Tirzepatide and retatrutide are both chains of 39 amino acids. One comes from a pharmacy, with a prescription. The other is still in Phase 3 trials and comes from a website, under a label that says “Not for human consumption.” (People inject it anyway.) That’s not a hack. It’s an unapproved drug trade, and the bill is starting to come due.

On October 13, CBS News reported that the FDA had received two reports of people who died after taking products sold as retatrutide, Eli Lilly’s experimental triple-hormone agonist for obesity. Neither report has been verified, and neither proves the product caused the death. Over the same months, America’s Poison Centers logged 249 retatrutide exposures in the first quarter of 2026, 460 in the second and 862 in the third. CBS notes that a steady rise is common when poison centers start tracking a new drug, and they began tracking this one late last year.

No regulator anywhere has approved retatrutide. Lilly is still running Phase 3 trials and has said it plans to ask the FDA for approval in early 2027. You can buy it anyway: from websites, from med spas and, CBS found, from at least one Brooklyn convenience store.

What makes that possible is a line of fine print on the vial: Research Use Only. Not for Human Consumption. Those words let a seller import powder that no regulator has inspected, call it a laboratory reagent, and sell it to people who inject it into their abdomens. The label doesn’t describe the product. It describes who the seller says is responsible when something goes wrong, which is you.

Three days after the CBS report, one online vendor published a guide summarizing the two death reports. It said, accurately, that causation was unproven. Directly under that summary sat a box marked “Available to order” and a button to buy research-use retatrutide.

I’ve spent my career in drug development, most recently focused on a drug and pipeline in metabolic health. Getting a legitimate medicine to a patient takes years of toxicology, manufacturing validation, stability testing and human trials before a single prescription can be written. The gray market skips all of it. In my view, it keeps the commercial upside of a medicine and hands the biological risk to the person holding the syringe.

It’s fair to ask why so many people are willing to take that risk. Black markets grow in the gap between innovation and access. When the new obesity drugs carried list prices of about $1,060 to $1,349 a month in 2023 and 2024, sat behind prior-authorization walls and ran short for two years, a lot of patients went looking for an exit. That explains the demand. It doesn’t excuse the supply, and it doesn’t make what’s in the vial any safer.

A peptide is a drug class, not a spa menu

The gray market runs on a word that has lost its meaning. In wellness culture, “getting on peptides” sounds like starting magnesium or buying a cold plunge. It’s common enough that in a May 2026 Sermo survey of 507 U.S. physicians, 48 percent said a patient had told them about using a non-FDA-approved peptide in the past year.

In drug development, “peptide” isn’t a lifestyle category. It’s a modality: a class of molecules built from short chains of amino acids. The FDA draws the line at 40 amino acids or fewer. Above that, a chain is regulated as a protein. Peptides sit between small-molecule chemicals and large biologics like monoclonal antibodies.

Calling every peptide an anti-aging aid is like calling every small molecule a sleep aid. Melatonin is a small molecule. So are most chemotherapy drugs. (Nobody’s putting those on a spa menu.)

Here’s some of what the peptide class actually contains:

• Vasopressin, nine amino acids, regulates blood pressure and how much water the kidneys keep.
• Cyclosporine, an 11-amino-acid ring, suppresses the immune system so a transplanted heart or kidney isn’t rejected.
• Ziconotide, 25 amino acids, is a synthetic version of a cone snail’s venom, infused into the spinal fluid for severe pain.
• Semaglutide, 31 amino acids with a fatty-acid chain attached, is the active ingredient in Ozempic and Wegovy.

Insulin marks the edge. At 51 amino acids, it’s past the FDA’s line, and since 2020 it has been regulated as a biologic. A miscalculated dose can cause severe, sometimes fatal, low blood sugar.

These molecules are signals. They bind to specific receptors and switch on endocrine, cardiovascular, metabolic and neurological responses. When you inject one, you’re not applying a skin cream. You’re pulling a regulatory lever inside your own physiology. Rebranding that as wellness made it easy to treat experimental drugs like vitamins.

How the guardrails came off

People outside biotech ask me some version of the same question: if only approved drugs can be prescribed, how are licensed clinics putting patients on unapproved peptides? The guardrails didn’t fail at once. They came off in three stages.

