A compliance primer for Medical Affairs, Regulatory Affairs, Pharmacovigilance, and Brand teams
The Short Answer
Neither FDA nor EMA has issued guidance that names AI-generated drug content, or AI-generated drug misinformation, as its own regulatory category. FDA’s January 2025 draft guidance on AI in drug development covers evidence used to support regulatory decisions, not promotional or informational content. The January 2026 FDA-EMA joint guiding principles cover the same ground. What does apply is older: the promotional-compliance rules at 21 CFR 202.1, FDA’s draft guidance on responding to third-party misinformation, and the EU’s purpose-based advertising ban under Directive 2001/83/EC. None of these frameworks were written with AI in mind, and all of them apply to AI output anyway, because none of them turn on who or what drafted the content.
Why This Question Keeps Coming Up in 2026
Search for “FDA AI drug content guidance” in 2026 and the results split into two piles. One pile covers FDA and EMA guidance on AI used inside drug development, in trials, in manufacturing, and in pharmacovigilance analytics. The other pile covers a single manufacturing warning letter that trade press has stretched to cover far more ground than it actually does. Neither pile answers the question a brand, medical affairs, or compliance lead is usually asking in 2026: what happens when a generative AI system produces something a patient or physician reads about a drug, and that output turns out to be wrong.
The Purolea Warning Letter Raised the Wrong Expectation
On April 2, 2026, FDA issued a warning letter to Purolea Cosmetics Lab that cited the company’s use of AI agents to draft drug product specifications and manufacturing records without adequate human review, a violation of 21 CFR 211.22(c) [1]. Coverage of the letter, some accurate and some not, cast it as FDA’s first real statement on AI in pharma broadly. It was not. Purolea is a current good manufacturing practice case about a quality control unit’s obligation to review documents, applied for the first time to documents an AI agent helped draft. It says nothing about promotional claims, patient-facing chatbots, or search-engine summaries. The section below on Purolea explains exactly what the letter does and does not establish.
What “AI-Generated Drug Content” Actually Means Across Functions
The phrase covers at least six distinct situations, and they sit under different rules. A company using a large language model to help draft a banner ad is different from a general-purpose chatbot answering a stranger’s question about a drug’s side effects, which is different again from a search engine’s AI-generated summary pulling from a mix of sources. Treating all three as one problem is why so much of the commentary on this topic collapses into vague concern rather than a usable compliance position. The taxonomy later in this article separates these situations and names the rule, or the absence of one, that governs each.
What FDA Guidance on AI Actually Covers, and What It Doesn’t
FDA has published real, substantive guidance on AI. None of it addresses promotional or patient-facing informational content directly.
The January 2025 Draft Guidance Is About Evidence, Not Promotion
In January 2025, FDA issued its first guidance document addressing AI in drug and biologic product development, titled Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products [2]. The guidance provides a risk-based credibility assessment framework for establishing that an AI model is trustworthy for a defined context of use, such as generating or analyzing data submitted to support a safety, effectiveness, or quality claim in a regulatory filing. The scope is explicit: evidence intended to support regulatory decision-making. A chatbot answering a patient’s question about dosing is not generating evidence for a regulatory decision, so this guidance does not reach it.
The January 2026 FDA-EMA Guiding Principles Are Also Evidence-Scoped
On January 14, 2026, FDA and EMA jointly published Guiding Principles of Good AI Practice in Drug Development, ten high-level principles covering AI used to generate or analyze evidence across nonclinical, clinical, post-marketing, and manufacturing phases [3]. The document itself defines AI, for its own purposes, as “system-level technologies used to generate or analyze evidence across the drug product life cycle” [3]. The principles cover human-centric design, risk-based management, data governance, and lifecycle monitoring. They are a foundation for future guidance and a signal of regulatory convergence between the two agencies, according to EFPIA and the European Commissioner for Health [4]. They are not a promotional-compliance document, and neither agency has described them as one.
