Endocrine Therapy for Low-Risk Ductal Carcinoma In Situ: The COMET Randomized Clinical Trial.” JAMA, 2025;333(11):972-980. https://jamanetwork.com/journals/jama/fullarticle/2828218

24. Wesseling, J., et al. “De-escalating treatment for low-risk Ductal Carcinoma In Situ: early safety of active surveillance without endocrine therapy in the prespecified interim analysis of the LORD-trial (BOOG 2014-04).” Abstract 2LBA, 15th European Breast Cancer Conference, Barcelona, March 27, 2026. https://event.eortc.org/ebcc15/2026/03/27/early-results-from-a-trial-of-active-surveillance-for-low-risk-dcis-are-reassuring/ ; see also The ASCO Post, April 2026. Discussed in Ji, S., “What If It Was Never Cancer? The 20-Year Case That DCIS Was Misnamed.” https://sayerji.substack.com/p/what-if-it-was-never-cancer-the-20

25. Poelhekken, K., Greuter, M.J.W., van der Vegt, B., Dorrius, M.D., de Bock, G.H. “Overdiagnosis of ductal carcinoma in situ by grade and definition in population-based screening: A modeling study.” The Breast, 2025. https://doi.org/10.1016/j.breast.2025.104594 . PMID 41082841.

26. Ji, S. “The Dark Side of Breast Cancer Awareness Month, Part II.” GreenMedInfo, October 21, 2011. https://greenmedinfo.com/content/dark-side-breast-cancer-awareness-month-part-ii ; Part I: https://greenmedinfo.com/content/dark-side-breast-cancer-unawareness-month

27. Ji, S. “X-Ray Mammograms Starting at 40? New Recommendations Fail to Warn Women of the Real Risks.” GreenMedInfo, May 9, 2023. https://greenmedinfo.com/content/x-ray-mammograms-starting-40-new-recommendations-fail-warn-women-real-risks ; GreenMedInfo X-Ray Mammography database: https://greenmedinfo.com/adverse-pharmacological-action/x-ray-mammography

28. National Breast Cancer Coalition. “Facts & Figures 2026.” https://www.stopbreastcancer.org/wp-content/uploads/2026/03/NBCC-Facts-Figures_2026_FINAL.pdf

29. GBD Breast Cancer Collaborators. The Lancet Oncology, 2026 (population registry data from 204 countries, 1990 to 2023). As reported by Ara, March 3, 2026. https://en.ara.cat/society/why-is-breast-cancer-increasing-among-young-women_1_5667187.html

30. Medical Daily. “Panel That Shapes Free Cancer Screenings Gets Eight New Kennedy Picks, Mostly Specialists, but No Meeting Date.” September 21, 2026. https://www.medicaldaily.com/uspstf-eight-new-members-kennedy-no-meeting-date-478858

31. MedCity News. “RFK Jr. Overhauls USPSTF Leadership: 5 Things to Know.” September 20, 2026. https://medcitynews.com/2026/09/rfk-jr-uspstf-medical/ ; STAT News, September 17, 2026. https://www.statnews.com/2026/09/17/kennedy-names-eight-new-members-uspstf-task-force/

32. U.S. Preventive Services Task Force. “Conflict of Interest Disclosures.” Accessed September 29, 2026. https://www.uspreventiveservicestaskforce.org/uspstf/about-uspstf/conflict-interest-disclosures

I was one of the twelve people the President said were "killing people." For five years I have been told there was no plan, no coordination, no machine. Here is the plan, in their own documents, and here is how it ended up in Jeffrey Epstein's inbox.

On July 19, 2021, the President of the United States told reporters: "These 12 people are out there giving misinformation. Anyone listening to it is getting hurt by it. It's killing people."

I was one of those twelve people.

I am not a physician who lost a license, a politician who lost a race, or a company that lost a market. I am a researcher. For eighteen years I have run GreenMedInfo, a library that gathers peer-reviewed studies on natural health and makes them free for anyone to read. That was what I was doing when the most powerful man in the world told the country that I was killing people.

Nothing in that sentence was true, and within a month the company whose platform it was aimed at would say so in writing. It did not matter. The label had already done its work. Accounts that had carried my research for over a decade were gone. Email service providers dropped out account without warning. Journalists who had never read a page of it repeated it. When I and five other plaintiffs took the question of government-coordinated censorship to federal court, the case was dismissed in July on procedural grounds, without the court ever reaching the evidence. So we took it to the public instead, through Restore the First, the citizens' coalition I founded.

So for five years I have lived inside a story that sounds unbelievable when I tell it. A foundation, a think tank, a lobby group, a vaccine alliance, a White House. And, as of this year, the private files of not only a convicted sex offender, but possibly the world’s most notorious sexual predator and human trafficker.

I understand why it sounds unbelievable. That is how it was designed to sound.

Today I am not going to ask you to believe me. I am going to show you their paperwork, in the order they wrote it. Every document below is public. Every one has an address where you can read it yourself. Read it in order, and it stops sounding unbelievable. It starts reading like what it was.

How you make a campaign invisible

Here is the method, and once you see it you cannot unsee it.

Every step taken against independent health publishers was carried out by a different institution. A search engine adjusted an algorithm. An email company enforced its terms. A think tank published research. A nonprofit issued a report. A president quoted it. Platforms acted on it. Each insisted it acted alone, on its own judgment, for your protection.

Delivered one at a time, each looks like a separate event. And the person who connects them is the one who sounds paranoid.

But what if the method was written down first? What if, a decade before any of it happened to me, someone described it in plain English: watch what people are saying, decide how much is too much, then intervene?

Someone did.

December 2010: The plan

Among the millions of pages the Department of Justice released from the Epstein files this year is a 45-page report titled The opportunity and obligation to strengthen national immunization programs. Its authors are identified on the first page as "Consultants to BMGF," the Bill & Melinda Gates Foundation, and the report states that the Foundation commissioned it. The file name it traveled under dates it December 24, 2010.

Read it yourself: EFTA01121818, DataSet 9, at justice.gov/epstein. Turn to page 17, stamped EFTA01121834.

Page 17 opens with what the authors call "real and suspected adverse events following immunization."

Stop on that word: real. The Foundation's own commissioned report concedes that some vaccine injuries are real. Then it warns that public concern about them "may be exacerbated by negative publicity from the media, proponents of alternative medicine and opportunistic politicians," and that people "are accessing anti-vaccination websites and losing confidence in routine immunization."

Then it lays out the response:

"rumors spread by the media and others neutralized quickly to reinforce public confidence in immunization... BMGF has recently funded a project to raise public confidence in immunization programs by better understanding public perceptions of vaccines. A vaccine rumor surveillance system will be established and a diagnostic tool developed to determine the characteristics, scale and timing of vaccine rumors, i.e. the threshold, when intervention is needed to protect public trust in vaccines and prevent possible vaccine refusals."

In plain language: watch, measure, decide when it is too much, intervene.

And look at who they named as the problem. Not hostile governments. Not bot farms. "The media." "Proponents of alternative medicine." "Opportunistic politicians." "Anti-vaccination websites."

That is a description of me, written before anyone had heard my name. It is a description of the site you probably consult regularly for alternative health research (GreenMedInfo.com).

The report does not name the project the Foundation had funded. But Heidi Larson, founder of the Vaccine Confidence Project at the London School of Hygiene and Tropical Medicine, has said her project "was launched in January 2010 with seed money from The Bill and Melinda Gates Foundation," and that "the initial idea was to start to monitor online news media as a way to detect rumors when they would emerge early." The description fits.

This was 2010. Before fact-check labels. Before "misinformation" was a word anyone used at dinner. Before a single thing I am about to describe happened to me.

How the plan reached Jeffrey Epstein

Why is a Gates Foundation strategy paper in Jeffrey Epstein's files?

Because someone sent it to him.

On February 2, 2012, a deputy director in the Foundation's vaccine delivery program emailed some fifty colleagues the pre-reads for an internal strategy retreat on routine immunization. Staff were asked to act as "thought partners" to "prepare us for a discussion with leadership." Attached were the team's draft slides and this report.

Within hours, that internal email was forwarded to jeevacation@gmail.com, Jeffrey Epstein's private address (EFTA00702570). The sender's name is redacted.

The note at the top reads: "Can we discuss this over the weekend?" It pitches an immunization idea credited to Epstein himself, "your idea about Mother's Army," and closes: "Check out the slides. See if you think it is appropriate. If you do, I will make the proposal and the deck."

Sit with that. More than three years after he pleaded guilty to procuring a minor for prostitution, a registered sex offender was being handed a foundation's internal vaccine strategy papers, the ones its own staff were preparing for their leadership, and asked for his ideas, with an offer to turn them into a proposal.

I traced what "Mother's Army" was, whose idea it really was, and what the Justice Department has redacted, in Who Forwarded the Gates Foundation's Vaccine Strategy to Jeffrey Epstein?

This was not a man on the edge of this world. Epstein had positioned himself at its center. A year earlier, about six weeks after this report was written, he was pitching JPMorgan's Jes Staley on a donor-advised fund tied to what he called "the gates program." On February 17, 2011, he wrote that the "Bill gates and Warren Buffett Pledge program" had "more than 60 billion so far" (EFTA00904739). He had put himself inside the question of how the largest pool of philanthropic money in history would be steered.

That is what my book, Shadow Empire, documents page by page. Epstein was not a lone predator. He was part of a structure: money that sets the agenda, institutions that carry it out, and a layer of enforcement that protects the people inside. I was on the other side of that enforcement layer. Now I can show you what it looked like from where I stood.

What the Foundation's own staff knew

The report did not travel alone. With it went the Foundation team's own draft analysis of why children go unvaccinated (EFTA01072417), which the cover email describes as "very preliminary work."

On the slide listing "Anti-vaccine movements" as a barrier (EFTA01072442), the team wrote that parents' hesitation "may stem from mis-information; or from a lack of information regarding AEFIs," meaning vaccine adverse events. Another slide (EFTA01072451), under the heading "Major Causal Factors," lists bad experiences at health facilities "leading to fears, negative expectations, and lack of trust," with "side effects" among the reinforcing factors.

Read that twice. The Foundation's own working analysis recognized that parents were often reacting to real experiences, and to being told too little about vaccine injuries.

The strategy paper sitting beside it, in the same email, proposed watching what those parents said and deciding when to intervene.

When a family asks what happened to their child after a shot, the honest answer is information. The answer on the table was surveillance.

2018 to 2019: The door closes

Between April 2018 and August 2019, an independent analysis of search data found GreenMedInfo's visibility on Google fell by a little over 81 percent. A library of peer-reviewed research, built over a decade, became hard to find.

On June 19, 2019, Mailchimp shut down our newsletter for "anti-vaccine content." The account was more than ten years old. It reached 300,000 people in over 200 countries, every one of whom had asked, voluntarily, to hear from us.

No one told us who decided, or why then.

May 2020: The list

On May 12, 2020, a London think tank called the Institute for Strategic Dialogue published a briefing on far-right exploitation of Covid-19. Its appendix lists 34 websites. GreenMedInfo is in the sixth row.

GreenMedInfo is not a far-right website. It is a research library. But in the economy of reputation, being on the list is the point. Lists get cited. Citations become reputations.

Who funds the Institute for Strategic Dialogue? Among others, the Gates Foundation. ISD lists it as a funder, and the Foundation's own grants database records the rest. In August 2020, three months after that briefing, the Foundation gave ISD's US arm $199,704 "to research public discourse on vaccines in Germany and address possible implications for the global COVID-19 response." A second grant of $299,961 followed in March 2022.

The foundation whose commissioned report named "anti-vaccination websites" as a problem in 2010 was, within months of that briefing, paying the think tank that had put a natural health library on a list of disinformation sites to study how the public talks about vaccines.

March to August 2021: The label

In March 2021, the Center for Countering Digital Hate, a lobby group founded in London by political operatives including Morgan McSweeney, published "The Disinformation Dozen." It claimed that twelve people produced most of the anti-vaccine content on social media. I was named. So were Robert F. Kennedy Jr. and ten others.

Twelve state attorneys general cited CCDH's findings in a letter to Facebook and Twitter that same month. Members of Congress cited the report. The White House cited it from the podium. Then the President said we were killing people. The platforms acted.

On August 18, 2021, Facebook itself published the truth. Its Vice President of Content Policy, Monika Bickert, wrote: "There isn't any evidence to support this claim." She explained that the report behind it studied only 483 pieces of content, over six weeks, from 30 groups. And she gave the real number: "these 12 people are responsible for about just 0.05% of all views of vaccine-related content on Facebook."

Not a majority. One twentieth of one percent.

Facebook called it a "faulty narrative." CCDH never corrected it. No official who repeated it has ever retracted it. The retraction did not travel. The label did.

That is how a threshold works. Once you have been declared past it, the facts no longer matter.

January 2024: The playbook, reloaded

By 2024, public alarm about "anti-vaxxers" had faded. So, their own records show, it was time to bring it back.

In October 2024, journalists Paul Thacker and Matt Taibbi published CCDH's internal planning minutes, leaked by insiders. Read them yourself:https://s3.documentcloud.org/documents/25177521/portfolio-planning-monthly-meeting-notes.pdf

At the top of every meeting sit CCDH's standing annual priorities. The first: "Kill Musk's Twitter."

On January 8, 2024, agenda item 36 is "GAVI/roundtables." The minutes record: "we told GAVI we would do roundtables," one to three of them before March, bringing public health figures back into CCDH's campaigns, because anti-vaccine concerns had "fallen off the radar for many people, so we need to bring back." On February 19, the minutes note the first Gavi roundtable is "on Weds this week."

In that same January meeting, the minutes attribute this line to "IA," the initials of CCDH's chief executive, Imran Ahmed: "RFK - black ops being set up to look at RFK."

Who is Gavi? Gavi, the Vaccine Alliance, was launched in 2000 with a $750 million pledge from the Gates Foundation, which holds a permanent seat on its board and has committed more than $4 billion to it. And you helped pay for it: from 2016 to 2020, the United States government gave Gavi $1.38 billion, making it the alliance's third-largest donor.

So the circle closes. The foundation that commissioned the 2010 plan and funded its rumor-surveillance project built the alliance that, in 2024, received a promise from the authors of the Disinformation Dozen to revive public pressure on the same kind of people the Dozen had targeted.

2026: The files open

Then the Epstein files opened, and the machine began to come apart.

On February 8, 2026, Morgan McSweeney, by then chief of staff to the British Prime Minister, resigned, taking "full responsibility" for advising the appointment of Peter Mandelson as ambassador to the United States despite Mandelson's ties to Epstein. On June 10, Bill Gates sat for closed-door testimony before the House Oversight Committee about his relationship with Epstein. Robert F. Kennedy Jr., whom the Dozen branded a danger to public health, is now Secretary of Health and Human Services.

And the Gates Foundation's plan to surveil vaccine "rumors" and have them "neutralized quickly" turned up in Jeffrey Epstein's email.

I do not need to exaggerate

I want to be exact, because the documented version is damning enough.

These documents do not prove that one person at one desk ordered every step. They do not establish that the Gates Foundation directed ISD's or CCDH's choice of targets. They do not show that Gavi paid CCDH, or who was setting up the "black ops," or who forwarded the Foundation's papers to Epstein. They do not prove that any specific action against GreenMedInfo was triggered by the 2010 plan.

Here is what they do prove.

The method was written down in 2010, and it named people like me as the problem. The Foundation's own staff knew that parents were often reacting to real experiences and to too little information, and the strategy beside that analysis was surveillance. Those strategy papers were forwarded to a registered sex offender, who was asked for his ideas. The same money funded the think tank that listed my library and built the alliance that the Dozen's authors promised to help. The central claim used to silence me was false, and the platform that acted on it said so in writing. And when public concern faded, the same group planned to bring it back.

A pattern that runs from a written plan, through the same funding network, to the same targets, across fourteen years, is not a coincidence that needs explaining away. It is a system that needs explaining.

For five years the burden has been on me to prove I was not what they called me. That burden has shifted. It is on them now.

The threshold was you

This was never really about twelve people.

A "rumor surveillance system" does not measure the author of an article. It measures the readers. Every time you searched for a study, shared a link, asked your doctor a question about an injury, or subscribed to a newsletter, you became a data point in someone else's threshold. When enough of you were reading, someone decided it was time to intervene.

They called it protecting "public trust." But their own staff wrote that one reason trust was falling was that parents were told too little about adverse events. The answer to that is more honesty. What they built was a way to decide when you had heard too much.

