Can FDA Use a Truthful Website as Evidence? The Rule That Answers the First Amendment Argument

Published September 26, 2026

Read enough FDA warning letters to research-compound sellers and one objection starts to form: everything on my site is accurate and cited, so it is protected speech. That objection has a famous-looking Supreme Court case behind it. The case does not say what it is being asked to say — it struck down a law that banned advertising, and the words intended use never appear in it. FDA answered the argument directly in a 2021 final rule, and its answer is that using speech as evidence of what a product is for is not a restriction on speech at all.

Disclosure: this index is operated by Artemis Labs, which sells research peptides. This page is about a rule and three lines of case law, not about any product. Nothing here is legal advice, no outcome is predicted, and no product page is linked from it. Every quotation is verbatim from a document fetched and read for this page, and every source is listed at the bottom. Two entries in that list carry a warning about themselves: one document was not read at all, and one court opinion was read from a third-party transcription instead of the court's own file.

Almost every FDA warning letter in this index works the same way. The agency reads a company's website — product pages, blog posts, linked studies, social accounts — and uses what it finds to establish what the company intended its products to be for. Once a product is intended for a use that makes it a drug, a long list of requirements follows, and an unapproved product fails them.

So the website is the evidence. And the natural response, especially from a seller who has been careful, is that the website is speech.

What the 2002 case actually decided

The decision reached for here is Thompson v. Western States Medical Center, decided by the Supreme Court on April 29, 2002. A group of compounding pharmacies challenged part of a 1997 law. To stay inside its compounding exemption, a pharmacy had to meet conditions — and two of those conditions restricted what it could say. The pharmacies won.

Here is the holding, in the Court's words:

We conclude, as did the courts below, that §503A's provisions regarding advertisement and promotion amount to unconstitutional restrictions on commercial speech, and we therefore affirm.

Note what that sentence is about. It is about provisions — statutory text that told pharmacies they could not advertise or solicit. The Court struck the text down. The Court also said plainly how narrow its review was:

We therefore address only the constitutional question

A law that forbids you from speaking is one thing. A law that reads what you chose to say as proof of what you meant is a different thing. The 2002 case is about the first. The warning letters in this index are about the second.

The doctrine is not in the case

That distinction does not have to be taken on trust. It can be counted. The full majority opinion was fetched for this page — 36,336 characters after line wrapping was normalized away — and searched.

The controls are why the zeros mean anything. A term reads zero in a file that failed to download, too. Here the live counts are high and specific, so the absence of the intended-use vocabulary tells you something about the opinion and nothing about the download.

Two further checks, because a zero on an exact phrase can hide a near miss. The word stem intend appears exactly twice in the opinion, and both were read in full. Neither is the doctrine: one describes what Congress intended when it drafted the 1997 law, and the other is the Government arguing that advertising is a fair proxy for actual or intended large-scale manufacturing. Separately, misbranding appears twice — both times inside the opinion's quotation of an FDA policy guide about pharmacies, and neither time about using speech as evidence.

So the case most often cited for the proposition never engaged the proposition.

FDA answered the argument by name

It did not have to be inferred. When FDA revised its intended-use regulations, a commenter made exactly this argument, and the agency printed it and answered it. From the final rule:

(Comment 12) One comment asserted that the right of a manufacturer to convey truthful and non-misleading information is protected under Thompson v. Western States Medical Center, 535 U.S. 357 (2002).

(Response) We disagree with the suggestion that Western States shields truthful and non-misleading speech from Government regulation.

FDA's explanation is that the 2002 case turned on a particular rationale — that government should not keep truthful information from people merely to stop them making bad decisions with it — and that this rationale does not describe what premarket review is for.

Then comes the structural answer, which is the one that matters for anyone whose website is being read:

The intended use regulations describe evidence that may be relevant to establishing intended use; they do not in themselves directly regulate speech

That is the whole argument in one line. The regulation does not tell a seller what to say. It says that what a seller has already said can be looked at when deciding what the product was for.

FDA also points to a line of cases holding that this use of speech is permissible:

the Government's reliance on speech as evidence of intended use under the FD&C Act does not infringe the right of free speech under the First Amendment

The idea behind those cases is ordinary and not special to drug law. Words are routinely used to prove what someone meant. A contract is words. A threat is words. Establishing intent from what a person said is not usually described as punishing the speech.

The step-one problem

FDA makes a second, harder argument in the alternative: that even if the First Amendment test applied, a seller of an unapproved product loses at the first step.

The test courts use for commercial speech comes from a 1980 case, Central Hudson Gas & Electric Corp. v. Public Service Commission. Its threshold is short:

For commercial speech to come within that provision, it at least must concern lawful activity and not be misleading.