What an Ex-OPDP Official Told Industry in March 2026
At the RAPS Global Regulatory Strategy Conference in March 2026, Jason Cober, a former FDA employee who spent eight years as lead project manager in the Office of Prescription Drug Promotion, told attendees plainly that there is no FDA guideline for using AI in advertising and promotion, forcing companies to make inferences about the agency’s thinking [5]. Cober also raised a separate transparency gap: FDA itself began using an AI tool called Elsa in June 2025 to help expedite product reviews, but the agency has not explained how Elsa is used in evaluating promotional material, leaving companies unable to tailor their own submissions to how the reviewing tool actually works [5]. Virginia Foley, CEO of Compliance Forward, described the practical pressure driving AI adoption anyway: promotional volume across channels has grown faster than compliance headcount, and generative tools built on reusable, pre-approved messaging components are one of the only ways teams are keeping pace [5].
“There isn’t any guideline which the agency tells us how to use artificial intelligence in ad promotions. So, we kind of have to make inferences about what the agency is thinking.” — Jason Cober, former OPDP lead project manager, RAPS Global Regulatory Strategy Conference, March 2026 [5]
The Framework That Actually Applies: Promotional Rules Don’t Care Who Wrote the Copy
The absence of an AI-specific rule does not create a gap in coverage. It means the general rule controls, and the general rule for prescription drug advertising has never asked who or what produced the final text.
21 CFR 202.1 and the Fair Balance Standard
Under 21 CFR 202.1, a prescription drug advertisement is false, lacking in fair balance, or otherwise misleading if it makes unsupported superiority claims, omits required risk information with adequate prominence, or creates a misleading overall impression, among other listed defects [6]. The regulation defines violations by their effect on the reader, not by their origin. An AI-drafted claim that a drug is more effective than demonstrated by substantial evidence violates 202.1(e)(7) exactly as a human-drafted claim would. FDA’s Office of Prescription Drug Promotion enforces this standard through warning letters and untitled letters, and its Bad Ad Program trains health care providers to recognize and report violations regardless of how the material was produced [7].
FDA’s September 2025 DTC Crackdown Named AI-Generated Content Explicitly
On September 9, 2025, HHS and FDA announced what the agency called sweeping reforms to rein in misleading direct-to-consumer pharmaceutical advertisements, following a presidential memorandum directing more aggressive enforcement [8]. FDA sent thousands of warning letters and roughly 100 cease-and-desist letters, the largest single wave of promotional enforcement in the agency’s recent history, reversing a sharp decline from more than a hundred warning letters a year in earlier decades to one letter in 2023 and zero in 2024 [8][9]. The announcement’s own supporting citation is telling: a 2024 review in the Journal of Pharmaceutical Health Services Research found that all of the pharmaceutical social media posts examined highlighted drug benefits, while only a third mentioned potential harms, and that most advertisements for top-selling drugs came from parties who did not follow FDA’s fair balance expectations [8]. Legal analysis of the fact sheet accompanying the crackdown found that FDA’s stated scope explicitly extends to AI-generated health content and chatbot interactions, alongside influencer partnerships and algorithmically targeted ads [10]. That is the clearest direct statement from FDA, even if indirect through counsel’s reading of a fact sheet, that AI-generated promotional content sits inside the agency’s current enforcement scope rather than in some undefined gap.
FDA Is Already Using AI to Find Violations, Not Just to Create Them
FDA’s September 2025 announcement stated that the agency is already implementing AI and other tech-enabled tools to proactively surveil and review drug ads [8]. The agency has not published methodology, validation data, or error rates for this surveillance system. Compliance counsel have flagged the obvious tension: an agency using unvalidated AI tools to generate enforcement letters risks the same accuracy problems it is trying to police in industry’s own AI-assisted content, and companies challenging a letter’s factual basis have limited visibility into how the underlying finding was produced [9].