Stage one: the shortage. Federal law lets compounding pharmacies make patient-specific versions of drugs, and lets both those pharmacies and larger outsourcing facilities make copies of an approved drug while it sits on the FDA’s shortage list. When demand for semaglutide and tirzepatide outran supply in 2022, both went on that list, and a compounding boom followed. Telehealth platforms and cash-pay med spas built prescribing engines around it. The American Med Spa Association estimated 8,899 med spas in 2022 and 10,488 in 2023, in an industry it puts at about $17 billion.

Then the FDA declared the tirzepatide shortage resolved in late 2024 and the semaglutide shortage resolved in February 2025. The legal window for mass copies closed after short wind-down periods. My read is that many of these businesses didn’t close; they changed the menu. By June 2026, CBS had found more than 50 U.S. clinics selling or promoting retatrutide. The FDA has since sent warning letters in waves: about 80 in September 2025, 30 in March 2026 and 25 in June 2026, most of them aimed at telehealth marketing of compounded GLP-1s.

Stage two: the vote. A pharmacy generally can’t compound from bulk powder unless the substance is part of an approved drug, has a USP monograph, or is on the FDA’s 503A bulks list. When the FDA asked for nominations to that list, the industry sent thousands, including, as ProPublica reported, purified water and asparagus. In 2023 the FDA put 19 popular peptides, BPC-157 among them, in Category 2: substances that may present significant safety risks.

In April 2026, after the original nominators withdrew their nominations, the FDA took 12 peptides off that list. In July, its Pharmacy Compounding Advisory Committee reviewed seven of them. FDA staff recommended against all seven, citing thin evidence. BPC-157 is a good example: a 2025 systematic review in the HSS Journal found 36 eligible studies, 35 of them in animals or cells and one in people. The committee voted to recommend six of the seven anyway, by close margins. BPC-157 passed 8 to 6.

The members who voted yes made a real argument. Keep these peptides off the list, they said, and patients will keep buying them from the black market with no physician involved. I take that seriously. A pharmacy-compounded product with a prescriber is safer than an anonymous vial. But it treats a list of ingredients as if it were evidence the drug works. Compounding access doesn’t tell anyone whether BPC-157 heals a tendon or at what dose. It just makes an unproven product look vetted.

It also didn’t do what many people were told it did. Pharmacists spent the following weeks warning customers that the vote hadn’t made these peptides legal to compound, for three reasons:

1. Advisory votes aren’t binding. Nothing becomes law until the FDA finishes notice-and-comment rulemaking, which has no deadline. As of October 17, 2026, none of these peptides is on the list. HHS has sent the White House an interim policy that could let pharmacies compound some of them sooner, but its terms aren’t public and it hasn’t been issued.
2. Even the final list carries a warning. Under 21 CFR 216.23(d), the FDA states there are “inadequate data to demonstrate the safety or efficacy” of drugs compounded from listed substances, and calling such a drug FDA-approved or FDA-endorsed makes it misbranded.
3. None of it applies to retatrutide.

Stage three: no path at all. Retatrutide has no approval, no monograph and no place on any bulks list. There is no legal way to compound it in the United States. So it arrives the other way: as powder bought from overseas manufacturers nobody has inspected, imported, relabeled “for research use only,” and sold beside the supplies you’d need to inject it. One of Lilly’s October lawsuits alleges that adding retatrutide to a seller’s online cart automatically added a reconstitution solution.

Licensed medicine is part of this, too. A CBS investigation in June found more than 120 websites and more than 50 U.S. clinics selling or promoting retatrutide, many staffed by licensed physicians and nurse practitioners. At least 21 clinics changed their sites after reporters called. The second death report, filed in September, describes a 49-year-old patient whose nurse sent her a link, sold her vials from an online peptide vendor and recommended doses. That account isn’t verified. If it’s accurate, my view is simple: a clinician who does that isn’t practicing medicine. They’re laundering contraband with a medical license.