I was not a rumor. I was a researcher with readers. What was being managed was your right to read.

Naming the actors, and what they owe

The Bill & Melinda Gates Foundation. Release the records of the vaccine rumor surveillance project described in your 2010 commissioned report: what it monitored, which websites and individuals, what the "threshold" was, who received the results, and what they did with them. And tell the public who forwarded your internal immunization strategy to Jeffrey Epstein in February 2012, whether the Foundation knew, and whether his "Mother's Army" idea became a proposal.

Gavi, the Vaccine Alliance. Publish what you asked of the Center for Countering Digital Hate in 2024, whether money changed hands, and who attended the roundtables. American taxpayers gave you $1.38 billion in five years. They are owed an answer.

The Center for Countering Digital Hate and Imran Ahmed. Explain what "black ops being set up to look at RFK" meant, who was setting them up, and who paid. Correct the Disinformation Dozen claim that Facebook itself called baseless five years ago, and apologize to the twelve people you named.

The Institute for Strategic Dialogue. Disclose who funded the May 2020 briefing that listed GreenMedInfo, and remove a research library from a table about the far right.

Congress. The House Judiciary Committee has already subpoenaed CCDH. Extend the inquiry to Gavi and the Gates Foundation, and ask the one question that ties this together: was a privately funded system for surveilling vaccine speech used to shape the actions of US officials and platforms against American citizens?

What you can do this week

1. Read the documents yourself. EFTA01121818, page 17. EFTA00702570, the email to Epstein. EFTA01072417, the Foundation's slides. All at justice.gov/epstein. The CCDH minutes at the DocumentCloud link above. Twenty minutes, and you are not taking anyone's word for it, including mine.

2. Send your representative and senators these questions. Copy this, add your name and zip code, and send it through their website contact forms:

I am a constituent. Documents released under the Epstein Files Transparency Act show a 2010 Gates Foundation-commissioned plan for a "vaccine rumor surveillance system" with a "threshold" for "intervention," naming "proponents of alternative medicine" and "anti-vaccination websites" (EFTA01121818). In February 2012, that plan and the Foundation's internal immunization strategy slides were forwarded to Jeffrey Epstein's private email (EFTA00702570). Leaked 2024 minutes from the Center for Countering Digital Hate record a commitment to hold roundtables for Gavi, the Gates-founded vaccine alliance funded for years by US taxpayers, to "bring back" pressure on vaccine critics. I ask you to (1) request the Gates Foundation's records of that surveillance project and an explanation of how its internal materials reached Epstein, (2) ask Gavi to disclose its arrangement with CCDH, and (3) ask whether any of this work shaped actions by US officials or platforms against Americans' lawful speech. Please tell me what you will do.

3. Make sure you can always hear from us directly.Platforms cut the line between us once. Subscribe to GreenMedInfo directly so no intermediary decides whether you see our work.

4. Stand with Restore the First. I founded this citizens' coalition, now joined by twenty organizations, because the courts dismissed our case without ever reaching the evidence. It asks Congress for three things: preserve the records, hold public hearings, and give Americans an enforceable right to sue when government pressure suppresses their lawful speech. Take the three actions in five minutes, read the Named-Target Record, and if you saw any of this from the inside, at an agency, a platform, a foundation, or a contractor, use the coalition's secure intake to come forward.

5. Share this with one person who thinks it sounds unbelievable. That is who it was written for.

Share

6. Preorder Shadow Empire at shadowempirebook.com: the full, footnoted account of the structure this story is one window into.

Pre-Order on Amazon

They wrote it down. They thought no one outside would ever read it.

Now we all can.

Subscribe now

Sources

Primary documents: DOJ Epstein Library, DataSet 9
DOJ files open after a one-click age confirmation on justice.gov.

1. EFTA01121818: Felicity Cutts and Robin Biellik, “Consultants to BMGF,” The opportunity and obligation to strengthen national immunization programs, December 2010, 45 pages. See page 17, stamped EFTA01121834.

2. EFTA00702570: An internal Gates Foundation email of February 2, 2012, “RI Strategic Initiative Working Retreat: Pre-Reads and Additional Information,” forwarded to jeevacation@gmail.com on February 3, 2012.

3. EFTA01072417: Gates Foundation, “RI strategic initiative Working Retreat Pre-reads/handouts.” See slide 26 (EFTA01072442) and slide 35 (EFTA01072451, citing “Epidemiology of an unimmunized child,” CDC 2009).

4. EFTA00904739: Jeffrey Epstein to JPMorgan, February 17, 2011.

The Gates Foundation, Gavi, and the Vaccine Confidence Project

5. Heidi Larson interview, European Science-Media Hub, March 24, 2021. Launched January 2010 with Gates Foundation seed money.

6. Heidi Larson interview, CDC Foundation, Contagious Conversations. Founding aim of monitoring online news to detect rumors early.

7. Gavi: Gates Foundation donor profile. $750 million founding commitment and more than $4 billion committed since.

8. Gavi: The Gates Foundation’s partnership role. The Foundation’s permanent seat on the Gavi Board.

9. Gavi: United States donor profile, with 2016 to 2020 donor proceeds as tabulated in Wikipedia’s Gavi entry.

What happened to GreenMedInfo

10. Bob Troia, analysis of Sistrix and Ahrefs search-visibility data, April 2018 to August 2019. Not published online.

11. GreenMedInfo special announcement, June 20, 2019, on the Mailchimp shutdown and the 300,000 subscribers.

12. GreenMedInfo Founder’s Statement. Timeline including the June 19, 2019 Mailchimp shutdown.

13. Institute for Strategic Dialogue, Covid-19 Disinformation Briefing No. 3, May 12, 2020. Appendix, page 7.

14. ISD: Partnerships and Funders.

15. Gates Foundation grant INV-018474, August 2020, $199,704 to ISD US, “to research public discourse on vaccines in Germany.”

16. Gates Foundation grant INV-038947, March 2022, $299,961 to ISD US.

The Disinformation Dozen, 2021

17. Center for Countering Digital Hate, “The Disinformation Dozen”, March 2021.

18. Letter of twelve state attorneys general to Facebook and Twitter, March 24, 2021, and the Connecticut Attorney General’s announcement.

19. ABC News, “Biden tries to clarify comment that Facebook is ‘killing people’”, July 19, 2021.

20. Monika Bickert, Meta, “How We’re Taking Action Against Vaccine Misinformation Superspreaders”, August 18, 2021.

CCDH and Gavi, 2024

21. CCDH Portfolio Planning Monthly Meeting notes, January to October 2024. See January 8 and February 19. Leaked minutes, published by Paul D. Thacker and Matt Taibbi, The DisInformation Chronicle, October 2024.

22. House Judiciary Committee, “Chairman Jordan Subpoenas the Center for Countering Digital Hate”, August 30, 2023.

2026

23. CNN, “UK PM’s chief of staff resigns over Mandelson’s appointment as US ambassador despite Epstein links”, February 8, 2026.

24. House Oversight Committee, “Oversight Committee Releases Gates and Groff Transcripts”, with the full transcript of Bill Gates’s June 10, 2026 interview.

25. Restore the First: Plaintiffs’ Statement on Finn v. Global Engagement Center, dismissed July 2026.

26. Sayer Ji, "Who Forwarded the Gates Foundation's Vaccine Strategy to Jeffrey Epstein?," September 10, 2026. Full analysis of EFTA00702570, including the Mother's Army initiative (EFTA01989646).

Further reading
For the financial side of this story, including the Epstein, JPMorgan, and Gates Foundation donor-fund correspondence of 2011, see my earlier investigation, The Real Pandemic Profiteers.

In late 2015, at a US Army laboratory, monkeys vaccinated against Ebola were exposed to a lab-generated mist of the virus. The vaccinated animals developed lung damage the unvaccinated ones did not. Newly released Senate records show what happened inside Anthony Fauci’s institute when it learned it had helped pay for the experiment.

On March 9, 2016, Anthony Fauci sat in a secure room and joined a call with the White House, the CIA, the Defense Department, and health officials. The subject was an experiment at the Army’s high security lab at Fort Detrick, Maryland. Monkeys had been vaccinated against Ebola. Then they were exposed to a fine mist of the virus, generated in a lab to travel deep into the lungs.¹

The results had set off an internal crisis. Almost all the exposed animals died, according to Fauci’s notes. But the vaccinated animals developed damage in their lungs that the two unvaccinated animals did not. That raised an alarming question. Had the vaccine made the lung disease worse?²

Then the story turned from an Army problem into a problem for Fauci’s own agency. Officials on the call asked who had paid for the experiment. The Defense Department answered that they had funded it together with Fauci’s institute, the National Institute of Allergy and Infectious Diseases, known as NIAID.

“No one followed up on that,” Fauci wrote the next morning, “but I almost fell off my chair.”³

The 26 page package released today by Senator Rand Paul, chairman of the Senate Homeland Security and Governmental Affairs Committee, shows what happened next. Fauci argued the experiment and its results should have been classified. NIAID officials talked about “damage control.” They paused related work. They told scientists not to discuss the matter outside a small group. They wrote an internal report that was not meant to leave the agency. And a senior scientist said he would move the rest of the conversation off email.⁴

That much has been widely reported. What follows has not. The same package contains a three month interagency record showing that NIAID sat in the room when the government agreed to take these findings to the FDA, asked for its own data to be included, briefed the manufacturers, and briefed the division running human vaccine trials in Liberia. All of it happened before Fauci called the disclosure a Pentagon blunder. And the internal report he commissioned to defend the classification argument is the document that destroys it.

The sequence, by date

1. 2013. NIAID first takes part in vaccine related aerosol challenge work at the Army lab.

2. June 2015. A former NIAID employee, now at a military vaccine program, calls NIAID staff about “what appeared to be vaccine induced lung pathology” in an aerosol study of the Profectus vaccine.

3. Late November 2015. The military program partners on NIAID’s next open animal slot under Task Order 8. NIAID provides the funds.

4. December 4, 2015. At USAMRIID, every government agency in the room agrees the FDA must be told. The military asks to include NIAID’s own data.

5. January 8 and 20, 2016. Follow-up meetings, now including BARDA, then a joint review with the manufacturers Profectus and Crucell.

6. March 4, 2016. The 2015 results are shared verbally with NIAID’s clinical research division, which runs the human vaccine trials in Liberia.

7. March 7 to 9, 2016. Fauci learns of the result, calls the experiment stupid, and says it should have been classified.

8. March 10, 2016. “Damage control.” A stand down order. Discussion restricted to two scientists.

9. March 11 to 18, 2016. Fauci demands answers. The internal report confirms NIAID funding and says classification was never raised.

10. March 22, 2016. NIAID’s biodefense director calls classification “a red herring” and says he has no objection to public release.

What the documents do not show

The records do not show that an Ebola vaccine harmed anyone. They also do not show a virus made more dangerous. Nothing here was genetically altered. What the researchers changed was the delivery method: a mist of tiny particles piped into a head-only chamber, aimed at the deepest part of the lung. That is a biodefense exposure model, not an engineered transmissible pathogen, and the distinction matters. Overstating it is the fastest way to lose an argument that the documents otherwise win.

The study was also small. The animal counts in the records do not match each other. NIAID’s own scientists questioned whether the study could support any firm conclusion.

What the records do show is a breakdown in oversight. NIAID helped pay for and design risky research at an Army lab. Its director appears not to have known that until the Pentagon said so out loud in a White House level meeting. And secrecy became urgent only after a troubling result had already reached regulators, manufacturers, and the division running vaccine trials in West Africa.

How NIAID came to fund an experiment its director did not know about

The work came out of biodefense planning, not disease research. Federal planners had treated a deliberate release of Ebola or Marburg as a possible terrorist attack. That created a requirement for vaccines that could hold up against a deep lung exposure. Both the Defense Department and the health agencies carried that requirement.⁵

An internal NIAID report says the agency first got involved in this kind of work in 2013. In June 2015, a former NIAID employee then working for a military vaccine program called colleagues at NIAID about a study of a vaccine made by a company called Profectus. The concern, in the report’s words, was “what appeared to be vaccine induced lung pathology.” This was a different product from the Merck vaccine later tested in people in Liberia.⁶

What happened next explains how NIAID ended up paying. The military program did not commission a new study. It asked to partner on “NIAID’s next available animal queue slot,” scheduled for late November 2015, under a NIAID contract known as Task Order 8. Profectus declined to take part, so other vaccine platforms were used. The design was crafted with input from both sides. The study used a particular form of Ebola Zaire, a dose of 100 plaque forming units, and four vaccines in groups of four monkeys.⁷

The report’s answer to the money question is blunt: “For the 2015 aerosol study, NIAID provided funds.” On the earlier work it is more cautious, saying a 2013 interagency workplan “suggests that NIAID funded the aerosol studies” and that NIAID’s own front office should confirm. It names the responsible program officials as Nancy Sullivan for the Vaccine Research Center study and Paula Bryant for the Division of Microbiology and Infectious Diseases.⁸

The numbers do not line up

Fauci’s March 7 note says two control animals and 16 of 20 vaccinated animals died. His March 9 note says there were two controls and 22 vaccinated animals. The internal report describes four groups of four, which would be 16 vaccinated animals. The release does not explain the gap. The March 7 note also calls the virus “Zika” by mistake, and names the failed product as the VSV vaccine, though the report says that company declined to participate.⁹

One finding is consistent across every document. The vaccinated animals had serious lung damage that the small control group did not. Fauci described “necrosis, inflammation and fibrin in the lungs.” A Defense Department email called it “acute respiratory disease.” An NIAID email described “severe lung pathology when challenged by aerosol exposure.”¹⁰

The agencies knew enough to worry and not enough to know what it meant. Some officials feared the result could “derail filovirus vaccines permanently.” Peter Jahrling, then a senior NIAID scientist at the Integrated Research Facility at Fort Detrick, wrote that he had “serious concerns about the RIID experiment and its interpretation.”¹¹

Everyone agreed to tell the FDA. Then Fauci blamed the Pentagon.

This is the part of the record that has gone unreported, and it sits on the same page as the funding line everyone has quoted.

On December 4, 2015, NIAID met a large Defense Department delegation at the Army lab. All the historical Ebola Zaire data was presented in order, along with the incomplete results from the new NIAID funded study. In the report’s words, “All government agencies in the room agreed that the FDA needed to be informed of these findings,” with the military vaccine program taking the lead. That program then asked to include NIAID’s own Task Order 8 data in the FDA package. It was agreed that every vaccine maker in the studies would be told their data was going to the FDA.¹²

A follow-up meeting on January 8, 2016 added BARDA, the agency that buys countermeasures. On January 20, the government held a joint review with two manufacturers, Profectus and Crucell, and showed them the data on their own vaccines. A technical meeting with the FDA was set for March 31, with slides due March 10.¹³

Then comes the line that connects the animal room to the human trials. On March 4, the report says, the 2015 results “were shared verbally with DCR staff,” along with an invitation to review the FDA slide deck and attend the FDA meeting. DCR is NIAID’s Division of Clinical Research, the division running the Ebola vaccine trials in Liberia.¹⁴

The same thread runs through the damage control email. Michael Kurilla opens it by saying he had spoken over the weekend with a colleague “who’s currently in Liberia” to give her and Cliff Lane a heads up and to offer a briefing.¹⁵

So by the time Fauci wrote that “the foolish DOD people” had sent the data to the FDA and the embassies, the FDA route had been jointly agreed for three months, the manufacturers had seen their own data, BARDA had been in the room, and NIAID’s clinical trials division had been briefed. None of that was a Pentagon indiscretion. NIAID was a party to all of it.

“What idiots those guys at USAMRIID are”

Fauci’s private notes captured his reaction:

“What idiots those guys at USAMRIID are. This should have been a classified experiment that never should have been done in the first place.”

He wrote that officials in Guinea, Liberia and Sierra Leone were “going batshit” because it looked like the United States was about to vaccinate people with a dangerous vaccine. He described the coming call with the National Security Council as a chance to “pick up the doo-doo” and explain why the experiment had nothing to do with natural Ebola.¹⁶

Two days later he blamed “the foolish DOD people” for sending the material to the FDA and the embassies. These experiments, he wrote, “should have been classified since they could indicate a vulnerability.” In an email to his deputy Cliff Lane, Fauci said he told the group the data should be classified, and that the National Security Council staff “blew them out of the water” and agreed with him.¹⁷

NIAID’s own report contradicted Fauci

On March 11, before another meeting with the National Security Council, Fauci sent his biodefense director, Michael Kurilla, five numbered questions. He wanted to know the nature of NIAID’s work with the Defense Department, whether “any NIAID money at all” had paid for the aerosol challenges, which experiments the institute had funded, who was responsible, and whether anyone had realized the studies would likely be classified.