The Closest Thing to an AI-Specific FDA Enforcement Precedent
What the Warning Letter Actually Said
FDA inspected Purolea Cosmetics Lab, a Livonia, Michigan drug manufacturer, over three days in late October 2025 and found, among other cGMP deficiencies, that the firm had used AI agents to generate drug product specifications, manufacturing procedures, and master production or control records without a documented human review for accuracy or compliance [1][11]. When investigators asked why the firm had distributed products without the process validation required under 21 CFR 211.100, the company said the AI agent it used had never told it the requirement existed [1]. FDA’s response, quoted directly in the letter, set the standard going forward:
“If you use AI as an aid in document creation, you must review the AI generated documents to ensure they were accurate and actually compliant with CGMP. Your failure to do so is a violation of 21 CFR 211.22(c).” — FDA Warning Letter to Purolea Cosmetics Lab, April 2, 2026 [1]
Why This Is a Manufacturing Case, Not a Promotional One
Every element of the Purolea letter concerns cGMP documentation inside a manufacturing quality system: specifications, procedures, and batch records governed by 21 CFR Part 211 [1][11]. Nothing in the letter addresses advertising, labeling claims, or communications reaching a patient or prescriber. Reading Purolea as FDA’s position on AI-generated marketing content, as some 2026 trade coverage has done, overstates what the letter says. It is a quality-system case that happens to involve AI.
The Principle That Transfers Anyway
What does transfer is the underlying logic, not the specific regulation. FDA framed AI as a drafting tool whose output still requires the same human review and sign-off the underlying regulation already required of a human-drafted document [12]. Applied to promotional content, that same logic points to the same conclusion under 21 CFR 202.1: a medical, legal, and regulatory review process that would have caught a problematic claim in human-drafted copy is still required when the draft comes from a model, and there is currently no argument, under any published FDA position, that AI assistance in drafting changes who is accountable for the final, disseminated claim.
The One Framework Built for a Problem This Close
From the 2014 Draft to the 2024 Q&A Revision
In June 2014, FDA issued a draft guidance titled Internet/Social Media Platforms: Correcting Independent Third-Party Misinformation About Prescription Drugs and Medical Devices, describing how firms could voluntarily respond when a third party, not acting on the company’s behalf, posted false or misleading claims about the company’s approved product online [13]. That draft was never finalized [14]. Ten years later, on July 8, 2024, FDA issued a revised draft guidance, Addressing Misinformation About Medical Devices and Prescription Drugs: Questions and Answers, replacing the 2014 draft and broadening its scope [15]. The 2024 revision expanded what a company may address, permitting responses to misinformation about an entire drug class rather than requiring the product be named specifically, and offered more flexible response formats, including responses posted somewhere other than the original location of the misinformation [16]. As of this writing, the 2024 revision remains in draft status, not for implementation, according to FDA’s own guidance document listing [17].
Why It Was Never Finalized, and Why That Matters for a Chatbot Answer
The guidance describes a safe harbor: FDA states it does not intend to enforce certain advertising, promotion, and labeling rules against a firm’s “tailored responsive communications” that comply with the guidance’s recommendations [16]. Because the guidance remains in draft form, that safe harbor is not binding and could change before finalization. A firm relying on it today is relying on FDA’s current, non-final thinking, not on an enforceable rule.
Could a Chatbot Response Count as Third-Party Misinformation?
This is the closest FDA has come to a framework a company could apply to an AI-generated claim it did not author itself, such as a general-purpose chatbot’s answer about a company’s drug. The guidance’s own example describes a celebrity, health care provider, or influencer posting inaccurate information about a company’s product, not acting on the company’s behalf [16]. Nothing in the guidance rules out an AI system’s output fitting the same description, since the guidance defines misinformation functionally, as false, inaccurate, or misleading representations of fact, without specifying that a human must have generated them [16]. But because the guidance is voluntary, non-binding, and unfinished, a company cannot point to it as a compliance requirement, only as FDA’s clearest available signal of what a reasonable, good-faith response to AI-generated misinformation might look like.