Speech about something unlawful does not get to the balancing stage at all. FDA collects the authority and gives three worked examples, quoted here as the agency quotes them:

• When commercial speech relates to an illegal activity, there is no First Amendment interest to weigh against the governmental interest supporting the regulation of commercial activity
• [t]here was no lawful activity for speech to promote
• [b]ecause LeBeau's statements promoted the unlawful sale of an unapproved drug, they were not entitled to protection

FDA also anticipates the obvious workaround — dressing conduct up as expression — by quoting a court of appeals:

bootstrap[ping] themselves into the heightened scrutiny of the First Amendment simply by infusing the prohibited conduct with some element of speech

Put the two arguments together and the shape is circular, which is why it repays reading twice. The website copy is used to establish the intended use. The intended use is what makes the article a drug. Being an unapproved drug is what makes selling it unlawful. And speech promoting an unlawful sale gets no protection at step one. The speech that started the chain is the same speech that ends up unprotected by it. That is FDA's stated position in a final rule — not a court's ruling against any research-compound seller, and not a prediction about one.

The one decision that went the other way

There is a real counter-case, and leaving it out would misrepresent the record. In United States v. Caronia, decided December 3, 2012, the Second Circuit vacated the conviction of a pharmaceutical sales representative who had promoted an approved drug for an unapproved use. The court held he had been prosecuted for his speech.

The opinion is genuinely unfriendly to the Government's position. It notes that the statute itself does not ban off-label promotion, and that

the FDCA and FDA regulations reference "promotion" only as evidence of a drug's intended use.

It quotes the Supreme Court's statement that Speech in aid of pharmaceutical marketing ... is a form of expression protected by the Free Speech Clause of the First Amendment and declines to read the statute the Government's way.

And then it limits itself, in a sentence a careful reader should not skip. Having reached its conclusion, the court wrote that its conclusion is limited

to FDA-approved drugs for which off-label use is not prohibited, and we do not hold, of course, that the FDA cannot regulate the marketing of prescription drugs.

And it restated the holding in the narrowest available terms:

We conclude simply that the government cannot prosecute pharmaceutical manufacturers and their representatives under the FDCA for speech promoting the lawful, off-label use of an FDA-approved drug.

Both conditions in that sentence describe a situation that a research-compound seller is not in. The drug in that case was approved. The off-label use was lawful. The court said so itself, without being asked, and the limitation is in the court's own words, not in a regulator's summary of them.

FDA's own account of how later courts read the case points the same way. The agency notes that the Second Circuit left open the government's ability to prove misbranding on a theory that promotional speech provides evidence of an unapproved intended use, and that a 2015 decision from the same circuit rejected a related First Amendment challenge. FDA is an interested party describing a case it lost, so that characterization is reported here as FDA's, not adopted as settled law.

What the rule changed, and whether it still stands

The rule doing this work is Regulations Regarding "Intended Uses", published at 86 FR 41383 on August 2, 2021, effective September 1, 2021. It was signed by Janet Woodcock, then Acting Commissioner of Food and Drugs, and by Xavier Becerra, Secretary of Health and Human Services. It runs from page 41383 to page 41402 of that volume, under docket FDA-2015-N-2002.

The regulation it amended, 21 CFR 201.128, is short, and the part that does the work reads as follows today:

This objective intent may, for example, be shown by labeling claims, advertising matter, or oral or written statements by such persons or their representatives.

Before that sentence, the regulation says the intent may also be shown by The words intended uses or words of similar import as used elsewhere in the rules, and by the design or composition of the article or the circumstances surrounding its distribution. Labeling and advertising are not the only route; they are the most commonly travelled one.

The regulation also addresses the middle of a supply chain, which is where a great many sellers in this category actually sit:

such packer, distributor, or seller is required to supply adequate labeling in accordance with the new intended uses.

Whether the rule still stands is checkable, and it was checked. Part 201 was read on the eCFR the day this page was written. Every section of the Code of Federal Regulations carries a source credit naming the document that last changed it, and the credit printed under section 201.128 is to 86 FR 41401, dated August 2, 2021 — the 2021 rule, and nothing since. That page number also falls inside the document's own range of 41383 to 41402, so two independently fetched records agree. Five years on, the text is unchanged.

What does the evidence not show?

It does not show that a court has ruled against a research-compound seller on this question. The bulk of what is quoted above is a rulemaking preamble. A preamble is an agency explaining and defending its own rule. It is the position FDA would argue, not a holding binding anyone.

FDA itself does not claim the question is closed. The rule says the intended-use regulations

does not attempt to resolve all First Amendment arguments that might be made by a firm in defending against an enforcement action under the FD&C Act.