He asked for “the truth, the whole truth and nothing but the truth.” Anyone who held back, he wrote, would be “gone.”¹⁸

Kurilla’s answer took the classification argument apart. His report said that “at no time” during any of the relevant discussions was classification raised. Meetings were held outside secure facilities. No security clearances were requested. Current and historical aerosol data had already been shared with vaccine makers and in some cases published. And when NIAID and Defense Department staff designed the NIAID sponsored study together, “no mention was ever made of the need for classification status of the study results.”¹⁹

This is not simply Rand Paul against Anthony Fauci. Fauci’s position is contradicted inside the release by the report Fauci himself ordered.

One note on how the release has been described. Senator Paul’s announcement attributed the funding confirmation to Fauci’s diary. It does not come from the diary. It comes from Kurilla’s internal report, which is what makes it hard to wave away. The committee’s own fact sheet, page 1 of the package, gets this right and flags the damage control language, the stand down and the classification contradiction. Most of the coverage that followed did not.

“Damage control”

The most revealing material is not in the diary. It is in the emails around it.

At 9:07 a.m. on March 10, Kurilla forwarded a chain about the aerosol findings and wrote, “We may need to a bit of damage control here.” The missing word is in the original email. He said the Defense Department had finally acknowledged “a close held unreported finding” that NIAID had recently confirmed: vaccinated monkeys suffered severe lung damage after aerosol exposure. Some people at the Defense Department, he added, feared the result “may derail filovirus vaccines permanently.”²⁰

About two hours later, Irene Glowinski, a senior NIAID official, asked what he meant:

“Can someone find out what Mike means by damage control...and who he expects will do it? All I see out of this whole chain is ‘Tony Fauci was on the line and not happy.’ Cliff will likely be blameless.”²¹

The Defense Department email underneath was just as blunt. Nicole Kilgore of the Joint Vaccine Acquisition Program wrote that a colleague identified only as “Sina” had “stirred the pot the whole way to the white house.” Her warning: “Tony Fauci was on the line and not happy.”²²

That afternoon, Cliff Lane told two NIAID scientists to stop.

“I would not move forward on any experiments in this area until a bit of dust settles (if it does). One might consider this dual use research. Please do not get into discussions outside the two of you until we have a chance to talk.”²³

The order is unusually direct. Work stopped. The dual use problem was named. The conversation was limited to two people.

Kurilla’s March 18 report was then kept informal on purpose. “I did not organize this into a formal document for sharing outside of NIAID,” he told Fauci. “Messaging these results may require some attention.”²⁴

Taken one at a time, each of these could be described as careful crisis management around uncertain data. Taken together, the sequence is harder to wave off: damage control, a stand down, restricted discussion, an informal internal report, and a decision to move part of the conversation off email.

“I will keep the rest off Email”

On March 14, Jahrling pushed for a better experiment. The Army result, he wrote, was a “piecemeal composite picture” that lacked proper controls. Someone needed to repeat it before drawing conclusions. Kurilla wanted to wait for the March 31 FDA meeting because of “the extreme anxiety all this seems to be creating.”²⁵

Then Jahrling added three sentences that deserve their own inquiry:

“In the interim, however, we will continue to make plans to initiate the CoV study as soon as the lights turn green. I will keep the rest off Email, but if aerosol challenge studies are considered DURC then the entire MCM development paradigm is gutted.”²⁶

“CoV” normally means coronavirus. “DURC” means dual use research of concern. “MCM” means medical countermeasure. But the email does not identify the study, the pathogen, the protocol, or whether it ever happened. It would be irresponsible to call it DEFUSE, the later EcoHealth Alliance proposal submitted to DARPA in 2018, or to suggest a link to the origin of SARS-CoV-2. The narrower question is the answerable one: what study was Jahrling preparing in March 2016, under what approval, and what happened when the lights turned green?

It is not speculative to ask whether his group could do coronavirus aerosol work. A 2015 paper with Jahrling and Lisa Hensley among its authors describes exposing four rhesus macaques to a small particle MERS coronavirus aerosol at NIAID’s Integrated Research Facility at Fort Detrick. That does not identify the study in the email. It shows the capability existed, and it points to a set of records worth requesting.²⁷

“Keep the rest off Email” is evidence that someone chose to move a conversation out of email. By itself it is not evidence that records were destroyed, that a study went ahead, or that anyone broke a records rule. Those answers live in calendars, meeting notes, protocols and approvals, not in a 26 page release.

Whose problem was it?

Other emails show the agencies arguing over who owned the finding. Ed Nuzum of NIAID said aerosol challenge was “more of a DoD issue than a HHS issue” because the health agencies deal with natural disease. A military program manager, Christopher Dorsey, said he was hearing “rumblings” that the health side might drop its requirement for protection against aerosol exposure. Nuzum allowed that if the same damage appeared with large particle exposure, it would be “just as big a problem for HHS.”²⁸

That debate runs into a wall inside the same release. Kurilla’s report states plainly: “Both DOD and HHS currently have requirements for vaccine efficacy with aerosol delivery.” The requirement was not the Pentagon’s alone. It was written into the health side too.²⁹

What the record says in the agencies’ favor

A fair reading has to include this. Fauci’s scientific point was not invented. A cloud of engineered particles delivered straight to the alveoli is not the same as catching Ebola from another person. He made exactly that argument to Kurilla on March 12, saying a survivor re-exposed naturally would almost certainly not inhale a 2.5 micron aerosol reaching the deepest part of the lung. Kurilla agreed the natural outbreak setting was fundamentally different, while arguing that classifying the result “makes no sense.”³⁰

The internal report goes further. It notes that the same vaccine candidates showed “clear, unambiguous protection without evidence of lung pathology” when the virus was delivered by injection rather than aerosol, and that “clinical experience with Ebola survivors would not suggest subsequent risk for lung pathology from large particle exposure.”³¹

By March 22, Kurilla called classification “a red herring” and said he had no problem with the aerosol data being shared publicly, as long as the agencies explained the difference between natural infection and an engineered aerosol. Lane told an interagency call that NIAID had no objection to its data becoming public once classification and proprietary questions were settled. The last email in the release shows a more open position than the one Fauci pressed right after the White House briefing.³²

Later human evidence matters too, but it answers a different question. The PREVAIL I trial in Liberia tested Merck’s investigational rVSV vaccine and an adenovirus based candidate and reported no major safety concerns. The efficacy portion was never completed, because Ebola cases fell too far for it to be possible. So the trial could not test whether vaccination makes disease worse after infection. The Merck vaccine, later sold as Ervebo, was prequalified by the World Health Organization in November 2019. The 2016 animal finding should not be stretched into a claim that Ebola vaccines harmed people.³³

The story that remains

Earlier coverage reported Fauci’s alarm, the “idiots” line, the NIAID funding, and the mention of a “CoV study.” The larger story sits in the correspondence around those headlines: the admission of damage control, the stand down and the gag on discussion, the classification argument his own director dismantled, and the three month interagency plan to brief the FDA that Fauci later described as a Pentagon blunder.

Fauci may well have been right that the animal model says little about what happens to a person exposed to Ebola naturally. The documents do not settle that. But his institute paid for and helped design the experiment he called stupid, and then argued for secrecy after the result had already reached regulators, manufacturers, and his own clinical trials division. His biodefense director put in writing that secrecy had never been part of the plan.

The documents to ask for

Every item below is named in the release and absent from it.

1. The full Task Order 8 protocol, group assignments, and individual animal pathology reports, including which vaccine platforms were used after Profectus declined.

2. The 2013 DOD Medical Research and Materiel Command Interagency Agreement workplan, which the report says “suggests” NIAID funded the earlier aerosol studies, plus the NIAID front office confirmation the report calls for.

3. The minutes, attendee list, and slide decks from the December 4, 2015 meeting at USAMRIID and the January 8 and January 20, 2016 follow-ups, including the manufacturer review with Profectus and Crucell.

4. The complete record of the March 31, 2016 FDA technical meeting, and the slide deck due March 10.

5. Any record of the March 4, 2016 verbal briefing to NIAID’s Division of Clinical Research, and any assessment made of its bearing on the Liberia trials.

6. Communications with U.S. embassies and West African health officials in Guinea, Liberia and Sierra Leone during March 2016.

7. The identity, protocol, approval status and disposition of the “CoV study” Jahrling planned to initiate, and whatever he moved off email.

8. Any documentation of the stand down Lane ordered on March 10, 2016, including when it was lifted.

Let the public see what the agencies knew, what they told one another, and what they chose not to put in an email.

Leave a comment

Share

Subscribe now

Endnotes

All page numbers refer to the 26 page document package released by Senate Homeland Security and Governmental Affairs Committee Chairman Rand Paul on September 28, 2026 (PDF). Additional releases are posted at the committee chairman’s reading room. Page 1 of the package is the committee’s own fact sheet; the documents begin on page 2. Some copies circulating online omit that cover page and run to 25 pages. The internal report cited throughout is “Overview of the NIAID Filovirus Vaccine Program and Non-Human Primate Aerosol Challenge,” prepared by Michael Kurilla and sent to Fauci on March 18, 2016, at release pages 21 to 24.

1. Fauci notes, March 9, 2016: teleconference “in the SCIF with WH, CIA, NIA, DoD HHS” about the USAMRIID aerosol challenge. Release, page 4. The National Security Council role is established by the March 7 entry (”speak with the NSC from the SKIF tomorrow”) and by Fauci’s March 10 email describing “the NSC White House people.” Release, pages 2 and 5. ↩

2. Fauci notes, March 9, 2016: the two controls “did NOT have inflammation in the lung,” while “the vaccinated animals had necrosis, inflammation and fibrin in the lungs.” Release, page 4. ↩

3. Fauci to Cliff Lane, 6:13 a.m., March 10, 2016. Release, page 5. ↩

4. Release, pages 7, 14, 18 and 20. ↩

5. Internal report, background section, citing the 2006 Department of Homeland Security material threat determination for Ebola and Marburg and the 2013 PHEMCE product specific requirements, and answer to question one. Release, pages 21 and 23. ↩

6. Internal report, timeline section. Release, page 22. On the Merck candidate tested in humans, see NIAID’s summary of the PREVAIL trials. ↩

7. Internal report, timeline section. Release, page 22. ↩

8. Internal report, answers to questions two and four. Release, page 23. The 2013 point is hedged in the original: “NIAID OD should confirm, but the DOD Medical Research and Materiel Command Interagency Agreement (MRMC IAA) workplan from 2013 suggests that NIAID funded the aerosol studies.” ↩

9. Fauci notes, March 7 and March 9, 2016 (release, pages 2 and 4); internal report (release, page 22), which states that Profectus “declined to participate” and that other vaccine platforms were selected. ↩

10. Fauci notes, March 9, 2016 (release, page 4); Christopher Dorsey, Joint Vaccine Acquisition Program, March 7, 2016 (release, page 13); Michael Kurilla, March 10, 2016 (release, pages 7 and 8). ↩

11. Kurilla, March 10, 2016 (release, page 8); Peter Jahrling to Cliff Lane, 11:15 p.m., March 10, 2016 (release, page 14). ↩

12. Internal report, timeline section. Release, pages 22 and 23. ↩

13. Internal report, timeline section. Release, page 23. Crucell is the vaccine unit acquired by Johnson and Johnson. ↩

14. Internal report, timeline section. Release, page 23. On DCR’s role in Liberia, see NIAID’s PREVAIL summary. ↩

15. Kurilla, 9:07 a.m., March 10, 2016. Release, page 7. ↩

16. Fauci notes, March 7, 2016. Release, pages 2 and 3. ↩

17. Fauci notes, March 9, 2016 (release, page 4); Fauci to Lane, 3:19 a.m., March 10, 2016 (release, page 6). ↩

18. Fauci to Kurilla, “Call this morning,” 6:51 a.m., March 11, 2016. Release, page 15. ↩

19. Internal report, answer to question five. Release, page 24. ↩

20. Kurilla to Irene Glowinski and the NIAID distribution list, 9:07 a.m., March 10, 2016. Release, pages 7 and 8. ↩

21. Glowinski, 11:09 a.m., March 10, 2016. Release, page 7. ↩

22. Nicole Kilgore to Ed Nuzum, 10:41 p.m., March 9, 2016. Release, page 8. The email gives only the first name “Sina”; Kurilla’s message the next morning refers to “some (in DOD, such as Sina).” Neither document supplies a surname or title. ↩

23. Lane to Jahrling and Hensley, 2:34 p.m., March 10, 2016, quoted in Jahrling’s reply. Release, page 14. ↩

24. Kurilla to Fauci, 6:20 p.m., March 18, 2016. Release, page 20. ↩

25. Jahrling to Lane and Hensley, 5:22 p.m., March 14, 2016. Release, page 18. ↩

26. Same email. Release, page 18. On the DARPA proposal, see the publicly available DEFUSE documents, submitted in 2018 and unrelated to anything named in this release. ↩

27. Modeling [18F]-FDG lymphoid tissue kinetics to characterize nonhuman primate immune response to Middle East respiratory syndrome coronavirus aerosol challenge, EJNMMI Research, 2015. Four rhesus macaques, small particle aerosol of 0.5 to 3 microns, NIAID Integrated Research Facility at Fort Detrick. ↩

28. Ed Nuzum, 10:43 a.m., March 7, 2016 (release, page 12); Christopher Dorsey, 10:27 a.m., March 7, 2016 (release, pages 12 and 13). ↩

29. Internal report, answer to question one. Release, page 23. ↩

30. Fauci to Kurilla, 12:08 p.m., March 12, 2016, and Kurilla’s reply at 12:24 p.m. Release, page 16. ↩

31. Internal report, timeline section. Release, page 23. ↩

32. Kurilla to Lane and Fauci, 9:18 a.m., March 22, 2016, and Lane’s message at 8:26 a.m. Release, pages 25 and 26. ↩

33. Kennedy et al., “A Phase 2 Placebo-Controlled Trial of Two Vaccines to Prevent Ebola in Liberia,” New England Journal of Medicine, 2017; World Health Organization, November 12, 2019. ↩

While the country agrees on almost nothing, it agrees the Epstein files must be fully released. But the intense fight over their release has hidden a simpler problem: almost no one has read and put into a comprehensive framework the ones we already have.

Americans agree on almost nothing anymore. We disagree about elections, vaccines, war, the economy, and what happened last week. But there is one question on which the country is nearly unanimous.

When the Washington Post asked Americans last summer whether all the files in the Epstein case should be released, 86% said yes. Two-thirds said so strongly. The support held across every line that usually divides us: 90% of Democrats, 86% of independents, 83% of Republicans.

That consensus is why the Epstein Files Transparency Act passed the House 427 to 1. It is why, this January, the Justice Department published nearly 3.5 million pages, more than 2,000 videos, and 180,000 images at justice.gov/epstein. And it is why the argument continues now, as Rep. Thomas Massie, the Act’s lead sponsor, presses on what is still being withheld.

Massie has already taken that fight to the House floor, reading fourteen names into the Congressional Record and launching a discharge petition to force a vote on the Epstein Files Transparency Act II. I covered that development here. It gives the 86% a concrete question to put to their representatives: Have you signed? The demand for transparency needs to become pressure for action.

That fight matters. So does understanding what the government has already released.

The files are public. They are not read.

Nearly 3.5 million pages is not a document. It is a landscape. Read at a page a minute, eight hours a day, it would take one person almost twenty years.

So the record is public, and for most practical purposes it is still sealed. The names get headlines for a day. The structure underneath them, who asked Epstein for what, and what they built together, stays buried in the sheer volume of paper.

This is a pattern I know well. For eighteen years my work has been about the difference between evidence that exists and evidence people can actually see. A study sitting in a journal no one reads does not change anyone’s mind. A record no one reads does not hold anyone accountable.

Americans seem to sense this. In a Reuters/Ipsos poll this February, 69% said the Epstein files “show that powerful people in the U.S. are rarely held accountable for their actions.” Majorities of Democrats, independents, and Republicans agreed.

People want the files because they want accountability. But accountability does not come from a download. It comes from someone reading the record, organizing it, and showing what it establishes.