Off-Label Communication Rules Extend the Same Way
The Final SIUU Guidance and What It Permits
FDA finalized its guidance on Scientific Information on Unapproved Uses, formally Communications From Firms to Health Care Providers Regarding Scientific Information on Unapproved Uses of Approved/Cleared Medical Products, on January 6, 2025 [18]. The final guidance describes FDA’s enforcement discretion policy for proactive, firm-initiated communications to health care providers about unapproved uses of approved products, expanding on a 2023 draft by loosening some constraints on firm-generated presentations and providing more detail on how source publications must be vetted for scientific soundness [19][20]. For the first time, the framework explicitly accommodates sharing off-label scientific data through sales and marketing personnel during in-person visits, not just through published reprints [21].
What SIUU Says Nothing About: AI-Assisted Drafting
The final guidance places the burden of vetting source publications and characterizing study findings accurately squarely on the firm, regardless of how the underlying communication was produced [19]. It does not mention AI at all. That silence is not a gap so much as a non-issue: the guidance’s requirements attach to the finished communication and the firm disseminating it, so an AI-assisted first draft of an SIUU communication carries exactly the same compliance burden on the firm as a human-drafted one, with no separate standard to satisfy.
What EMA and EU Law Actually Say
The AI Reflection Paper Covers the Same Ground as FDA’s Draft Guidance
EMA’s Reflection Paper on the use of Artificial Intelligence in the medicinal product lifecycle, adopted by CHMP on September 9, 2024, addresses AI applications across drug discovery, non-clinical development, clinical trials, manufacturing, and post-authorization activities including pharmacovigilance [22][23]. Like FDA’s January 2025 draft guidance, the paper’s stated concern is that AI systems introduce new risks to patient safety and the integrity of clinical study results, and it sets out a risk-based, human-centered approach to AI throughout the medicinal product lifecycle [22]. Nothing in the paper addresses promotional communication, patient-facing chatbots, or AI-generated search content.
The EU’s Advertising Ban Is Purpose-Based, Not Format-Based
Directive 2001/83/EC prohibits advertising prescription-only medicinal products to the general public across the European Union under Article 88, and defines advertising broadly under Article 86 as any form of information, canvassing activity, or inducement designed to promote a medicinal product’s prescription, supply, sale, or consumption [24][25]. The Court of Justice of the European Union reaffirmed in its February 27, 2025 ruling in Apothekerkammer Nordrhein that the decisive factor in classifying a communication as advertising is its purpose, whether its primary aim is to promote prescription, supply, sale, or consumption of a medicinal product, not its format, channel, or author [26].
A Chatbot That Promotes a Prescription Drug in the EU Is Still Illegal Advertising
Because the CJEU’s test is purpose-based, an AI system that a company deploys, whether a branded chatbot or an AI-assisted marketing tool, and that functions to promote a prescription medicine to the general public falls squarely within Article 88’s prohibition, with no exception created by the fact that a model, rather than a copywriter, generated the output. The EU’s general pharmaceutical legislation reform, proposed in 2023 and amended by the European Parliament in April 2024, retains this same prohibition structure in its proposed replacement for Directive 2001/83/EC, while adding new provisions for what may be advertised to the public and preserving member states’ discretion to impose additional restrictions [27]. As of this writing, that reform has not addressed AI-generated content as a distinct category either.
Original Analysis: Six Categories of AI-Generated Drug Content, Mapped to the Rule That Actually Governs Each
Because no single document answers the compliance question directly, the table below sorts AI-generated drug content by who controls it and maps each category to the framework that applies today, based on the guidance and enforcement actions cited throughout this article.
Content type
Who controls it
Governing framework
Current status
Company-drafted promotional copy, AI-assisted
Manufacturer
21 CFR 202.1, OPDP review, Form FDA-2253 submission
Fully in force; AI assistance changes nothing
Branded chatbot on company site or app
Manufacturer
21 CFR 202.1; may also implicate labeling if dosing or use claims appear
Not manufacturer speech; company may respond under misinformation Q&A
Draft guidance only, non-binding, July 2024
Search engine AI-generated summaries/overviews
Search platform
Same as above; outside direct FDA/EMA jurisdiction over the platform
No dedicated framework
Methodology
This taxonomy was built by cross-referencing FDA and EMA’s own guidance scope statements, described in the sections above, against documented enforcement actions and guidance documents current as of September 2026. It reflects publicly available guidance and enforcement history rather than any regulator’s own published classification, and it will need revision if FDA finalizes its misinformation guidance or issues AI-specific promotional guidance.