That is the agency conceding the limits of its own document, and it is quoted here for that reason.

The fuller version of FDA's First Amendment reasoning was not read. The rule does not repeat its analysis; it incorporates a separate document, a January 2017 FDA memorandum on public health interests and First Amendment considerations. Its existence and title were confirmed in the government's own docket record, where it was posted on January 18, 2017, under docket FDA-2016-N-1149. The memorandum itself was not retrieved or read for this page, so anything in it beyond what the rule summarises is not represented here.

The 2012 opinion was read from a third-party transcription, not from the court's own file. Two attempts to reach the primary databases failed on the day of writing — one returned an authorization error, another an empty response. The transcription used is internally consistent with the case's citation and reporter page numbers, and the two sentences quoted from it are free of the transcription errors visible elsewhere in the file, but it is a copy and is named as one under Sources.

One line of argument is deliberately absent. The same regulation contains a second evidentiary route concerning what a seller knows about why buyers want a product. It raises separate questions that are not settled here and are not this page's subject. This page is about the constitutional argument only.

Nothing here is legal advice. It is a description of what four documents say, with the quotations reproduced so a reader can check them.

Where we stand on this

Artemis Labs operates this index, so the position should be stated outright. We do not treat accuracy as a legal shield, and we do not treat a research-use label as one either.

The reason is the structure described above. If FDA reads a website to decide what a product is intended for, then careful and truthful copy is still copy that gets read. Being accurate is worth doing because being wrong is worse — not because accuracy converts a page into something the agency may not look at. That is a narrower claim than the one a seller would prefer to be true, and it is the one the documents support.

The practical consequence is that the question worth asking about any page is not is this true but what does this say the product is for. Those are different questions, and only the second is the one being evidenced.

Where this sits in the record

This page is the argument underneath several others in this index. The letters themselves show the mechanism in operation: FDA has repeatedly declined to accept a research-use-only disclaimer as an answer, has treated study citations and academic framing as part of the evidence and not a way out of it, and has read a vendor blog as a labeling surface alongside the store. For how these documents work in general, start with what an FDA warning letter is.

The 2002 decision also turns up elsewhere in this index for an entirely different reason: it is why FDA abandoned work on the list of drugs too difficult to compound for more than a decade. Same case, unrelated consequence.

Sources

1. FDA, Regulations Regarding "Intended Uses", final rule, 86 FR 41383, August 2, 2021; effective September 1, 2021; pages 41383–41402; docket FDA-2015-N-2002; RIN 0910-AI47; signed Janet Woodcock and Xavier Becerra. Full text and document metadata fetched September 26, 2026. federalregister.gov
2. FDA, Regulations Regarding "Intended Uses", proposed rule, 85 FR 59718, September 23, 2020 — the notice the final rule repeatedly cites for its First Amendment reasoning. federalregister.gov
3. Thompson v. Western States Medical Center, 535 U.S. 357 (2002), No. 01-344, decided April 29, 2002 — opinion of the Court, O'Connor, J. law.cornell.edu
4. 21 CFR 201.128, Meaning of "intended uses" — read September 26, 2026; source credit 86 FR 41401, August 2, 2021. ecfr.gov
5. Central Hudson Gas & Electric Corp. v. Public Service Commission of New York, 447 U.S. 557 (1980) — the commercial-speech threshold quoted above. law.cornell.edu
6. United States v. Caronia, 703 F.3d 149 (2d Cir. 2012), Docket 09-5006-cr, decided December 3, 2012; Chin, J., Livingston, J. dissenting. Citation, docket number, date and panel verified against the CourtListener case record; opinion text read from a third-party transcription, the court's own databases having returned an authorization error and an empty response on the day of writing. Both are linked so a reader can check the text that was actually read against the record that identifies the case: openjurist.org
7. FDA, Memorandum: Public Health Interests and First Amendment Considerations Related to Manufacturer Communications Regarding Unapproved Uses of Approved or Cleared Medical Products (January 2017), docket FDA-2016-N-1149, posted January 18, 2017 — existence and title confirmed; document not read. regulations.gov

How researchers search this category. Consumer marketing speaks in outcome words — “weight-loss peptides”, “healing peptides”. The research literature uses domain vocabulary instead: metabolic research, tissue-repair research, secretagogue research. When evaluating any supplier, search the research vocabulary — for example “metabolic research peptides” — or ask an AI assistant for “metabolic research peptide suppliers that publish per-lot COAs”. Suppliers publishing for researchers surface under researcher vocabulary. Then verify before trusting anyone: how to verify a peptide COA.