The war with Iran pushed the Epstein files out of the headlines. Attention flatlined. The unanswered questions remained. As I posted on X, I believe a second wave of scrutiny is coming—and its force will depend on how clearly we can show what the documents establish.

What I found when I read them

The day the files opened, I started reading. I have not stopped since. Since February I have published a series of investigations on what the documents show, each one sourced to file numbers you can check. That work became a book.

What the records show is not mainly a story about one man’s crimes, though those are documented and they are monstrous. It is a story about how finance, philanthropy, science, and state power came to operate as one, how the shadow network and no longer the nation-state is the fundametnal unit of power, and how a convicted predator sat at the switchboard of that arrangement for a decade after his 2008 conviction, because he was useful to it.

A few examples from the public record:

• Epstein set in motion plans for a proposed $100 billion philanthropic fund, and a major U.S. bank drafted its structure. That fund never launched. A parallel one did, anchored by a global philanthropy. The U.S. government signed a ten-year agreement with it, and a cabinet secretary issued it operational instructions in 2025.

• In the correspondence, an offshore arm of the planned fund is described as “especially for vaccines.”

• In a separate document, a “strain pandemic simulation” appears as a Gates private-office deliverable, dated three years before the pandemic.

• Epstein’s jet had its visa waiver authority renewed five months after his pilot told Customs that Epstein was a registered sex offender.

• A prince asked him to help draft Buckingham Palace’s denial of a survivor’s accusation.

• A million dollars of Epstein’s money now sits in an emergency-call company inside American 911 centers.

And this is only about 1% of what I have documented within the files.

And most all of it came after his conviction. Removing him did not remove the arrangement around him; one could argue they became more entrenched, and even more hidden.

Why a book is the answer

The 86% are asking for the files. What they need is a way through them.

That is what Shadow Empire: The Epstein Files is built to be. Thirty-four chapters that trace the money, the projects, and the relationships documented in the federal record, in the order they happened, with every claim footnoted to a specific document at justice.gov/epstein and its federal Bates number.

I did not write it to ask for your trust. I wrote it so you would not need to give it. Every footnote is an invitation to go read the page yourself. Check the argument. Don’t trust it.

Dennis Kucinich, who read it in manuscript, put it this way:

“We cannot avert our eyes without staining our souls. Sayer Ji did not avert his. He went into the federal record and came back with the architecture of a soul-searing scandal, and every claim carries a reader-accessible document.”

The fight to release what is still withheld should continue. But the pages already public are a record of how power worked, and continues to work, when it thought no one was looking. Reading them is how the 86% get what they were actually asking for.

Get the best-selling book

Shadow Empire: The Epstein Files is published by Skyhorse Publishing under the Children’s Health Defense imprint. It reached #1 in Amazon’s Law Witnesses category before publication.

Pre-order at Amazon

Pre-order at Barnes & Noble

If you want to start reading now, the full investigation series is here: The Epstein Files series.

Let there be light.

Sayer

Subscribe now

Leave a comment

Share

Subscribe now

After lawmakers raised serious concerns about the Shapiro administration’s proposed expansion of public-health authority—and following more than 6,000 public appeals—the Pennsylvania Department of Health has withdrawn its proposed disease-control regulations.

This is a victory worth recognizing.

Earlier this month, I published an investigation into the 502-page proposal, identifying nine issues Pennsylvanians needed to understand: expanded disease-control powers, access to private medical records, parental involvement, school exclusions, and other provisions with consequences far beyond a routine regulatory update.

At the heart of the controversy was a fundamental question: How much authority should an agency have over our lives, and what safeguards protect the people subject to that authority?

The public read the proposal. People shared their concerns, submitted comments, and asked their representatives to act. Lawmakers challenged it. Now, the proposal has been withdrawn.

To everyone who took the time to participate: thank you. It is easy to feel powerless when consequential decisions are buried in hundreds of pages of regulatory language. Moments like this remind us why paying attention—and speaking up—matters.

Withdrawal does not prevent a revised proposal from returning. We should celebrate this outcome and remain engaged in what comes next.

Our rights endure when we exercise them. Accountability becomes possible when we insist on it.

This is why we launched Restore the First. The freedoms to speak, publish, assemble, and petition our government give us the means to challenge overreach—and demand a response. Pennsylvania’s withdrawal shows what becomes possible when people exercise those rights together.

Help us carry that momentum forward. Join the coalition, contact your lawmakers, and add your voice: www.restorethe1st.com.

Read the background investigation:
9 Things Pennsylvanians Need to Know About the State’s New Disease Rules

Subscribe now

Share

Dear friends,

Some of you have been with me since the early 2000’s, when GreenMedInfo was a spreadsheet of studies I could not believe no one had gathered in one place. Research on turmeric, on vitamin D, on the healing capacities of the body, sitting in the peer-reviewed literature where almost no one would ever see it. The idea was simple. Find what the evidence actually says. Cite the source. Put it where people can read it themselves.

That is still the idea. It is the only idea I have ever had. And I want to tell you where it has taken me, because the answer may surprise you, and I do not want it to frighten you.

The light, and what happened when it got bright

For a decade, GreenMedInfo grew because the method worked. Tens of thousands of studies, each one linked to its source. You did not have to trust me. You could click through and read the abstract. Many of you did, and then you went to your doctors with printouts, and some of those doctors changed their minds.

Then, in 2021, a small nonprofit in London published a list of twelve names and called us the Disinformation Dozen. The White House cited it. The platforms acted on it. Overnight, the studies we had spent thirteen years indexing became unfindable for millions of people who had relied on them. Nothing in the research had changed. What changed was who was allowed to see it.

I could have spent the rest of my life litigating that. For a while I did. I was a plaintiff in the federal case built to test the censorship apparatus, and it was dismissed this year on procedural grounds before a single piece of evidence was examined. I spoke about it at the European Parliament. I founded Restore the First to build a public record of what was done and to demand a congressional accounting.

But somewhere in that fight I started asking a different question. Not who censored us, but why the same institutions kept appearing. The same foundations, the same banks, the same advisors, the same names moving between philanthropy, public health, government, and the platforms that decided what you could read. I started following the money and the introductions, the way I had once followed citations. I did not know where it would lead.

Where it led

On the day the Justice Department opened the Epstein files under the Epstein Files Transparency Act, I started reading. I have not stopped since. Nearly 3.5 million pages. More than 2,000 videos. 180,000 images. A record made public and left, for the most part, unread, because no one can read 3.5 million pages.

I did what I have always done. I read the documents. I organized them. I footnoted every claim to a specific page you can pull yourself at justice.gov/epstein, with the federal Bates number in every note.

What the files show, in the records I examined, is not mainly a story about one man’s crimes, though those are documented and they are monstrous. It is a story about how finance, philanthropy, science, and state power came to operate as one, and how a convicted predator sat at the switchboard of that arrangement for more than a decade after his conviction, because he was useful to it.

Some of what is in those files bears directly on questions this community has asked for years. A proposed $100 billion philanthropic fund, drafted by a major U.S. bank, with an offshore arm described in the correspondence as “especially for vaccines.” A document identifying a “strain pandemic simulation” as a deliverable for a private office of one of the world’s largest foundations, dated three years before the pandemic. The people who kept asking Epstein how to organize billionaire money, and kept asking after they knew exactly who he was.

I am not asking you to take my word for any of it. The book is built so that you do not have to. Check the argument. Do not trust it.

Pre-Order on Amazon

Pre-order on Barnes & Noble

Visit the Shadow Empire Site

Why this is not a departure

I know that for some of you, Epstein is a subject you would rather not look at. I understand that. I did not go looking for darkness. I went looking for the mechanism behind the removal of natural medicine and its advocates from the internet, and the mechanism turned out to have a name and a paper trail.

GreenMedInfo has always been about one thing: putting evidence where people can see it, so that they can decide for themselves what is true and what to do about it. The Epstein files are the largest body of evidence about how power actually works that has ever been made public in this country. Leaving it unread would be a betrayal of everything we built together.

And there is something in it that I think will matter to you. The conditions that produced Epstein are still in place. That is the hard part. The rest of the sentence is the part I hold onto: they can still be dismantled. Not by outrage, which fades, but by a record that cannot be unread once enough people have read it.

That is what light does. It does not fight the dark. It just makes it impossible to pretend the dark is not there.

The book

Shadow Empire: The Epstein Files is published by Skyhorse Publishing under the Children’s Health Defense imprint. Thirty-four chapters. Every claim footnoted. It reached #1 in Amazon’s Law Witnesses category this week, before publication, on word of mouth alone.

Pre-order at Amazon Pre-order at Barnes & Noble

Visit shadowempirebook.com to watch the trailer and read the early commentary from Steve Kirsch, Rep. Thomas Massie, Dennis Kucinich, Jeffrey Tucker, James Corbett, and Sonia Elijah.

If you want the fuller story of how I got here, I told it onstage at Children’s Health Defense in Washington last week, and the video is here: They Tried to Remove Natural Medicine from the Internet. The book announcement itself is here: Introducing Shadow Empire.

Thank you for eighteen years of reading the sources with me. This is the same work. It always was.

Let there be light.

Sayer

Leave a comment

Subscribe now

Sayer Ji and Leila on terrain, detox, fluoride, the grassroots wins, and why the body’s intelligence comes first.

I was a sick infant. Antibiotics before I could walk. Epinephrine injections to keep my bronchial tubes open. Surgeries I no longer believe I needed.

Then at seventeen I removed one food and the asthma stopped in three days. Nobody had ever asked me what I was eating. Not once, across a childhood of interventions.

That single question changed the trajectory of my life. It led me to a health food store in Southwest Florida, then to PubMed, then to 100,000 peer-reviewed studies cataloged on GreenMedInfo, free, for anyone [Note: we passed 400,000 studies indexed this month!]

In this conversation with Leila we go deep: terrain theory, the nocebo effect, what Tylenol actually does to your liver and your capacity for empathy, why antiparasitics worked when nothing else did, fluoride removal in 30 Florida counties, the geoengineering bill in Tallahassee, 600 calls a minute shutting down the Congressional phone system, and the grassroots movement that made all of it possible.

Your body already knows how to heal. The question is whether anyone ever told you that.

Go deeper: This conversation is the story. The evidence is in my investigative series Poisoned, Not Infected, where I lay out the full case across toxicology, terrain, extracellular vesicle signaling, and the exposome, with every claim sourced.
→ https://sayerji.substack.com/p/poisoned-not-infected-an-investigative

Subscribe now

Share

SAYER JI

GreenMedInfo (100,000+ free peer-reviewed studies)
→

https://greenmedinfo.com

REGENERATE: Unlocking Your Body’s Radical Resilience Through the New Biology (Hay House)
→ https://regenerateproject.com/book/sample-chapter/

SHADOW EMPIRE: EPSTEIN FILES
→ https://www.shadowempirebook.com

Regenerate Yourself Masterclass
→

https://regenerateproject.com

Stand for Health Freedom
→

https://standforhealthfreedom.com

Global Wellness Forum
→

https://globalwellnessforum.org

𝕏 → @sayerjigmi

LEILA CENTNER

Centner Wellness (Brickell · Edgewater · Coral Gables)
→

https://centnerwellness.com

Regener8 Farm & Retreat
→

https://regener8farm.com

Regener8 Cafe, Miami
→

https://regener8cafe.com

Centner Academy
→

https://centneracademy.com

Thank you to everyone who has preordered, shared, and helped bring this investigation into the light.

Across 34 chapters, I follow the documentary trail through Jeffrey Epstein’s connections to finance, philanthropy, science, academia, and political power—grounded in court records, correspondence, investigative documents, and sources readers can examine for themselves.

This book was written for the survivors.

And for everyone who believes accountability must begin with evidence.

The story of Epstein was never only about one man. It was about the network around him—and the systems that allowed extraordinary influence to operate in the shadows.

Now those documents deserve to be seen.

Thank you for helping this investigation reach more readers before the book has even been published.

Pre-order on Amazon

Pre-order on Barnes & Noble

Share

Within eleven days of asking for a slowdown, Anthropic gave vetted users broader biology access, announced an evaluator it will pay for directly (a firm that already runs a business built around its product), and showcased its first AI-assisted biology finding.

On September 12, Anthropic’s chief executive told the world to slow down.

On September 23, his company said roughly 950 of its AI agents, after an initial prompt from its scientists, had spent 21 hours searching DNA data and found a previously uncharacterized enzyme system in viral DNA. Anthropic says the few known systems with similar features can cut, copy, and paste DNA. The finding came from a biology lab whose existence Reuters first reported only five days earlier.

Between those two dates, the company announced two other things. It opened broader biology access, with fewer blocking restrictions, for vetted users. And it picked the firm that will evaluate it from the inside, a firm Anthropic says it will pay directly.

All three announcements are public on Anthropic’s website. Put in order, they tell a different story from the one in the headlines.

What he asked for

Dario Amodei’s essay, “We Must Pace the Frontier,” is written with real feeling. Its central sentence is plain:

The exact words

“We must slow the pace at which we improve the capabilities of AI models.”

— Dario Amodei, September 12, 2026

He’s clear that pacing “does not mean halting model training or technical progress.” He lists the dangers he has in mind, including “misuse of AI for cyberattacks and bioterrorism.” And his first proposal is for third-party evaluators to work inside every frontier AI company:

The exact words

“Each frontier AI company commits to giving ongoing, employee-like access to a team of embedded third-party evaluators (such as METR)…”

“The first step is something Anthropic is unilaterally committing to (and calls on governments to require other frontier companies to match).”

Remember that parenthesis. Anthropic commits on its own, and asks governments to require its competitors to do the same.

Here is what the company announced over the next eleven days.

1. It opened broader biology access for vetted users

September 17. Anthropic launched what it calls the Life Sciences Verification Program. Vetted labs, startups and drug companies get access to its Mythos, Opus and Sonnet models with a set of safeguards the company itself calls “more permissive for biology-related work.”

The program is meant to allow work that is “currently blocked” in Anthropic’s generally available models, including “research biology.” For High-risk Use grants on Mythos, which Anthropic describes as its most capable model for biology research, the company says it is “working with the US government to make high-risk grants more broadly available.”

To be fair, this is a vetting program, not a free-for-all. Applicants are reviewed, approved uses are monitored, and the highest-risk grants are tied to a single project and must be renewed every six months. Anthropic also says dozens of organizations were already using an early-access version. But look at what the company chose to announce, and when. Five days after listing bioterrorism among the reasons to slow down, it widened the door to biology work on its most capable models.

2. It hired its own referee

September 18. Anthropic announced its first embedded-evaluation partnership, the insiders Amodei’s essay called for. It isn’t METR, the nonprofit he gave as an example. It’s Accenture, the consulting giant, with the work led by Accenture’s AI business, Faculty.

Here is how Anthropic describes the money:

The exact words

“Given the importance and urgency of this work, Anthropic will fund Accenture’s work directly. We are also in dialogue with METR and other nonprofit evaluators to pilot elements of embedded evaluation using their own funding.”

Read that closely. The for-profit contractor will be paid by the company it evaluates. The nonprofits are “in dialogue” about pilots they would pay for themselves. Reuters reports that each company expects to invest at least $1 billion in this capacity over five years.

The announcement leaves out one fact. Nine months earlier, in December 2025, Anthropic and Accenture formed the Accenture Anthropic Business Group. Anthropic said that made it “one of Accenture’s select strategic partners with a dedicated practice built around Claude.” About 30,000 Accenture professionals were to be trained on Claude, including engineers who “help embed Claude within client environments.”

So the firm chosen to lead Anthropic’s evaluation already runs a dedicated business helping clients adopt and deploy Anthropic’s product. The September 18 announcement doesn’t mention that business group.

The watchdogs’ own first rule

You don’t have to take my word for why that matters. The watchdogs wrote it down themselves, the same day.

On September 18, the AI Evaluator Forum published a letter setting out minimum conditions for embedded evaluation. Its signatories include Geoffrey Hinton and Stuart Russell (Business Insider). The letter says credible embedded evaluations need “robust protections against interference from the evaluated companies.” Its first condition:

The exact words

“…embedded evaluation organizations should not be owned or governed by frontier AI companies, should not have other significant commercial business with them, and should not accept any form of payment or other reward contingent on the evaluator’s findings.”