The OPDP Capacity Problem
Submission Volume Against Review Staff
Promotional material submissions to FDA rose from 78,696 in 2012 to 149,516 in 2024, an increase of roughly 90 percent over twelve years, according to a Senate document reviewing FDA’s promotional oversight [9]. Over the same broad period, annual warning letters for promotional violations fell to one in 2023 and zero in 2024, before the September 2025 enforcement reversal [8]. The same Senate document reported that four OPDP leaders, including the office’s director and deputy director and the director and deputy director of its Division of Promotion Policy, Research, and Operations, had departed the agency, some under reduction-in-force notices [9].
What This Means for How Fast Violations Get Caught
A review office handling a growing submission volume with a shrinking senior staff is a review office more likely to rely on automated tools, exactly the AI-assisted surveillance FDA has already said it is implementing [8], and less likely to catch novel AI-generated violation patterns quickly through manual review alone. For companies, that argues for building internal review capacity that does not depend on FDA catching a problem first, since the agency’s own capacity to catch it has been visibly strained.
What This Means for Regulatory Affairs
Regulatory teams should stop searching for an AI-specific promotional rule that does not exist and instead confirm that existing medical, legal, and regulatory review workflows explicitly capture AI-assisted drafts before submission through Form FDA-2253, the same way they capture human drafts. The Purolea precedent, even though it is a manufacturing case, establishes the general principle FDA will apply if an AI-drafted promotional claim reaches enforcement: the absence of documented human review is itself treated as the compliance failure, independent of whether the underlying claim was also substantively wrong [1][12].
What This Means for Medical Affairs
Medical affairs teams sit closest to the gap this article identifies: FDA’s misinformation guidance is the only document that plausibly reaches independent AI chatbot output about a company’s drug, and it is a voluntary, non-binding draft [16][17]. Teams choosing to respond to an inaccurate AI-generated answer about their drug are operating in the same discretionary space FDA has described for responding to a mistaken influencer post, not under a defined AI obligation. Documenting the decision process for when and how to respond protects the company’s position if FDA ever finalizes the guidance and applies it retroactively to the company’s prior conduct.
What This Means for Pharmacovigilance and Brand Teams
Pharmacovigilance teams should treat AI-generated content the way they already treat any external source describing their drug: as a potential channel for adverse event signals, not exempt from case-intake processes simply because the source is a chatbot rather than a person. Brand and marketing teams should assume that AI-assisted copy receives no compliance benefit of the doubt; under 21 CFR 202.1 and FDA’s stated September 2025 enforcement scope, the standard for a claim’s accuracy and fair balance is identical whether a human or a model produced the first draft [6][10].
A Practical Compliance Checklist for AI-Generated Drug Content
Confirm every AI-assisted promotional draft passes through the same medical, legal, and regulatory review your MLR process already requires for human-drafted copy, with documentation of that review.
Classify any AI-generated content reaching a patient or prescriber by who controls it: your own branded tool is company speech; an independent AI platform’s output is not, and falls under the misinformation framework instead.
If using AI to draft CGMP, regulatory, or SIUU-related documents, apply the same human sign-off standard FDA articulated in the Purolea letter, regardless of whether the document is promotional in nature.
Build a recurring, documented process for sampling what major AI platforms say about your products against current prescribing information, since no regulator requires this today but the gap in independent oversight makes it a practical necessity.
Keep a record of any decision not to respond to AI-generated misinformation, since FDA’s current framework for responding is voluntary and discretionary rather than mandatory.
Track finalization of FDA’s misinformation guidance and any EU pharmaceutical legislation reform, both of which could convert current voluntary recommendations into binding requirements.
FDA’s July 2024 revised draft guidance on addressing misinformation remains unfinalized as of this writing, a decade after its 2014 predecessor was first proposed and also never finalized [14][17]. Its eventual finalization, in whatever form, is the single most direct regulatory development to watch for anyone managing AI-generated content risk today.