You’d expect a pharmacist somewhere in this chain. Prescription drugs usually pass through one. Here’s how this one skips it. Most states let physicians buy and dispense drugs from their own offices, at least in some circumstances. A research-use vendor will sell to anyone without a prescription, because on paper it’s selling a chemical. Put those together and the whole supply chain fits in a shopping cart: the clinic orders the vials, and the clinic gives the shots. Lilly’s August lawsuits allege that kind of “med-spa administration,” and accuse one pharmacy of selling retatrutide to med spas and providers. The simplest version skips the vial too. A clinician sends the patient a link, as the nurse in that death report did. As far as I can tell, no pharmacist ever sees the order.

What’s actually in the vial

When nobody is accountable for what’s in the glass, nobody can promise what’s in it. Lilly says it has found products sold as retatrutide with too much, too little or the wrong active ingredient, plus contaminants. Independent testing points the same way.

• In September, WIRED bought four vials of injectable peptides at a Brooklyn bodega and sent them to Finnrick, an independent testing company in Austin. None contained a peptide.
• In Australia, a patient who had injected a product sold as retatrutide was hospitalized with a torn esophagus after uncontrollable vomiting. The Therapeutic Goods Administration tested the vial. It contained no retatrutide, and about eight times the semaglutide found in approved products.
• In June, Victoria’s chief health officer warned of six cases of acute liver injury linked to products labeled retatrutide. A cause hasn’t been confirmed.

To be fair to the vendors, not every vial fails. Among the samples people have chosen to send it, Finnrick has reported that nearly 30 percent were mislabeled, under- or overdosed, or contaminated. That isn’t a random sample, but it suggests many vials pass. The problem is that you can’t know which group your vial is in, and nobody is on the hook if it’s the wrong one.

Even a real attempt at retatrutide is hard to make well. It’s a 39-amino-acid chain of about 4,731 daltons, with non-natural amino acids and a fatty-acid chain attached. Aspirin, by comparison, is 21 atoms and 180 daltons. The simplest way to build a peptide like this is one amino acid at a time on a solid resin, which means 38 coupling cycles to reach full length, and every cycle that falls short leaves behind a slightly wrong chain.

Here’s an illustration of why that matters. It’s not a measurement of any product.

0.98³⁸ ≈ 0.46

If each step, counting every way it can go wrong, worked 98 percent of the time and nobody purified the result, a little under half the chains would come out at full length. The rest would be truncated or missing an amino acid somewhere. Good chemistry usually does better than 99.5 percent on any single coupling, but in crude batches of longer peptides the by-products often outweigh the target anyway. Legitimate manufacturers don’t ship crude material. They purify it, test each lot against limits and prove the process is reproducible.

That process control is what you lose in the gray market:

• Wrong chains. Peptide impurities that closely resemble the drug are a recognized immunogenicity risk, which is why the FDA asks generic peptide makers to characterize them closely. Antibodies against a drug could, in principle, cross-react with the body’s own hormones.
• Leftover chemistry. Peptides are cut off the resin with trifluoroacetic acid. That leaves trifluoroacetate bound to the peptide, and pharmaceutical makers typically exchange it for acetate or chloride. Left in place, it can make up a meaningful share of the powder’s weight. That throws off the real dose, and in cell studies trifluoroacetate has slowed the growth of bone and cartilage cells. A research-use vial may not say which salt you’re getting.
• Endotoxin. Some people filter reconstituted peptides through a 0.22-micron syringe filter and assume they’re sterile. The filter stops live bacteria. It doesn’t remove endotoxin, the cell-wall fragments of dead bacteria, which can cause fever and chills.

Gray-market sellers answer all of this with a PDF certificate of analysis showing 99 percent purity. A purity trace alone doesn’t confirm the molecule’s identity, and it doesn’t check for endotoxin or sterility. Any test speaks only for the sample someone tested, not the vial in your hand. Certificates can also be forged. In the Paradigm Peptides case below, prosecutors said the company forged its certificates. You’re not biohacking. You’re injecting chemistry nobody has vouched for.

Who answers when it goes wrong

The sharpest difference between a drug developer and a gray-market seller isn’t the chemistry. It’s who answers when a patient gets hurt.

A company with a molecule in clinical trials can’t sell a single dose to the public, or promote it as working, until the FDA approves it. Every batch it gives a patient is made under current good manufacturing practice, under an investigational new drug application the FDA has cleared, in trials an institutional review board oversees. Serious, unexpected adverse events go to the FDA on a deadline, often within 15 days. When something goes wrong, named people answer to regulators, investors and the courts.