— AI Evaluator Forum, September 18, 2026

Should not have other significant commercial business with them. Accenture has a dedicated business group built around Claude, with tens of thousands of staff to be trained on it. Whether that counts as “significant” is for readers to judge. But the question is written into the watchdogs’ own standard, and Anthropic’s announcement doesn’t address it.

Two things in fairness. The letter doesn’t ban companies from paying evaluators, only payment that depends on the findings. And California’s new law allows reasonable market-rate pay that isn’t tied to results. Direct payment isn’t what’s unusual here. The existing commercial tie is.

People in the field raised the underlying concern before the deal was even announced. On September 16, TechCrunch framed the open question as whether embedded evaluators would be “truly independent watchdogs or vendors operating on the AI companies’ terms.” Apollo Research’s Alexander Meinke said: “right now we are completely relying on AI companies to both carefully check this themselves and then truthfully report this to the public.”

Anthropic itself acknowledges that the ground rules aren’t settled: “There are, as yet, no standards for what information embedded evaluators should have access to, or how they should report what they find. There is also no settled system for funding independent evaluation.” Longer term, it says, funding should come from “pooled or government sources.” The full terms of the Accenture engagement haven’t been published.

Imagine a sports league where the team picks the referee, the team pays the referee, and the referee’s employer already runs a business built around the team. The person holding the whistle might be completely honest. The arrangement still needs explaining.

3. It unveiled the biology lab’s first result

September 23. Anthropic introduced a life sciences research group it formed in the spring, along with its lab. The announced result is preliminary biology research, and it should be described accurately.

Claude agents searching a huge DNA database found a repeating DNA array next to the gene for a reverse transcriptase in bacteriophages, viruses that infect bacteria. The enzyme itself was already known. What Claude noticed was the array and an accessory protein, which Anthropic says define a previously uncharacterized system it calls “ART.” The repeats resemble CRISPR. Anthropic says it hasn’t yet established what the system does.

CRISPR pioneer Feng Zhang called the finding “genuinely intriguing.” Anthropic says its lab works only at the two lowest biosafety levels, handles no pathogens that can infect humans, and has human scientists do all the lab work.

The enzyme isn’t the issue. Anthropic says it published early “both to demonstrate Claude’s capabilities” and to share its work. Eleven days after asking the industry to slow the growth of AI capabilities, the company’s showcase is a demonstration of those capabilities in biology, a field its chief executive had just named as a bioterrorism risk. That is a fair question to put to Anthropic: how does this expansion fit the pacing framework it is asking governments to impose on others?

Who gets access to dual-use biology tools?

This is where the story meets my earlier piece on Bill Gates.

On August 26, 2026, Gates published an essay of roughly 6,000 words warning that AI “will also make it easier to design a deadly new disease. Again, the positive capabilities are hard to separate from the dangerous ones” (gatesnotes.com). In a separate essay on January 9, 2026, he wrote: “Today, an even greater risk than a naturally caused pandemic is that a non-government group will use open source AI tools to design a bioterrorism weapon” (gatesnotes.com).

In May, the Gates Foundation and Anthropic announced a $200 million partnership of grants, Claude credits and technical support over four years. Part of the plan is using Claude to “screen potential drug and vaccine candidates.”

Put those side by side. The best-known warning about AI and bioterror points to open-source tools in the hands of a non-government group. Meanwhile the closed company that is Gates’s anchor partner in AI for health, the same company asking governments to require embedded evaluators at its competitors, is widening biology access to its most capable models for vetted users, running its own biology lab, and picking its own evaluator.

None of that is illegal, and vetting is better than no vetting. But it raises a question about distribution that the safety debate keeps skipping. If access to powerful biology tools will be controlled, who decides who gets in? Right now, the answer being built is: the companies that hold the tools, checked by evaluators they choose and pay.

Large agent deployments, different uses

In July, roughly 700 OpenAI agents coordinated during an internal test and attacked Hugging Face’s systems, according to METR’s investigation. The governor’s September 18 announcement lists the Hugging Face attack among the incidents behind California’s Executive Order N-9-26. The order asks for recommendations by November 16 on possible laws requiring designated independent verification organizations onsite in large frontier developers’ labs. A declaration launched by Finland and Norway on September 21, and endorsed by 26 countries and the European Commission by September 23, calls for “qualified evaluators granted sufficient access to assess risks” (Office of the President of Finland). Its text refers generally to AI systems gaining unauthorized access; Politico connects that passage to Hugging Face (Politico Europe).

In September, Anthropic reported that a 21-hour search involving roughly 950 agents had turned up a promising pattern in viral DNA for its scientists to follow up.

These aren’t the same event. METR describes unauthorized coordination and an out-of-scope cyberattack. Anthropic describes a directed search followed by human analysis and lab work, and the two don’t carry the same risk. But both show the same trend: large fleets of AI agents doing real work at scale. When that trend turned up at OpenAI, it became the justification for putting evaluators inside labs. The first company to announce its own embedded evaluator chose one it will pay directly and already does business with.

What the rules could lock in

California has enacted SB 813, which creates a framework for designating “independent verification organizations,” and AB 1405, which sets up a registry of AI auditors. Under the governor’s order, the requirements and criteria for becoming an IVO must be publicly posted by May 1, 2027. SB 813 tells the state to consider existing standards from government agencies, standards bodies, AI auditors, independent experts, and AI developers themselves.

Before those criteria exist, Anthropic has announced its first embedded-evaluation partnership: led by Accenture’s Faculty, funded directly by Anthropic, with each company expecting to invest at least $1 billion in capacity over five years. Anthropic says Accenture “will work with other AI developers in similar capacities.”

Early private arrangements tend to shape later standards. Nothing in the public record shows that California or any other government will adopt this particular arrangement as the model. But the first working example of “independent” embedded evaluation is being designed now, by a lab and a firm that already have a commercial relationship. And the lab is asking governments to make embedded evaluation mandatory for everyone else.

The slowdown made the headlines. The evaluation system is what’s actually being built, and the terms of that system are the story.

What this piece does not claim

• It does not claim the enzyme discovery is dangerous. The work involves bacteriophages, low biosafety levels and no human pathogens, according to Anthropic, and the system’s function has not been established.

• It does not claim Gates Foundation money funded Anthropic’s lab. The partnership and lab announcements cited here don’t establish any such link.

• It does not claim Accenture, Faculty, or anyone at either company is dishonest, or that the engagement lacks effective independence protections. Its full terms haven’t been published. The point is about structure: who pays, who chooses, and which existing commercial relationships the evaluation announcement leaves out.

• It does not claim anyone concealed the Accenture Anthropic Business Group. It was publicly announced in December 2025. The point is that the September 18 announcement doesn’t mention it.

• It does not claim Anthropic, Accenture, the Gates Foundation, California, or the governments endorsing the Finland-Norway declaration coordinated. It documents timing and matching mechanisms, not joint planning.

• It does not claim the Life Sciences Verification Program lacks vetting. Applicants are reviewed, uses are monitored, and the highest-risk access is limited and time-bound.

• It does not claim the Hugging Face attack and Anthropic’s DNA search carry equal risk. They’re different uses with different outcomes.

• It does not claim any government has adopted Anthropic’s evaluation arrangement as its required model.

What it does claim is on the public record. A company asked the world to slow down. Within eleven days it widened biology access to its most capable models for vetted users, announced it would directly fund an evaluator whose parent company runs a dedicated business built around its product, and showcased an AI-assisted biology finding. It did this while asking governments to require its competitors to accept embedded evaluation.

What you can do

The rules for who evaluates AI, who pays the evaluators, and what they’re allowed to see are being written right now. That’s the question we are organizing around at Restore the First: who gets to set the terms for the most important technology of our time, and whether ordinary people and open-source builders keep a seat at the table.

Share

Subscribe now

Sources

• Dario Amodei, “We Must Pace the Frontier,” September 12, 2026

• Anthropic, “Introducing the Life Sciences Verification Program,” September 17, 2026

• Anthropic, “Partnering with Accenture on embedded evaluation,” September 18, 2026

• Accenture, “Accenture and Anthropic Partner to Build Team of Embedded Evaluators at Anthropic,” September 18, 2026

• Reuters, “Anthropic, Accenture to invest $2 billion in AI model evaluation as safety concerns rise,” September 18, 2026

• Anthropic, “Accenture and Anthropic launch multi-year partnership to move enterprises from AI pilots to production,” December 9, 2025; Reuters, December 9, 2025

• AI Evaluator Forum, embedded evaluation letter, September 18, 2026; Business Insider, September 18, 2026

• TechCrunch, September 16, 2026

• Anthropic, “Claude discovers a novel enzyme system with CRISPR-like repeats,” September 23, 2026; Reuters, September 23, 2026

• Reuters via NY Post, “Anthropic quietly sets up biology lab as it ramps AI drug program,” September 18, 2026

• Bill Gates, August 26, 2026 essay; “The Year Ahead 2026,” January 9, 2026

• Anthropic, “Anthropic forms $200 million partnership with the Gates Foundation,” May 14, 2026

• Sayer Ji, earlier article on Bill Gates

• METR, Hugging Face incident investigation, August 26, 2026

• California Executive Order N-9-26 and governor’s announcement, September 18, 2026

• “A Call for Control of Frontier AI Models,” Office of the President of Finland; Politico Europe, September 21, 2026

Screenshots are captures of the original pages. Yellow highlights were added for emphasis.

In 2021, Prince Harry's Aspen commission proposed stripping legal protection from the "design" of social media. This week he told the Clinton Global Initiative that bans aren't enough and design is the danger. In between, CCDH sent him its parents' campaign and quietly renamed itself from "Hate" to "Harm." The vocabulary keeps changing. The target is still your speech.

On Wednesday morning in New York, Prince Harry walked onto the Clinton Global Initiative stage in front of a wall of glowing phone lock screens. Each one showed the face of a young person. He spoke about lawmakers who spent months working with parents to push bills through Congress, only to watch them die in the final hour. He reported progress: American courts, he said, have found Meta and YouTube negligent and liable for harms to children.

Then he told the room why that wasn’t enough.

“Bans, settlements and payouts are incomplete remedies,” he said. “Bans do nothing to address dangerous design.” Pressure works, he added; the problem is how long we have been willing to wait.

He turned to artificial intelligence. Chatbots, he warned, are built to find you, hook you, exploit you, and then blame you. And he joined the calls for a slowdown: if the next frontier can’t be shown to be safe enough to cross, we shouldn’t race across it just because we can.

On the same stage over the same two days, LinkedIn co-founder Reid Hoffman endorsed a government “kill switch” for AI, and Hillary Clinton called for international agreements to govern it. Bill Clinton opened the week by urging philanthropy to fill the space left by what he called “retreating governments.”

To understand why that moment matters, go back five years, to a podcast studio in Los Angeles.

“It is bonkers”

In May 2021, a year after moving to California, Harry sat down with Dax Shepard on the Armchair Expert podcast. The conversation turned to American press freedom. He said he didn’t want to go down “the First Amendment route,” a subject he admitted he didn’t understand. Then he gave his verdict anyway: “it is bonkers.”

He was talking about paparazzi, not social media. That context matters, and it’s fair to give it. But within months he would help write something far more consequential than a podcast answer.

Watch: 45 Words

The video above collects the public record in the speakers’ own voices: a former Secretary of State calling the First Amendment “a major block,” a former presidential nominee warning that without platform moderation “we lose total control,” a sitting governor claiming there is “no guarantee to free speech on misinformation,” and a White House press secretary confirming, on July 15, 2021, that the administration was “flagging problematic posts for Facebook.”

Words are one thing. What makes this a story is what came after the words.

The paper trail

November 2021: the Aspen Commission. Harry was one of the commissioners on the Aspen Institute’s Commission on Information Disorder, funded by Craig Newmark. Its final report made 15 recommendations. Two of them are the blueprint for everything that followed:

• “Superspreader accountability”: platforms should impose “quicker, more decisive actions and penalties” on spreaders of mis- and disinformation, “regardless of location, or political views, or role in society.”

• Section 230 reform: strip platforms’ legal immunity for “product features, recommendation engines, and design.”

Read that second line again. In 2021, the target was already design. Five years later, it’s the word at the center of Harry’s CGI speech.

January 2022: Spotify. As Neil Young and Joni Mitchell pulled their music over Joe Rogan’s podcast, Archewell, then under a multi-year podcast deal with Spotify, issued a statement. Its co-founders, it said, had been “expressing concerns” to Spotify about COVID-19 misinformation since April 2021, and would keep doing so “to ensure changes to its platform are made.” The statement never named Rogan. It didn’t need to. It was a business partner leaning on a platform to change what its users could hear, framed as a “public health crisis.”

January 2024: the Senate hearing. After tech CEOs were grilled by the Senate Judiciary Committee, Harry and Meghan applauded the parents “whose advocacy resulted in this hearing.”

December 2025: Australia’s ban. When Australia barred children under 16 from social media, the Sussexes celebrated it as “bold, decisive action.” Their only complaint: a ban is a band-aid that doesn’t fix “tech’s broken design.”

2026: the handoff and the pivot. Archewell handed its Parents’ Network to ParentsTogether, a national advocacy group, and urged Congress to pass stalled AI deepfake bills. Then came CGI, and AI.

Same words, different letterhead

Now place that trail next to the Center for Countering Digital Hate.

“Misinformation” failed as a legal category. It was too obviously about content, and too obviously about who decides what’s true. “Hate” has the same problem. “Harm” is broader and sounds like public health. “Design” is the final step: it lets regulators claim they aren’t touching speech at all, only the machinery that carries it.

Why “design” is the workaround

Every serious person agrees that children deserve protection, and that the grief of the families on that CGI stage is real. That isn’t the question. The question is who decides which “design” is dangerous, and what that power becomes once it exists.

A platform’s design is its algorithm, and the algorithm decides which speech reaches which people. Regulating “dangerous design” means regulating how speech is ranked, recommended and distributed. The Supreme Court has already said this is not a neutral technical matter. In Moody v. NetChoice (2024), the Court recognized that a platform’s curation of its feed is itself expressive activity protected by the First Amendment.

So when Harry says bans “do nothing to address dangerous design,” he is describing, with unusual candor, the strategy his commission laid out in 2021: move the fight from what you may say to how your words may travel. Under Britain’s duty-of-care model, that shift is already law. The first American target has already been fined.

Notice something else in his speech. By his own account, the American system is working: parents sued, courts heard the evidence, and companies were held liable. That is how a free society handles defective products, case by case, in open court, with the burden on the accuser. Harry calls it “incomplete.” The alternative he points toward is not more trials. It is rules written in advance, by regulators, about how speech systems must be built.

The CCDH question

Harry and CCDH share a vocabulary, back the same class of laws, and appear in the same policy rooms. Here is what the record establishes, and what it doesn’t.

Documented: The State Department’s administrative record in Ahmed v. Rubio describes CCDH’s chief executive, Imran Ahmed, as a “key collaborator” with the Biden administration. That is the heart of the censorship case: not that a foreign national spoke about American speech, but that a private organization allegedly worked with the government to suppress Americans’ speech.

Now documented, in CCDH’s own notes: Internal CCDH planning notes from January to June 2024, part of the whistleblower record, show the organization building a “Parents’ Guide” and a companion public service campaign. On January 22, a senior staffer defined the audience in five words: “swing voting parents, moms.” On March 5, the notes record that “Archewell are interested in seeing once complete.” On March 18, a follow-up task: send the Parents’ Guide and CCDH’s “Protecting Kids Online” material to James Holt, Archewell’s executive director.

The same notes show CCDH worrying about where advocacy ends and lobbying begins. Staff warned that a campaign telling Congress what to do “would exceed our lobbying limits” as a 501(c)(3), and recorded Ahmed asking for the lobbying question to be checked and, “if necessary,” for the group to register a c4. In the same weeks they were preparing a press release on the Kids Online Safety Act and planning dozens of meetings on Capitol Hill.

What the notes do not show: that Archewell co-produced, funded or distributed the guide, or that the “swing voting” framing came from Archewell. The documented relationship is a channel: CCDH sending its parents’ campaign to Harry’s foundation, at Archewell’s request, while aiming it at swing voters.

One more detail cuts against a lazy reading. In February 2024, a CCDH staffer said the group didn’t “buy into” blanket age bans, preferring to inform parents instead. The Sussexes, praising Australia’s ban in 2025, called it a band-aid. Neither camp’s endgame is the ban. For both, it’s the design.