Pending: EU Pharmaceutical Legislation Reform
The EU’s proposed overhaul of Directive 2001/83/EC, amended by the European Parliament in April 2024, was still moving through the legislative process as of mid-2026 reporting, with the Council and Parliament negotiating remaining amendments [27]. The reform, as proposed, preserves the existing purpose-based advertising prohibition rather than replacing it with a format-specific or AI-specific test.
What to Watch for the Rest of 2026
Trade press reporting mid-2026 noted continued FDA warning letters citing AI-adjacent deficiencies following Purolea, an EU Annex 22 public consultation on AI in GMP manufacturing that included a multistakeholder workshop on July 1, 2026, and reports that the UK’s MHRA has flagged AI-generated inspection responses containing fabricated regulatory citations in compliance correspondence [28]. None of this reporting describes a dedicated promotional or misinformation rule for AI-generated content taking shape, reinforcing that the gap this article describes was still open as of September 2026.
Key Takeaways
As of this writing, neither FDA nor EMA has issued guidance that names AI-generated drug content, or AI-generated drug misinformation, as its own regulatory category.
FDA’s January 2025 draft AI guidance and the January 2026 FDA-EMA joint guiding principles both cover AI used to generate or analyze regulatory evidence, not promotional or patient-facing content.
FDA’s September 2025 DTC advertising crackdown, and legal analysis of its accompanying fact sheet, explicitly extends current enforcement scope to AI-generated content and chatbot interactions.
21 CFR 202.1’s fair balance and misleading-impression standards apply to AI-assisted promotional copy exactly as they apply to human-drafted copy.
FDA’s April 2026 Purolea warning letter is a manufacturing quality case, not a promotional or misinformation case, though its underlying logic, that AI output requires the same human review a human draft would require, extends by analogy.
FDA’s closest framework for AI-generated third-party misinformation is a July 2024 draft guidance that remains non-binding and unfinalized.
The EU’s advertising prohibition under Directive 2001/83/EC is purpose-based, and the CJEU confirmed in February 2025 that format and authorship do not change whether a communication counts as advertising.
Promotional material submissions to FDA rose roughly 90 percent from 2012 to 2024 while OPDP lost senior leadership, a capacity gap that predates and compounds the AI-content monitoring challenge.
FAQ
Has the FDA issued guidance specifically on AI-generated drug misinformation?
No. FDA’s January 2025 draft guidance on AI addresses evidence generation for regulatory submissions, not promotional or informational content, and a former OPDP official said publicly in March 2026 that no guideline tells industry how to use AI in ad promotion [2][5].
Does the FDA-EMA joint AI guidance apply to marketing or promotional content?
No. The January 2026 guiding principles define their own scope as AI used to generate or analyze evidence across nonclinical, clinical, post-marketing, and manufacturing phases, not AI used to draft or answer questions about promotional or medical information [3].
Can a company be held responsible for what a chatbot says about its drug?
It depends on control. A company-branded assistant is company speech and is subject to the same fair-balance and misbranding rules as any other promotional material. A general-purpose model like ChatGPT, Claude, or Gemini answering a question is not the manufacturer’s speech, so ordinary advertising rules do not directly apply, though the manufacturer can choose to respond under FDA’s voluntary misinformation framework [16].
What is the closest thing to an FDA enforcement action involving AI-generated drug content?
FDA’s April 2, 2026 warning letter to Purolea Cosmetics Lab, which cited a violation of 21 CFR 211.22(c) after the company used AI agents to draft manufacturing specifications and records without adequate human review. It is a manufacturing case, not a promotional case, but it establishes that FDA treats AI output as subject to existing review obligations rather than as its own regulatory category [1].
Is there an FDA framework for responding to AI-generated misinformation about a drug?
The closest fit is FDA’s July 2024 draft guidance, Addressing Misinformation About Medical Devices and Prescription Drugs: Questions and Answers, which replaced a 2014 draft on correcting third-party misinformation. It describes voluntary, non-binding recommendations and remains in draft status [15][17].