As far as I can tell, a research-use seller needs little more than an LLC, a storefront and a supplier. Its terms of service ask buyers to confirm they’re researchers. If a customer ends up in an emergency room, the seller points to the label.

Warning letters haven’t closed that gap. Public Citizen found that the FDA sent 14 warning letters to companies selling retatrutide starting in December 2024. As of May 2026, 11 of the 14 were still advertising retatrutide or other unapproved peptides, and eight were still selling retatrutide. An FDA warning letter is a piece of paper, not a padlock. Seizing inventory or shutting a business down takes a referral to the Justice Department and time in court.

That route does work eventually. In July, Matthew Kawa, owner of the online vendor Paradigm Peptides, was sentenced to 70 months in federal prison after pleading guilty to introducing unapproved drugs into interstate commerce with intent to defraud or mislead. He agreed to forfeit $5 million. His company had sold to about 54,000 customers under research-use labels while, he admitted, knowing the products were being sold for human use. Prosecutors also pointed to FDA warning letters he received in 2020 and 2022 and ignored. The research-use label wasn’t a shield. It took years to prove that.

Why a drug company is doing the policing

In my view, federal enforcement was built to inspect factories, not to chase thousands of storefronts. It’s catching up, and the numbers show how big the problem has become. In all of fiscal 2025, Customs and Border Protection intercepted more than 690 shipments containing over 31,000 units of illicit GLP-1 drugs. In July 2026 alone, it recorded more than 1,400 seizures and nearly 90,000 vials. Those figures cover illicit GLP-1 products broadly, not just retatrutide.

The most visible enforcer so far is the company whose molecule is being copied. On August 12, Lilly announced six lawsuits against U.S. businesses it accuses of selling retatrutide, including med spas, online sellers and a pharmacy. It said it had referred more than 200 people and businesses to regulators, law enforcement and licensing boards, and had reported more than 14,000 websites, ads, social posts and listings marketing retatrutide in over 100 countries. On October 8, it filed five more suits, against three online retailers, a weight-loss clinic and a med spa. These are allegations, though Lilly told The Washington Post it has settled two of the August cases, one with a seller that agreed to stop selling and tell customers to stop using its products.

Lilly is also pressing payment companies, platforms and shippers to stop doing business with these sellers. My bet is that’s the part that moves fastest. A seller that can’t take a card payment or ship a box doesn’t need to be sued.

The skeptic’s reading is obvious: a drug company is protecting a future blockbuster. That’s partly true. Lilly has a commercial stake, and it has said openly that it worries bad reactions to bootleg products will be blamed on the real one. Both things can be true at once. A sponsor protecting its molecule and a sponsor protecting patients from fake versions of it are, in this case, doing the same work. My harder question is why so much of the work is falling to a private company rather than the agencies we fund to do it.

Real speed vs. counterfeit shortcuts

The gray market’s best defense is that the FDA is too slow. On the first half of that, I agree. Review times should be shorter, and the country already has legitimate ways to make them shorter.

Accelerated approval lets a drug for a serious condition reach patients years earlier on a surrogate endpoint, if the sponsor runs confirmatory trials. Emergency use authorization put COVID-19 vaccines and treatments into use within months. And the FDA’s Commissioner’s National Priority Voucher pilot, launched in 2025, aims to bring final review down to one or two months with a team-based review. In April 2026 the FDA approved orforglipron (Foundayo), Lilly’s daily GLP-1 pill for obesity, through that program. In November 2025, the FDA’s Vinay Prasad and Commissioner Martin Makary outlined a “plausible mechanism” pathway in the New England Journal of Medicine, aimed first at bespoke therapies for rare diseases. In February 2026, they said one adequate, well-controlled trial plus confirmatory evidence would become the FDA’s default for novel drugs.

None of those pathways remove accountability. A sponsor’s name is still on the application. The plant is still inspected. Every batch is still tested for identity, purity, sterility and endotoxin before release. Adverse events are still reported. If something goes wrong, everyone knows who answers for it.

The gray market doesn’t cut red tape. It cuts out the science.

What would help

I think three things would do more than another round of warning letters.