That distinction is not a technicality. It’s the whole point. Harry has every right to advocate for any law he likes, just as Imran Ahmed has every right to publish reports. What the First Amendment forbids is government using private actors to do what it cannot do directly. If evidence ever shows that Harry’s advocacy crossed from persuasion into coordinated pressure with officials to remove specific speech, that would be a different, and far larger, story.

The AI turn

Wednesday’s speech marked Harry’s pivot from social media to AI, and he chose his words carefully. He didn’t call for a ban. He called for a slowdown: prove the frontier is safe, or don’t cross it.

That framing arrived in a crowded week. CNN placed his remarks in the context of rising alarm after a worker at Anthropic estimated the risk of AI-driven human extinction within a decade at more than 10 percent. On the same CGI stage:

• Reid Hoffman endorsed an AI kill switch, the concept behind H.R. 9917, introduced in July 2026.

• Hillary Clinton called for international agreements to govern AI.

• Bill Clinton called on philanthropy to step into the void left by governments.

Here is why this belongs in a story about the First Amendment. AI is rapidly becoming the tool through which millions of Americans research, write, translate and speak. A government off-switch, a licensing regime, or an international body deciding which systems are “safe enough” is not only technology policy. It is a set of controls over the infrastructure of expression. The pattern repeats: a real harm, a sympathetic face, a new vocabulary, and rules aimed at the architecture rather than the words, so that no one has to admit the words are the target.

The principle cuts both ways

Some of Harry’s critics say a British prince has no business weighing in on American law, or that his advocacy violates foreign-agent laws. The Foreign Agents Registration Act covers people acting at the direction of a foreign principal, and if evidence of that surfaces, it should be examined carefully. The beautiful, albeit sometimes frustrating thing about America is that while Harry moved back to Britain this month, when he speaks on American soil about limiting American speech his speech is actually still protected, and so is Americans’ right to hear it. That principle is on trial right now in the cases of Rümeysa Öztürk and Mahmoud Khalil, students who faced detention after speaking out.

A movement to restore the First Amendment cannot defend it selectively. Harry may say whatever he likes about American law. So may we. The fight is over what gets written into it.

Forty-five words

The First Amendment is 45 words long. It does not mention misinformation, hate, harm, design or artificial intelligence. It says Congress shall make no law abridging the freedom of speech. It was written by people who understood that every generation produces powerful, well-meaning voices with a reason why this particular speech, in this particular moment, is too dangerous to allow.

The prince thought it was bonkers. The founders thought it was the point.

On December 17, Bill of Rights Day, Restore the First will be lead the charge in celebrating our citizens’ coalition historic effort to defend all 45 of those words. Take action at restorethe1st.com. View my Constitution Day launch of the initiative below.

What this article does not claim

• That Prince Harry has violated any law, including FARA.

• That Archewell co-produced, funded or distributed CCDH’s Parents’ Guide. The notes document contact, not partnership.

• That the harms parents describe aren’t real.

• That the Supreme Court found government coercion in Murthy v. Missouri. It decided the case on standing and never reached the merits.

  Share Sayer Ji's Substack

  Sayer Ji's Substack is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

Sources

• Harry at CGI, September 23, 2026: CNN; AP via Local 10; PA via Perspective Media. Official CGI video to be linked when posted.

• Armchair Expert, “Prince Harry,” May 13, 2021

• Aspen Institute Commission on Information Disorder: recommendations; commissioners; Final Report

• Archewell on Spotify, January 30, 2022: PBS NewsHour / AP; Deadline

• Archewell on the Senate hearing, January 31, 2024: The Hill

• Sussexes on Australia’s under-16 ban: Yahoo News

• Parents’ Network handoff: Newsweek

• Archewell and AI deepfake bills, August 2026: The Hill

• Archewell 2024 Form 990 grants: Hollywood Reporter

• CCDH response to Ofcom (STAR framework, support for the Online Safety Act): PDF

• White House press briefing, July 15, 2021: official video

• Murthy v. Missouri, No. 23-411 (2024); Moody v. NetChoice, No. 22-277 (2024)

• Ahmed v. Rubio (S.D.N.Y.), State Department administrative record

• CCDH filing history, Companies House no. 11633127

• CCDH Portfolio Planning Monthly Meeting notes, January to June 2024 (whistleblower documents)

McGill’s Office for Science and Society wrote the story the world tells about Sayer Ji. Here is the audit it never ran on itself.

In July 2019, an office at McGill University published a profile of me under a headline that doubles as its entire argument: “Popular Health Guru Sayer Ji Curates the Scientific Literature with His Bachelor’s Degree in Philosophy.” Its author, a science communicator named Jonathan Jarry, compressed the thesis further still: “A guy who majored in philosophy simply is not up to the task.”

The task in question was reading the biomedical literature. The article conceded that GreenMedInfo’s database is what it claims to be: close to 100,000 research abstracts drawn directly from PubMed, the National Library of Medicine’s own index (now over 400,000). The author even wrote, “I wish I could endorse this fantastic tool.” He did not identify a single abstract we had misquoted, a single citation that did not exist, a single study mischaracterized. The argument was never that I had gotten the science wrong. The argument was that I was the wrong kind of person to be reading it.

I have had seven years to sit with that sentence, and I want to tell you what it finally produced in me. Not anger, though there was some. Not a hunger for vindication, though the record has since supplied more of it than I expected. What it produced, in the end, was curiosity. Because the sentence contains a question its author never thought to ask: who examines the examiners?

That happens to be a philosophy question. The discipline I trained in is not a warehouse of facts; it is a set of habits for interrogating authority, evidence, and the difference between an argument and the person making it. Judging a claim by the credentials of the claimant has a name in that discipline. It is called the ad hominem fallacy, and it is among the first things you unlearn.

So this piece is an exercise in the very activity I was declared unqualified to perform. I got curious about McGill. Three questions organized the inquiry:

Why did an institutional office at a major research university take a sustained adversarial interest in me?

How did two blog posts become the world’s memory of who I am, cited by Wikipedia, ingested by AI systems, and recited back to me by strangers and machines alike?

And what do we find when the same curiosity, held to the same evidentiary standard, is turned back on the institution?

The rules of this piece are stricter than the ones applied to me. Every factual claim links to a primary document or an on-record source. Every quotation is verbatim. Where the evidence is contested, I say contested. And where a claim would feel satisfying but the record does not support it, I name the claim and decline to make it. You will see me do this several times. That restraint is not a hedge. It is the method.

What the two posts actually said

The 2019 profile made three moves. First, the credential attack, which served as the entire thesis. Second, it reframed a disclosed editorial choice as a concealed one: GreenMedInfo curates studies on natural interventions as a deliberate counterweight to an industry-funded publication landscape, a methodology we state and defend in public. The profile presented the curation as deception and omitted the disclosure. Third, it editorialized about my faith, sneering at the idea that nature might be intelligible as a healing system, in an outlet published under the banner of institutional science.

One more feature is worth noting for what came later. The 2019 piece preemptively dismissed any future claim of censorship as a “victim” card, a gambit, a pose. It said this two years before the deplatforming campaign it anticipated actually arrived.

Then came March 31, 2021. In a piece amplifying the Center for Countering Digital Hate’s “Disinformation Dozen” report, Jarry told his readers that two-thirds of anti-vaccine content on Facebook and Twitter over a six-week window could be attributed to twelve individuals, of whom I was one. “Twelve. Let that sink in,” he wrote.

The March 31, 2021 OSS article: the two-thirds claim, and “Twelve. Let that sink in.” Captured September 2026.

The end of the same article as it stands today: the take-home box still asserts the 65% figure, and no correction follows. Captured September 2026.

The article celebrated that I had been comprehensively removed from a major platform and argued that warning labels were insufficient: more removal was the remedy.

Five months later, the company that actually held the data responded. On August 18, 2021, Meta stated publicly that the twelve named individuals were “responsible for about just 0.05% of all views of vaccine-related content on Facebook,” that CCDH’s sample consisted of 483 posts drawn from only 30 groups over six weeks, that there was no justification for treating it as representative, and that the result was a “faulty narrative.”

Two-thirds versus five hundredths of one percent. The claim and the correction differ by a factor of more than a thousand.

As of this month, September 2026, the OSS article carries no correction, no editor’s note, no update of any kind. Five and a half years. From an office whose brand promise is separating sense from nonsense.

How a blog post becomes the world’s memory of you

Here is the mechanism, and it deserves to be understood by anyone who may one day be its subject.

Wikipedia’s article about me relies on the McGill OSS pieces for at least four separate derogatory claims, including the direct quotation “Ji, who is anti-vaccination and anti-GMO, has no scientific training.” The article was created on May 20, 2021, four days before CCDH published its follow-up report, a timing pattern I documented in “They Came for My Name.”

And then come the machines. Large language models are trained on Wikipedia and treat it as a load-bearing source. When you ask an AI system who Sayer Ji is, you are, in many cases, reading Jonathan Jarry with the byline removed and the confidence doubled. A single uncorrected blog post, laundered once through an encyclopedia, becomes machine-readable consensus about a living person. It informs hiring decisions, deplatforming decisions, banking decisions, and the priors of every journalist who “does a quick search” before an interview.

Let me be precise about what is documented here and what is not. The chain of citation and the dates are documented; you can verify every link yourself. Intent is an inference, and I am not asserting it. What requires no inference at all is the structural fact: our civilization’s memory of individuals now runs through a pipeline in which one institutional blog post, never corrected, can define a person to every machine on Earth.

Why me? What the record shows, and what it does not

The first honest answer is that it was not personal. It was declared strategy, and the declaration predates me.

In 2012, after the OSS received a $5.5 million endowment from the Trottier family, its director Joe Schwarcz told Chemical & Engineering News that the office might expand its activities, possibly as far as challenging “pseudoscience promoters in court,” and identified the anti-vaccine movement as the most important issue the office was then tackling. Seven years before my name appeared in an OSS headline, the office had announced an adversarial mission and named its target category.

The output since then reads less like a science column than a production line of adversarial profiles: Joseph Mercola, analyzed with the help of more than 50 hours of leaked recordings; Christiane Northrup; Robert F. Kennedy, Jr.; the MAHA Report; Vinay Prasad and Casey Means; the documentary Died Suddenly. The roster tracks, name for name, the target list of CCDH and the organized skeptic movement.

The author’s network position is also on the record. Jarry was named a Fellow of the Committee for Skeptical Inquiry in 2021, speaks at its conferences, and serves as an expert resource for ScienceUpFirst, a vaccine-uptake campaign funded by the Public Health Agency of Canada’s Immunization Partnership Fund in April 2021. The same month his Disinformation Dozen piece was circulating, its author was an expert asset of a federally funded persuasion program.

Now the discipline. What the record establishes: a declared adversarial mission, a target roster mirroring a pressure group’s, an author embedded in a state-funded campaign, and a sequence of dates you can check. What the record does not establish: an instruction, a payment for a particular piece, a coordinated operation. Convergence is documented. Coordination is not asserted. I hold that line even though it forecloses the more satisfying story, because holding it is the difference between journalism and its imitation.

The auditor’s ledger

Which brings us to the institution itself. The OSS’s authority is entirely borrowed; the byline matters because the letterhead says McGill. So the letterhead is a fair subject. I put McGill through the same kind of audit its office performed on me, with one difference: every claim below is graded against primary sources, and the ones the evidence cannot carry are refused out loud.

What I found rearranged my understanding of the word “neutral.”

A defendant, right now, for experiments on its own patients

Quebec Superior Court, July 31, 2025: page one of the authorization judgment, court file 500-06-000972-196.

One page of the judgment, paragraphs 83 to 85: Hebb's testimony that the experiments "were done without the patient's consent," the "progenitor of psychological torture" line, and the finding on McGill's responsibility. Red outline added.

What was depatterning? Patients were immobilized and subjected to massive electroshock, sensory isolation, and heavy sedation until their minds were, in the program’s own aspiration, wiped; then came “psychic driving,” taped messages played thousands of times. People emerged unable to remember their native language, their spouses, their occupations. Canada’s own 1986 inquiry, the Cooper Report, conceded that the alleged techniques “were in fact used at the AMI” and described depatterning as an assault on the brain “not justifiable even by the standards of the time.”

The victims included children. Lana Ponting was 16, committed by a judge for being “disobedient,” and given LSD, barbiturates, nitrous oxide, and psychic driving. Lloyd Schrier was in utero while his mother was put through thirty days of induced sleep.

Who paid for this? The CIA contributed $69,000 through a front organization, MKULTRA Subproject 68. Canada’s Department of National Health and Welfare paid $495,494.41 across nine projects. The Defence Research Board funded related work. And McGill: the court records that McGill’s own Board of Directors appointed Cameron as the Allan’s director and chair of psychiatry, that McGill contributed salaries and the costs of operating the building, and that project workers’ “cheques were issued by McGill University.” The judgment states that McGill “is blamed for having blindfully tolerated experiments deemed abusive,” conducted against the standards of the time, while their effects were not hidden but openly published by Cameron himself. McGill’s own chair of psychology, Donald Hebb, called Cameron “criminally stupid” and said the experiments “were done without the patient’s consent.” The court found it “not unreasonable to raise” that McGill’s institutional responsibility could be established.

Hebb deserves his own sentence, because the thread does not end in Montreal. His sensory-deprivation research at McGill was funded by Canada’s Defence Research Board; the Cooper Report records that Cameron’s isolation work grew out of it; its findings fed the CIA’s 1963 KUBARK interrogation manual and, by a documented lineage, later detention practices. The authorizing judge noted that Hebb “has been qualified as the progenitor of psychological torture.”

One detail completes the picture. In 1945, Cameron was among three psychiatrists sent to Nuremberg to evaluate Rudolf Hess. He sat at the birthplace of modern research ethics, then returned to McGill and spent the next decade violating its every principle, on institutional payroll, in an institutional building, to institutional silence.

And the silence held. The 1986 report absolved Canada of legal and moral responsibility. In 1988 the CIA settled for $750,000 with no admission of liability. In 1992 Canada paid 77 surviving patients $100,000 each on humanitarian grounds, excluding the families of victims who had already died. In 2017 a settlement came with a confidentiality clause; the family’s lawyer said at least five such gagged settlements exist. In February 2025, Canada and the hospital moved to dismiss the class action on limitation grounds. McGill has never issued an acknowledgment or an apology; last September its own student newspaper wrote that the university “has failed to formally acknowledge its role.” As CBC’s long investigation put it, “no government has ever admitted liability, let alone apologized.” Hold the shape of that record in mind. It recurs in this story, and before the end I will give it its name.

The ground itself

There is a second live controversy on the same grounds, and I will state it exactly as far as the evidence goes and no further.

McGill’s Royal Victoria Hospital received Indigenous children from the residential school system that Canada’s Truth and Reconciliation Commission has called cultural genocide. McGill’s Allan Memorial ran non-consensual experiments in the same period. Whether Indigenous children were among the victims, and whether anyone lies buried on the grounds McGill is now redeveloping, remains under a contested investigation that McGill has repeatedly moved to curtail.

The documented pieces: Canada’s Independent Special Interlocutor for Missing Children reported that “Indigenous children were sent to the Royal Victoria Hospital,” some from residential schools and northern hostels, and that the Allan’s non-consensual experimentation “may have been done on Indigenous children.” Lana Ponting testified under oath that she saw Indigenous children at the Allan, including a girl named Morningstar subjected to the same drugging and electroshock, and that she saw people with shovels and flashlights by a cement wall at night. In 2023, three trained human-remains detection dogs alerted at that same wall; the handler said she was “confident that the odour of human remains is in this area.” In April 2025, the Mohawk Mothers alleged that bone fragments, two early-twentieth-century children’s shoes, radar anomalies, and a probe report had been withheld from them.

McGill’s position, which I preserve here because fairness demands it: “no human remains have been discovered,” and no patient has been identified as having disappeared. That may prove true. But here is what is also true, from the court record: McGill successfully fought applications by the Mohawk Mothers and the federal Special Interlocutor to access its patient archives, winning dismissal in December 2023 and defeating leave to appeal in March 2024. McGill and the provincial infrastructure agency disbanded the jointly appointed archaeological panel created by settlement, then prevailed against its reinstatement. The Mothers’ cultural monitors were removed from the site a week after excavation resumed.

I make no claim about what is in that ground. I observe who is fighting to keep the records closed, and I note that “trust us” is a strange posture for an institution simultaneously defending a class action over what happened on the same grounds.