Does EU law treat AI-generated promotional content differently from human-written content?
No. Directive 2001/83/EC defines advertising by its purpose, and the CJEU confirmed in February 2025 that the decisive question is whether a communication is designed to promote prescription, supply, sale, or consumption of a medicinal product, regardless of format or author [24][26].
Has EMA published guidance on AI chatbots specifically?
No. EMA’s September 2024 Reflection Paper covers AI in drug discovery, clinical development, manufacturing, and pharmacovigilance. It does not address AI-generated conversational or search content directed at patients or clinicians [22].
Is FDA using AI to enforce advertising rules against companies?
FDA said in its September 9, 2025 crackdown announcement that it is already implementing AI and other tech-enabled tools to proactively surveil and review drug ads. The agency has not published details on how these tools are used or validated [8].
What happened to FDA’s capacity to review promotional material as AI content volume grows?
Promotional material submissions rose from 78,696 in 2012 to 149,516 in 2024, about a 90 percent increase, while four OPDP leaders, including the office’s director and deputy director, departed the agency, according to a Senate review document [9].
How can medical affairs and regulatory teams monitor what AI systems say about their drugs?
Because no regulator requires a specific methodology, teams generally build their own prompt libraries, sample multiple AI platforms on a recurring schedule, and compare responses against current prescribing information. Purpose-built tools such as DrugChatter’s AI monitoring platform support that comparison, though the compliance obligation to act on what is found comes from the existing promotional and misinformation rules described throughout this article, not from any AI-specific mandate.
References
U.S. Food and Drug Administration. (2026, April 2). Warning Letter to Purolea Cosmetics Lab, Ref. No. 722591. https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/purolea-cosmetics-lab-722591-04022026
U.S. Food and Drug Administration. (2025, January). Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-artificial-intelligence-support-regulatory-decision-making-drug-and-biological
U.S. Food and Drug Administration & European Medicines Agency. (2026, January). Guiding Principles of Good AI Practice in Drug Development. https://www.fda.gov/media/189581/download
Health Policy Watch. (2026, January 16). EU and US regulators reach landmark accord on AI principles in drug development. https://healthpolicy-watch.news/eu-and-us-ai-principles/
Eglovitch, J. (2026, March 10). Experts offer tips for using AI for drug advertising and promotion. Regulatory Focus, Regulatory Affairs Professionals Society. https://www.raps.org/resource/experts-offer-tips-for-using-ai-for-drug-advertisi.html
Legal Information Institute, Cornell Law School. 21 CFR § 202.1, Prescription-drug advertisements. https://www.law.cornell.edu/cfr/text/21/202.1
U.S. Food and Drug Administration. The Bad Ad Program. https://www.fda.gov/drugs/prescription-drug-advertising-and-promotional-labeling/bad-ad-program
U.S. Food and Drug Administration. (2025, September 9). FDA Launches Crackdown on Deceptive Drug Advertising [Press release]. https://www.fda.gov/news-events/press-announcements/fda-launches-crackdown-deceptive-drug-advertising
Durbin, R. J. (n.d.). FDA OPDP and DTC advertising oversight [Senate document]. https://www.durbin.senate.gov/imo/media/doc/Durbin_FDA%20OPDP-DTC.pdf
XDS. (2026). AI-Generated Pharma Content: FDA Compliance Guide. https://blog.madebyxds.com/ai-generated-pharma-content-fda-compliance-2026
DLA Piper. (2026, April). FDA Warning Letter highlights risks of using AI in drug manufacturing. https://www.dlapiper.com/en-us/insights/publications/2026/04/fda-warning-letter-highlights-risks-of-using-ai-in-drug-manufacturing