1. Go after the infrastructure. Regulators can’t take down 14,000 listings one letter at a time. Payment processors, e-commerce hosts, social platforms and shippers make money from these transactions, and they’re far easier to find than the sellers. The FTC, the FDA and state attorneys general should make that business more expensive than it’s worth.
2. Hold licensed professionals to their licenses. State medical, nursing and pharmacy boards should treat a clinician who sources research-use chemicals for patients, or takes affiliate commissions from an unapproved peptide vendor, as a serious discipline case, up to and including the license. Some boards have started. In May, Alabama’s medical board told physicians that recommending, supplying, prescribing or administering a research-grade peptide is off-limits, and that every prescription drug must come from a source its pharmacy board permits. South Carolina’s board has issued a similar notice.
3. Speed up inside the guardrails, never outside them. Physicians, industry and patient groups should back faster review pathways and modern trial designs. And patients should hold one personal line: if a clinic offers you a molecule that’s still in clinical trials, or hands you a vial marked “Not for Human Consumption,” walk out.

When you skip regulated drug development to save time or money, you don’t get rid of the cost of proving a medicine is safe. You pay for the clinical trial with your own body.

Disclosure: I invest in biotechnology, pharmaceutical and technology companies and may hold positions, directly or through index funds, in companies mentioned in this piece, including Eli Lilly. This is my personal view, not investment advice, and it does not speak for any company I lead.

Sources

• CBS News: Two deaths were reported to the FDA (Oct 13, 2026)
• Drug Topics on CBS’s June investigation: 120+ websites, 50+ clinics
• Hoodline on CBS’s Brooklyn convenience store footage
• Becker’s: Lilly’s filing timeline, settlements
• Sermo Barometer: 507 physicians, 48%
• HSS Journal: BPC-157 systematic review
• AmSpa 2024 State of the Industry recap
• National Law Review: FDA telehealth warning letters, 2025 to 2026
• ProPublica via Daily Kos: nominations, asparagus
• Goodwin: April 2026 Category 2 removals
• McDermott: PCAC votes, July 23 to 24, 2026
• Becker’s: harm-reduction argument at PCAC
• Partnership for Safe Medicines: pharmacists on the vote
• 21 CFR 216.23 (eCFR)
• Buchanan: Lilly’s October 8 lawsuits
• WIRED via DNYUZ: bodega peptides
• TGA safety advisory
• Victoria liver injury alert, as reported (swap for the Victorian health department alert)
• Finnrick testing data (NY Mag)
• Public Citizen: retatrutide report, June 16, 2026
• McDermott: Kawa sentencing
• DOJ: United States v. Matthew Kawa
• Yahoo Finance / PharmaVoice: CBP seizures
• Healio: Lilly’s six lawsuits and 14,000+ listings
• AJMC: orforglipron approval
• AP via ABC7: orforglipron through the voucher program
• KFF: GLP-1 list prices ($936 to $1,349)
• KFF: weight-loss drug prices vs. peer nations (2023)
• Quartz: Zepbound launch price
• Kent 2025, Journal of Peptide Science: stepwise yields and crude by-products in SPPS
• Cornish et al. 1999: trifluoroacetate and osteoblasts (PubMed)
• Sikora et al. 2018: counter-ion effects (PMC)
• Roux et al. 2008: trifluoroacetate exchange (PubMed)
• BioPharma Dive: plausible mechanism pathway, NEJM (Nov 12, 2025)
• Troutman: single pivotal trial default and draft guidance (2026)
• Wouters et al., JAMA 2020: R&D cost per new drug (LSHTM summary)
• NABP: 503A compounding outside federal adverse-event reporting (July 2026)
• Frier Levitt: Category 2 designations in 2023
• Foley: tirzepatide shortage resolved (Oct and Dec 2024)
• McDermott: semaglutide shortage resolved (Feb 21, 2025)
• Frier Levitt: Lilly’s retatrutide lawsuits (August 2026)
• Drug Topics: physician in-office dispensing (2014)
• Alabama Board of Medical Examiners: notice on research-grade peptides
• 1819 News: Alabama board’s peptide notice (May 27, 2026)
• South Carolina Board of Medical Examiners: notice re peptides

Cover: illustration by Torpedo Publishing, in PD brand colors.