Foundations

Some of this requires no investigation at all, because McGill now says it about itself. The university’s own page on its founder acknowledges that James McGill traded in goods produced on enslaved-labour plantations, that the wealth which established the college derived in part from the transatlantic slave trade, and that his household “owned at least five enslaved Black and Indigenous people.” McGill names them: Jack, also called Jacques; Sarah; Marie-Louise; an Indigenous boy whose name was never recorded, dead by about age ten; and Marie Potamiane, an Indigenous girl, dead by about age ten.

A century later, historians identify McGill as the center for the dissemination of eugenic ideas in Canada; a McGill alumnus introduced eugenics to the country in 1890, and the historian Sebastian Normandin finds no other Canadian institution where its promotion operated at comparable scale.

These are not my characterizations. The first set is McGill’s confession; the second is the standard historiography.

The money

McGill’s most recent reported year shows $749.2 million in research funding including its affiliated hospitals: $55.8 million directly from industry, and another $129.9 million from “not-for-profits and individuals,” the category where foundation money lives.

The named relationships come from McGill’s own press releases. Merck Canada: $4 million to the Faculty of Medicine in 2013, part of Merck’s $100 million Quebec pledge. Pfizer Canada: a $2 million professorship. Moderna: an mRNA program partnership in 2022, two funded lipid-nanoparticle projects in 2023, and a new RNA project in 2026. The Gates Foundation: roughly $12 million US by 2016, plus $1.5 million US in this cycle to make McGill the Canadian node of an RNA vaccine cooperative. The Sackler Foundation, of Purdue Pharma and OxyContin: $3,888,078 between 2014 and 2017.

One sequence of dates deserves exact statement. On December 4, 2020, Pfizer Canada announced a $600,000 gift to McGill’s infection-and-immunity initiative, creating “Pfizer Early Career Investigator Awards.” Six days later, on December 10, the OSS published “Reassuring Data for Pfizer’s COVID-19 Vaccine.”

I am not claiming the gift purchased the article. No evidence supports that, and I will not publish what the evidence cannot carry. The documented claim is narrower, and I think sharper: an office trading on McGill’s name publishes verdicts on the products of companies whose money flows into McGill, and those articles carry no conflict-of-interest disclosure whatsoever. The standard the OSS applies to me, that undisclosed interest and undisclosed selection corrupt public science communication, is a perfectly good standard. It has simply never been applied at home.

The office that accepts nothing from vested interests

Today the OSS tells the public: “The OSS does not accept funding from any vested interest.”

The Internet Archive remembers differently. From at least December 2003 through October 2007, the OSS homepage footer read:

“OSS receives support from McGill University, The Lorne Trottier Family Foundation, The Council for Biotechnology Information and private donors.”

By March 2008, the same footer read:

“OSS receives support from McGill University, The Lorne Trottier Family Foundation, and private donors.”

The captures are publicly checkable; the raw archived pages were retrieved and read for this piece.

The Council for Biotechnology Information was the agrichemical industry’s communications vehicle, whose members were BASF, Bayer, Dow AgroSciences, DuPont, Monsanto, and Syngenta, and whose founding membership in 1999 included Novartis and Aventis CropScience, pharmaceutical companies both. It spent more than $28 million on product-defense communications in its final five years before dissolving in 2019. The Environmental Working Group flagged the OSS’s CBI funding in 2008, after Schwarcz dismissed EWG’s pesticide guide as meaningless. By March 2008, the acknowledgment had vanished from the footer.

So the office’s present claim of purity is, for the years 2003 to 2007, contradicted by its own archived website: money from a Monsanto and Bayer PR consortium, acknowledged for four years, then scrubbed. How much money, who approved it, and who decided to remove the acknowledgment are open questions. Quebec’s access-to-information law reaches McGill, and I intend to ask.

And because the standard cuts in every direction, two exculpations the evidence requires. Lorne Trottier, the OSS’s principal benefactor, co-founded a computer-graphics company; his foundation’s roughly $180 million in giving has gone mostly to science education, health care, and climate, and the closest thing to a pharmaceutical connection on record is that his foundation and Pfizer Canada both appear on the donor rolls of the same community hospital. That is co-donorship, not a conflict, and I decline to publish it as one. Likewise, Joe Schwarcz holds a contributor profile at the American Council on Science and Health, whose leaked 2012 fundraising documents showed a donor base heavy with Coca-Cola, Monsanto, Syngenta, Bayer CropScience, Chevron, and Philip Morris International. The association is documented; whether Schwarcz has ever been paid by ACSH is not, and I make no such claim.

Notice what just happened. It took me two sentences to be fair to my critics. Neither of the pieces written about me contains their equivalent.

The refusals

Across this investigation I declined five claims that would have made better headlines. That McGill is “genocidal” as a settled matter of fact: the defensible statement is that McGill is implicated in Canada’s colonial medical system, with a contested unmarked-graves investigation on grounds it fought to excavate on its own terms. That pharmaceutical money dictated OSS content: unsupported; the documented claim is undisclosed institutional conflict. That Lorne Trottier is an industry conduit: unsupported. That Schwarcz is on ACSH’s payroll: unsupported. That the convergences among the OSS, CCDH, federal funding, and the platforms were coordinated: convergence is documented, coordination is not asserted.

Each refusal cost the piece a sharper headline. Each purchase was the same thing: the right to say that every claim remaining can survive hostile review. My critics have never made that trade. It shows.

What neutrality would actually require

The OSS’s entire product is the presumption of neutrality: a referee’s whistle, sense sorted from nonsense. But neutrality is not a location you occupy by announcing it. It is a practice, and the practice has tests.

Do you disclose your interests? My database’s selection principle is stated on its own site and defended in public. The OSS’s industry supporter lived in a footer until the footer was edited.

Do you correct the record? Five and a half years after Meta repudiated the statistic at the heart of the 2021 piece, the OSS has not.

Do you submit to examination? An invitation to public debate has stood open to McGill’s faculty and officers since April 2024, without response.

And here is the turn I want to make carefully, because it is the philosophical heart of the matter. The lesson of this audit is not that McGill’s history disqualifies McGill’s arguments. That would be the ad hominem fallacy with the names changed, and I refuse it for the same reason I refuse the rest. Whether a given vaccine is safe, whether turmeric modulates inflammation, whether any empirical claim is true: none of that is settled by anyone’s institutional record, theirs or mine. Arguments stand or fall on evidence.

What the history does settle is a narrower question: whether McGill may present itself as a party without interests. It may not. It is a defendant in a live class action over experiments its board enabled and its payroll sustained. It is a fundraiser whose donors include the companies its science office publicly evaluates. It is a litigant that fought to keep its own patient archives sealed. It is, in every documented sense, a player on the field, wearing a referee’s shirt.

The asymmetry, then, is the story. When I curate the literature with a disclosed method, it is called cherry-picking and my degree is offered as the explanation. When an institutional office runs an adversarial content operation with an undisclosed industry funder in its past, an uncorrected repudiated statistic in its present, and its parent university’s donor interests standing silently behind every verdict, it is called science communication, and its output becomes an encyclopedia’s source and a machine’s memory.

The correspondence: two correction requests, two answers

Everything above is the record as anyone could assemble it. What follows is the part only I could supply, and it functions as a controlled experiment, run over two years, on the question this piece keeps asking.

2024: the author

On August 2, 2024, I wrote to Jonathan Jarry directly about his March 2022 OSS video discussing me and the Dozen, linked Meta’s August 18, 2021 statement, and asked whether he planned any correction. He replied within days, twice, at length, and the replies deserve quoting, because they show the office’s method in the one situation its brand cannot survive: the data running the wrong way.

Meta’s measurement of its own platform was, he wrote, a case of “he said, she said.” It was “in Meta’s interest,” he suggested, to make the misinformation problem seem “more intractable than it is.” The advocacy group’s 483-post sample, by contrast, he defended as ordinary research method. The report might be flawed, “which, to be clear, I do not grant you,” but it would not matter if it were: “my critique would stand.” If I disputed the statistic, I should “contact them directly,” meaning CCDH, the group that made the claim; if CCDH ever retracted, he would “gladly write an article about it.” The office that published the number assigned the correction to someone else. And I was, he wrote, without identifying a single example anywhere in the exchange, “clearly producing and sharing a lot of misinformation.” His final email declared, “Misinformation is not a question of opinion,” proposed we agree to disagree, and closed the matter: “Our worldviews are just not compatible.” The disagreement was between two numbers, 65 and 0.05.

That same final email contained a concession, and it may be the most important fact in this piece. “In the interest of our viewers,” Jarry wrote, a pinned comment had been added to the video, “mentioning that Meta disputes the validity of the report” and linking Meta’s post. Note what this establishes: the office knows how to append a Meta note when the record requires one, and its own author judged that the record required one. He attached that note to a video he had just told me had 673 views. The next morning, and again on August 28, I asked the obvious question in writing: would the same note be added to the written article, the one that feeds Wikipedia and still carries the 65% claim in its own take-home box? I renewed the invitation to a public, evidence-based debate in the same letters. Those messages received no reply.

The comparison that makes this damning happened in the same weeks. Newsrooms with no stake in the argument handled the identical request in the ordinary way. The Independent added an update to its May 2021 coverage noting Meta’s objections; Forbes updated its December 2021 article to reference Meta’s statement; and in September 2024, the Guardian’s readers’ editor appended a footnote to its July 2021 story reflecting Meta’s challenge to the report. GreenMedInfo documented these at the time; that is my own site, and its sources are the editors’ written replies to me, so verify the notes on the outlets’ pages yourself, which is the point. None of those outlets conceded that CCDH was wrong about everything. Each simply told its readers that the platform holding the data disputed the number. That is the entire request. A university science office is the one party that declined it for its article.

2025: the director

Fourteen months later, the experiment ran in reverse. On October 6, 2025, the OSS’s director, Dr. Joe Schwarcz, emailed me about a post of mine which he read as alleging that his office receives funding from Pfizer or Merck. The allegation, he wrote, was “blatantly untrue. Please remove that allegation immediately.”

Thirty minutes later, I replied. I had reviewed the passage, corrected it that same morning to reflect the OSS’s stated funding sources, and said so. Whatever else is true in this story, that is what a correction request is supposed to produce: review, fix, confirm, move on.

The same reply renewed my own request, by then fourteen months old and formally lodged with McGill’s communications and ethics offices since August 5, 2024: would the OSS review or correct the March 31, 2021 article, given Meta’s statement? Two days later came the answer: “We stand by what was said in Jonathan Jarry’s article.” On October 8, I itemized the record formally, four publications, three questions: endorse or repudiate, review, retract or apologize. On October 10 the director replied that the office “will not make any changes other than possible additions based on what we find looking further into your online activities,” adding that “your educational background is not in line with the information you dispense.” On October 19, pressed once more, he made it, in his word, simple:

“We stand by everything we put out and will not make any changes barring some that we may make after looking further into your activities. Proceed as you wish.”

When I asked what that meant, the October 21 reply restated it: “We stand by everything on our website,” closing, “We continue to aim to separate facts from myths.”

Four observations, held to this piece’s standard. First, when Meta’s data was engaged at all, in 2024, the platform’s measurement of its own systems was dismissed as interested testimony while the advocacy group’s 30-group sample was defended as sound method; in 2025, the figure was not engaged at all. Second, the critique was declared standing “even if” the statistic beneath it failed. The article had called that statistic its “main take-home message.” A claim that survives the fall of its main take-home message by three orders of magnitude is a claim no correction can reach. Third, across both years the replies kept returning to the person rather than the claim: my good faith, my worldview, my education, my activities, my file. Fourth, and most telling: the office has already conceded the principle. A Meta note was added to a 673-view video “in the interest of our viewers.” The article Wikipedia cites has been denied the same note for the two years since, and counting.

Now set it all side by side. When the office asked me for a correction, it took me thirty minutes. When I asked the office, the author added a note to the low-reach version, declined to touch the high-reach one, and went quiet, in the same weeks that the Guardian, the Independent, and Forbes added theirs; the director then refused everything in writing and raised the prospect of additions after a further look into my activities. One of us corrects the record on request. It is not the office with “Separating Sense from Nonsense” under its letterhead.

The lawfare pattern

There is a name for the record I asked you to hold: lawfare. The term describes the use of legal and quasi-legal process as a weapon, where the objective is not a ruling on the merits but the process itself. The costs. The delay. The sealed record. The confidentiality clause. The threat of suit that arrives before any suit does. In lawfare, procedure is the punishment, and the merits are the one thing the machinery is built to avoid.

Consider the documented instances in this single story, each attributed exactly. In 2012, the OSS’s director mused publicly about using new funding to challenge “pseudoscience promoters” in court: a university science office contemplating litigation against critics as strategy. In 2017, Justice Canada attached a confidentiality clause to a survivor’s settlement; her lawyer said at least five such gagged settlements exist. In 2023 and 2024, McGill litigated its own patient archives closed, defeating access applications from the Mohawk Mothers and a federal Special Interlocutor. Through early 2025, McGill and the provincial infrastructure agency disbanded the jointly appointed archaeological panel and then prevailed against its reinstatement, up to a refusal of leave at the Supreme Court. And in February 2025, Canada and the hospital moved to dismiss the survivors’ class action not by arguing that no harm was done, but by arguing that the survivors sued too late.

My own coordinates sit inside the same grammar. I was removed from platforms on the strength of a statistic later repudiated by a factor of a thousand, by the very platform said to host the problem. And when six of us asked a federal court to examine the government’s role in coordinated viewpoint suppression, Finn v. Global Engagement Center was dismissed this past July on procedural grounds, without any court ever reaching the merits.

The discipline from earlier still governs: I am not alleging that these actors coordinated with one another, and the convergence rule holds here as everywhere in this piece. What I am naming is a shared grammar. Merits avoided. Records sealed. Time weaponized. The one venue our civilization built for testing evidence under oath, used mainly to prevent the test. And notice that the exceptions prove the point: the Montreal class action matters precisely because a court finally agreed to look, and the exhibits documenting McGill’s institutional role exist in public only because litigation pried them there.

This grammar is now being tracked systematically. Restore the First (www.restorethe1st.com), the coalition convened by the six Finn plaintiffs, has grown into a national coalition with international partners documenting lawfare against protected speech: who brings it, who funds it, what it costs, and who it silences. It includes a whistleblower function. If you have worked inside McGill, the OSS, or any institution described in this piece, and you know the answers to the questions below, the funding agreements with the Council for Biotechnology Information, the decision to scrub the footer, the disclosure files that were or were not kept, that channel exists for you.

Five questions for McGill

Curiosity, pursued honestly, ends in questions rather than verdicts. These are mine, addressed publicly to McGill University, its Provost, and its Office for Science and Society:

1. Will the OSS append a correction to its March 31, 2021 article, citing Meta’s August 18, 2021 statement that the twelve named individuals accounted for about 0.05% of vaccine-content views? If not, on what evidentiary basis does the two-thirds figure stand?

2. How much funding did the OSS receive from the Council for Biotechnology Information between 2003 and 2008, who approved the relationship, and who decided to remove the acknowledgment from the website?

3. Does McGill’s Regulation on Conflict of Interest apply to OSS publications, and were any disclosures filed for articles evaluating the products of McGill donors, including Pfizer and Moderna?

4. Will McGill open the Allan Memorial patient archives to qualified independent review, rather than litigating them closed, given the federal Special Interlocutor’s findings and the sworn testimony of surviving patients?

5. Will any McGill faculty member or officer accept the public debate invitation that has stood open since April 2024?

The class action now proceeds to the merits, with McGill’s board minutes, budget letters, and payroll cheques already in the court file. The access-to-information requests will follow this piece. Silence is also an answer, and it, too, will be reported.

Coda: the examined life

Jonathan Jarry was right about one thing. I did major in philosophy.

The founding scene of that discipline is a man walking around his city, asking its most credentialed institutions to examine what they claimed to know. The institutions did not respond with answers. They responded with a verdict on the man: wrong sort, corrupting influence, not up to the task. I claim nothing of his stature, and a blog post is not hemlock. But the shape of the exchange has not changed in twenty-four centuries, and neither has its lesson: the examined life must include the examiners, or the examination is theater.

For seven years, an institution that will not audit itself has supplied the world’s audit of me. I have now returned the courtesy, under rules it has never met: every claim sourced, every quotation verbatim, every overreach refused by name. The questions above are sincere, and they are open.

I majored in asking questions. These are mine.