Greenlight Guru. (2026, April 28). What the Purolea warning letter really means for AI in medtech. https://www.greenlight.guru/blog/purolea-warning-letter-ai
Federal Register. (2014, June 18). Draft Guidance for Industry on Internet/Social Media Platforms: Correcting Independent Third-Party Misinformation About Prescription Drugs and Medical Devices. https://www.federalregister.gov/documents/2014/06/18/2014-14221/draft-guidance-for-industry-on-internetsocial-media-platforms-correcting-independent-third-party
Outside GC. (2024, August 6). Recent FDA Guidance Addressing Misinformation About Medical Devices and Prescription Drugs. https://outsidegc.com/blog/recent-fda-guidance-addressing-misinformation-about-medical-devices-and-prescription-drugs/
King & Spalding. (2024). FDA Issues Updated Draft Guidance on Addressing Misinformation About Medical Devices and Prescription Drugs. https://www.jdsupra.com/legalnews/fda-issues-updated-draft-guidance-on-7721762/
BioSpace. (2024, July 8). FDA Updates Guidance to Further Empower Companies to Address the Spread of Misinformation. https://www.biospace.com/article/releases/fda-updates-guidance-to-further-empower-companies-to-address-the-spread-of-misinformation/
U.S. Food and Drug Administration. Addressing Misinformation About Medical Devices and Prescription Drugs: Questions and Answers [Guidance document status page]. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/addressing-misinformation-about-medical-devices-and-prescription-drugs-questions-and-answers
Foley Hoag LLP. (2025, February 7). FDA Finalizes Guidance on Scientific Information on Unapproved Uses (SIUU) Communications. https://foleyhoag.com/news-and-insights/publications/alerts-and-updates/2025/february/fda-finalizes-guidance-on-scientific-information-on-unapproved-uses-siuu-communications/
King & Spalding. (2025, January 13). FDA Finalizes Guidance on Scientific Information on Unapproved Uses (SIUU). https://www.kslaw.com/news-and-insights/fda-finalizes-guidance-on-scientific-information-on-unapproved-uses-siuu
Ropes & Gray LLP. (2025, January 22). FDA Finalizes Guidance on Communication of Scientific Information on Unapproved Uses and Releases Updated First Amendment Analysis. https://www.ropesgray.com/en/insights/alerts/2025/01/fda-finalizes-guidance-on-communication-of-scientific-information
DistilINFO Publications. (2026, May 13). FDA SIUU Guidance Opens Off-Label Communication. https://distilinfo.com/2026/05/13/fda-siuu-guidance-opens-off-label-communication/
European Medicines Agency. (2024, September 9). Reflection paper on the use of Artificial Intelligence (AI) in the medicinal product lifecycle, EMA/CHMP/CVMP/83833/2023. https://www.ema.europa.eu/en/documents/scientific-guideline/reflection-paper-use-artificial-intelligence-ai-medicinal-product-lifecycle_en.pdf
NSF. (2024, October 16). EMA Reflection Paper on AI in the Medicinal Product Lifecycle. https://www.nsf.org/life-science-regulatory-news/ema-reflection-paper-on-ai-in-the-medicinal-product-lifecycle
Inside EU Life Sciences. (2023, January 31). CJEU Rules on the Advertising of “Unspecified Medicinal Products” in the EU. https://www.insideeulifesciences.com/2023/01/27/cjeu-rules-on-the-advertising-of-unspecified-medicinal-products-in-the-eu/
EuroGCT. Advertising of ATMPs. https://www.eurogct.org/research-pathways/commercialisation/advertising-atmps
BDK Advokati. (2025, March 20). New CJEU clarification on advertising of medicinal products under Directive 2001/83/EC. https://bdkadvokati.com/new-cjeu-clarification-on-advertising-of-medicinal-products-under-directive-2001-83-ec-relating-to-medicinal-products-for-human-use
Munungu, K. EU Pharmaceutical Law Reform: Key Changes on Advertising of Medicinal Products. https://biotalk.twobirds.com/post/102krpr/eu-pharmaceutical-law-reform-key-changes-on-advertising-of-medicinal-products
Sakara Digital. (2026, July 14). FDA AI Warning Letter Trends 2026: What Followed Purolea. https://sakaradigital.com/blog/2026-ai-warning-letter-trend-analysis-what-followed-purolea/