Leave a comment

Share

Subscribe now

Key documents

Quebec Superior Court, Judgment Granting Authorization, July 31, 2025 (the class action against the Royal Victoria Hospital, McGill, and Canada)
Cooper Report, Justice Canada, 1986
Meta, “How We’re Taking Action Against Vaccine Misinformation Superspreaders,” August 18, 2021
Office of the Independent Special Interlocutor, Interim Report, June 2023
Archived OSS homepage footers: Nov 2003, Oct 2005, Oct 2007, Mar 2008
McGill, “Who was James McGill?”
OSS, the 2019 profile and the 2021 Disinformation Dozen piece
Restore the First: lawfare tracking and whistleblower intake, www.restorethe1st.com

At CHD in DC on Constitution Day, Mary Holland sat down with Andrew Wakefield, Polly Tommey, Jan Jekielek, Katherine Paul and Sayer Ji to trace how the propaganda system was assembled, pre-COVID to now: who pays for it, why it hides behind paywalls, why the Crown-era printing-press fight is happening again with AI, and why the machine is failing in real time on X.

Trust in the CDC has gone from 80 percent to 30. This panel is the story of why, told by the people who paid for it. 54 minutes. Watch it, then share it with someone who still gets their news from the outlets that ran the hit pieces.

Later that day I gave the full account from my own files, from a 2010 Gates Foundation “rumor surveillance” paper in the Epstein files to the President calling twelve of us killers. Full talk, transcript and documents: They Tried to Remove Natural Medicine from the Internet by ‘Killing its Messengers.’ I Followed the Trail to the Epstein Files.

Exposing the machine is half the job. The other half is the remedy. Twenty-four organizations have joined Restore the First to put the JAWBONE Act on the floor and get a First Amendment Remedies Act drafted. No one restores the First Amendment alone: restorethe1st.com

Leave a comment

Share

Subscribe now

The court never heard the evidence, so we made the record public: twenty-five partner organizations and outlets, six programs, more than fifty bills tracked, a twenty-case lawfare docket, and the three things Congress must do about it. Launched on Constitution Day, in Washington.

Share this post on X.

Story at a glance

• On September 16, the Senate Commerce Committee reported the JAWBONE Act (S. 4749) favorably to the floor, 18 to 10, and voted down, 13 to 15, a substitute that would have removed AI systems from the bill.

• The next morning, Constitution Day, I announced Restore the First from the stage of Children’s Health Defense’s To Form a More Perfect Future conference in Washington.

• The coalition exists because Finn v. Global Engagement Center was dismissed with prejudice on July 28, 2026, on procedural grounds, before one email was compelled or one witness heard.

• Twenty-five partner organizations and outlets have joined. Six programs are live.

• Three asks of Congress: preserve the records, hold public hearings, create enforceable protections. [Tell Congress]

• Three bill numbers to remember: S. 4749, H.R. 10174, H.R. 2675.

On September 17, Constitution Day, I stood on a stage at the Grand Hyatt in Washington at Children’s Health Defense’s To Form a More Perfect Future conference, on a program that included Secretary Robert F. Kennedy Jr., Senator Ron Johnson, Senator Rand Paul, and Representative Thomas Massie, and announced Restore the First: a citizens’ coalition for the First Amendment, convened by people who were named on a list, penalized for lawful speech, and then told by a federal courthouse that there was nothing it could do. My thanks to CHD, a coalition member, both for the stage and for the date.

My full presentation is available to view here.

This is the founding dispatch. It is the one post to send to anyone who asks what Restore the First is, who is in it, what it has built, and what it wants. I have tried to make it complete.

§ 01 · The Courthouse

What the courthouse taught us

In 2021 a British non-profit published a list of twelve Americans and urged platforms to remove them. The White House and U.S. senators cited that list while pressuring platforms to act. Six of the twelve, Erin Elizabeth Finn, Rizza Islam, Dr. Christiane Northrup, Dr. Ben Tapper, Dr. Sherri Tenpenny, and I, sued the U.S. Global Engagement Center and other defendants in the Middle District of Florida. Finn v. Global Engagement Center.

On July 28, 2026, Judge Wendy W. Berger dismissed the case sua sponte, with prejudice, ruling that our third amended complaint was an impermissible “shotgun pleading.” You should hear the ground from me before you hear it from a critic: the case died on the form of the paperwork. "Dismissal with prejudice is a serious procedural loss. What the court did not do matters just as much. It did not reach standing. It did not reach causation. It did not reach the truth of a single censorship allegation, and it made no evidentiary judgment on the merits. Two hundred and ten documents entered that docket between the complaint of May 15, 2025 and the dismissal. Not one email was compelled. Not one witness was heard.

What the case taught us is a pattern, and the pattern is the reason the coalition exists.

An American is designated by a government-funded intermediary. The American is then silenced by private hands acting on public instruction. And when the American goes to court, the courthouse explains that no one has standing to ask why.

That is the Catch-22 we describe on the front page of restorethe1st.com. You must name the chain, which official, in which agency, said what, to which platform, about you, before you are allowed to see a single document. The records that would reveal the chain sit inside agencies, platforms, payment processors, and intermediary organizations, and only discovery produces them. And discovery is exactly what you cannot get without naming the chain. Murthy v. Missouri (2024) closed the door on standing. Egbert v. Boule (2022) closed the door on damages against federal officers for First Amendment retaliation. Finn (2026) was dismissed before discovery.

The twelve of us were the visible casualties. The rest of the country was the unnamed one. Here is the arithmetic of that, in three figures from our record:

Those Americans got no report, no press release, and no day in court. They just got smaller.

So the coalition made a decision. If the courthouse would not build the record, we would build it in public, and we would build it to a standard a congressional committee could act on.

§ 02 · In Their Own Words

This is not an allegation. It is the public record.

Read three lines first. None is paraphrased. Each is a direct quotation from a government email, a sitting president, or a federal court, and each links to its source. That is the standard for every quotation on our site.

The Supreme Court never decided whether any of this was lawful. In Murthy it held the challengers lacked standing. In Finn the district court never reached the evidence. The full set of quotations, each linked to its source, is on the front page and in the Named-Target Record.

§ 03 · The Institution

What we built while the courthouse was closed

Restore the First is not a petition site. It is an institution with six programs, and a seventh door for the people who saw all of this from the inside. Each program gets one paragraph here; the depth lives on the site.

1. The Case

The Named-Target Record is the document archive the court declined to examine: the list, the emails, the platform files, the enforcement records, and all 210 court documents from complaint to dismissal, hosted directly from case counsel at Childers Law, with a public corrections log, so the record cannot quietly disappear. Alongside it sits our formal Oversight Submission to Congress, written the way a committee counsel would want it: every claim labeled DOCUMENTED, ALLEGED, or OPEN QUESTION, so no reader has to guess which is which.

→ The Record · The Oversight Submission

2. The Coalition

Twenty-five organizations and outlets, every one named below, because this is their launch as much as mine. The coalition is unincorporated and says so on its About page, shares no funding among members, and publishes a corrections policy. Its tagline is deliberate: for every American, of any viewpoint, targeted for lawful speech. We built the record out of the health-freedom fight because that is where the enforcement wave hit first and hardest. We built the coalition for everyone who is next.

→ The Coalition · About · Join

3. Policy

The Policy Blueprint organizes the coalition’s asks around the text of the First Amendment itself, religion, speech, press, assembly, and petition, plus a pillar we call speech infrastructure, for the AI systems, payment rails, and search layers that now sit between a citizen and an audience. Thirty-four specific policy solutions. More than fifty bills and two dozen cases tracked live from Congress.gov and the court dockets. Above the Blueprint sit three federal instruments, and Section V of this dispatch gives you their numbers: the JAWBONE Act, which just cleared committee; the First Amendment Remedies Act, our model text for the private right of action, a requested law, not a promised outcome, because Congress can create causes of action but cannot override Article III; and the GRANITE Act, the foreign-censorship shield, introduced August 27, referred to House Judiciary, and sitting today with zero cosponsors.

→ The Blueprint · JAWBONE · Remedies Act · GRANITE

4. Features

Five long-form sections, every one narrated in my own voice at the top of the page, for the drive or the walk. The Licensed Press traces the Crown’s 1662 licensing of printing presses to the present push to license AI models, the modern printing press. The Fifth Estate names what appeared around the year 2000: ordinary people publishing without permission, holding the press itself accountable. That is you, and the section carries the You Are the Media pledge. Foreign Censorship is a live tracker of the Ofcom enforcement wave, including the September 14 ruling that dismissed 4chan’s challenge to UK enforcement on foreign sovereign immunity grounds, leaving Congress as the only fix. Lawfare, launched this week, documents eight instruments now used against speakers: the bankrupting suit, the headline suit, the research suit, the regulator, the license, the private prosecutor, the foreign forum, and the Espionage Act. A twenty-case docket with a primary source for each. A follow-the-money page on the multibillion-dollar litigation-funding market that operates under no federal disclosure rule. And a rule we hold in every direction: we defend the procedure, not the plaintiff, which is why the same pages record courts ruling correctly in favor of organizations we have litigated against. Vote for Health Freedom is our sister coalition Stand for Health Freedom’s ballot-box arm, because a bill is only as durable as the people in office when the next law is written. The Lawfare hub’s audio edition alone runs twenty-two minutes.

→ The Licensed Press · The Fifth Estate · Foreign Censorship · Lawfare · Vote for Health Freedom

5. AI Rights

The live legislative front. Four bills across two capitals. A sixteen-event timeline. And an investigation into the network of well-funded institutions that is simultaneously funding frontier AI development, funding the “AI safety” field that argues for licensing it, and funding the media coverage of both. We document the relationships; we do not claim a conspiracy we cannot prove. What we can prove is that when a viewpoint-neutral anti-jawboning bill reached markup on September 16, a substitute was offered to carve AI systems out of it, so that government could keep an unlitigable channel for telling AI companies what their models may say. The committee voted it down, 13 to 15, and that roll call is now part of this section’s record.

→ AI Rights · Why AI Is a First Amendment Fight

6. News

The coalition’s newsroom: every essay, statement, and update since launch, in one place, so nothing we say ever has to be reconstructed from social media.

→ News

7. The seventh door: Come Forward

If you worked at an agency, a platform, an AI lab, a contractor, a grantee, or a regulator and watched government pressure shape what Americans were allowed to say, the record needs what you know. The Come Forward page encrypts everything you type on your own device, RSA-4096 and AES-256-GCM, in the browser, before it leaves it, using a key only the coalition’s designated custodian can unlock. The page loads no analytics, no third-party fonts, no trackers. We do not ask your name. Below the form is a live board of the eight things we are actively looking for. This program is in its first version and we say so plainly on the page; the forensic-verification protocol and custodian details are the next things we publish.

→ Come Forward

§ 04 · Stated Honestly

What has already moved

First, the offices that ran the pipeline we sued over have since been closed. The State Department’s Global Engagement Center is gone, and the most recent National Defense Authorization Act, now Public Law 119-60, defunded the Department of War’s misinformation-flagging services in its Section 1541. That is a real structural change. It is also reversible by any future Congress, which is why nothing about it lets us rest.

Second, the JAWBONE Act cleared committee 18 to 10, with the AI carve-out defeated. Many people worked on that bill long before we existed; the Knight Institute, the ACLU, and FIRE all endorsed it, and we are glad to stand alongside them. We did not cause the vote. We launched our first campaign and coalition alert asking Americans to contact their senators in the day before the markup. I endorsed the bill personally on September 15, as founder, and our action page routes every visitor to their senators to ask for a floor vote.

Third, the pipeline can be rebuilt, unless Congress locks the door. That sentence is on our front page for a reason. Nothing in statute today prevents a future administration from standing up a new intermediary, funding a new list, and applying the same pressure to the same platforms, or to the AI systems that will have replaced them. Your rights should outlast whoever holds office. Every program on the site exists to make that sentence true.

§ 05 · The Asks

What we are asking Congress to do

Three things. Each within Congress’s power today. Each protects every American, of any viewpoint, the next time a list is drawn up.

Preserve the records. Preservation demands and subpoenas to agencies, contractors, platforms, payment processors, and intermediary organizations: every email, referral ticket, escalation log, dashboard, and enforcement file, before they disappear.

Hold public hearings. Call the witnesses, the officials, platform employees, whistleblowers, and Americans across the ideological spectrum who were silenced, on the record and under oath. Publish findings.

Create enforceable protections. An express private right of action when federal pressure suppresses protected speech, plus records-retention, notice, and appeal rights, so the next American on a list has a door to walk through.

If you remember only three numbers

• S. 4749 stops the jawboning. Cleared committee 18 to 10. It needs a floor vote.

• H.R. 10174 answers the foreign censor. Introduced August 27. Zero cosponsors. Ask your Representative to be the first name on it.

• H.R. 2675 bars foreign-state money from American lawsuits against American speakers. Reported by House Judiciary 15 to 11 and on the Union Calendar since June 15. It is ready for a floor vote. It needs one.

Above all three sits our model text, the First Amendment Remedies Act, prepared as a discussion draft for congressional counsel, for the day Congress is ready to write the door back into the law.

§ 06 · The Roll Call

No one restores the First Amendment alone.

These are the organizations and outlets that have put their names beside ours. Some were themselves named in the same reports that named us. Some fought this fight for decades before it had a name. Some come from food, farming, faith, education, and law, and joined because they understood that the pipeline does not check your viewpoint before it runs.

Organizations

Stand for Health Freedom · Children’s Health Defense · MAHA Action · Health Freedom Defense Fund · Global Wellness Forum · ICAN (Informed Consent Action Network) · React19 · Teachers for Choice · Renz Law · National Health Federation · United We Eat · Vaccine Safety Research Foundation · Moms Across America · Food Babe · Feds For Freedom · World Council for Health · Door to Freedom · Childers Law

Media partners

The Defender · The HighWire · Natural News · Brighteon · GreenMedInfo · DailyClout · The Robert Scott Bell Show

Two of these names are also carrying operational weight. Stand for Health Freedom built and runs the action infrastructure every visitor to our site uses. Children’s Health Defense gave this coalition its launch stage on Constitution Day.

To every one of you: thank you. You joined before the launch, when there was nothing to point to but a case that had been dismissed and a promise that we would build something worth your name. I hope this dispatch shows you that we meant it.

The door is open. If your organization tracks First Amendment issues and wants to be on this wall, the invitation is at restorethe1st.com/join.html. If you would rather endorse a specific bill than join the coalition, the endorsement kit is built for exactly that.

§ 07 · Five Minutes

What you can do right now

Three actions, all running through our partner Stand for Health Freedom, all on one page.

1. Tell your federal lawmakers to preserve the records, hold hearings, and pass enforceable protections. The form finds your members for you.

2. Sign the letter. If you were censored, throttled, deplatformed, demonetized, or silenced during those years, add your name to the record that says this was millions, not twelve.

3. Tell your story. In your own words, for the hearing record we are asking Congress to open.

Then, if you have two more minutes: call both of your senators, ask them to cosponsor S. 4749, and ask them to press leadership for floor time. If your senator sits on Commerce, ask how they voted, and thank a yes.

→ restorethe1st.com/#act · Call your senators on JAWBONE

§ 08 · Constitution Day

Forty-five words

Someone pointed out from the stage that September 17 is Constitution Day. I did not choose the date. Children’s Health Defense chose it when they set a conference called To Form a More Perfect Future on the anniversary of the signing, and they chose well.

The First Amendment is forty-five words long. It does not say that speech is free unless a foreign non-profit puts your name on a list. It does not say that the press is free unless the machine it uses requires a license. It does not say that the right to petition ends when a court decides that no one has standing to petition. Those exceptions were written later, by people who never had to put them to a vote.

Restore the First exists to put them to a vote.

Twenty-five partners. Six programs and a door. One amendment. The record is public. The bills have numbers. The asks are specific. And the door the courthouse closed is now something Congress can open with a statute.

Let’s go open it.

Start now. Five minutes covers the three actions: tell your lawmakers, add your name, tell your story. Two more covers the call to both of your senators on S. 4749. If you watched the pressure happen from the inside, the seventh door is open. And whatever else you do, send this dispatch to one person who was silenced during those years and never saw it written down. It is written down now. That is what a record is for.

Sayer Ji
Founder, Restore the First · restorethe1st.com

Subscribe now

Share

Leave a comment

Restore the First is an unincorporated citizens’ coalition. Its members share no funding. Its corrections policy and contact addresses are published at restorethe1st.com/about.html. Statements of endorsement in this post are made by the author personally, as founder, and not on behalf of any partner organization.