The Prevalence of AI in FDA Correspondence

All sections have been abridged for brevity from original submissions; please contact us for the full preprint.

Background (Abridged)

The U.S. Food and Drug Administration (FDA) has responded to generative artificial intelligence (AI) on three fronts: internal adoption, industry guidance, and industry correspondence. The agency has deployed its own internal generative AI tool, such as the publicly announced Elsa, to accelerate scientific review, protocol assessment, and inspection targeting [8,9]. It has also issued a series of discussion papers and draft guidance for industry covering AI use across the drug, biologic, and device lifecycle [3,4,5,6,7]. Third, it has begun referencing industry AI use directly in regulatory correspondence, including warning letters and establishment inspection reports (EIRs), which is the focus of this review.

On April 2, 2026, CDER issued a warning letter to Purolea Cosmetics Lab that included a dedicated subsection titled "Inappropriate Use of Artificial Intelligence in Pharmaceutical Manufacturing," citing the firm's use of AI agents to help comply with FDA regulations [10]. That letter is the first CDER warning letter to identify AI as a contributing factor to an adulteration finding for a finished pharmaceutical. Figure 1 plots both trajectories against each other. The purpose of this paper is to highlight the pace at which agency responses to this technology have accelerated, identify overlooked aspects of widely circulated media coverage of warning letters and other documents, and establish protocols for analyzing new technologies used by the agency.

Figure 1. FDA AI-related guidance, policy, and internal tooling (top band) vs. FDA regulatory correspondence referencing AI (bottom band), January 2022 through April 2026. Source: keyword search ("Artificial Intelligence") across the FDA warning letter database, the FDA CDER AI in Drug Development guidance index, and a regulatory intelligence corpus that indexes EIRs and inferred Form 483 content.

Methods (Abridged)

This review conducted a keyword-based retrospective search of FDA-issued correspondence, consistent with recently published extraction methods for FDA warning letters that combine regex filtering and web scraping to parse warning letter text into structured citations [11,12,13]. The search term "Artificial Intelligence" was applied to the FDA warning letter corpus and cross-referenced against an independent regulatory intelligence database that also indexed FDA establishment inspection reports (EIRs), Form 483 observations inferred from warning letter text, and FDA guidance documents.

Results (Abridged)

The Purolea letter is a CDER-issued CGMP letter for a finished pharmaceutical, citing 21 CFR 211.165(b), 211.84(d), and 211.22, with a dedicated subsection on AI misuse as part of a broader adulteration finding under Section 501(a)(2)(B) [10]. Figure 1 shows the issuance cadence of AI-related guidance accelerating through 2025 and into 2026, with enforcement references appearing roughly twelve months after corresponding guidance issuance.

Discussion (Abridged)

In each of the five regulatory correspondence documents reviewed, the AI reference is one of eight to nine observations, and in every case, the AI language is paired with more severe, better-established citations: unapproved device marketing, failure of the quality unit, inadequate microbiological testing, and inadequate CAPA.

Conclusion (Abridged)

While a wave of news coverage centers on AI-specific warning letters and related topics, the reality is far more grounded. The most important finding of this review is not that AI-related observations exist, but that they remain a small and heterogeneous subset of what the FDA is actually citing. In each of the five documents reviewed, the AI reference is one of eight to nine observations, and in every case, the AI language is paired with more severe, better-established citations.

Citations (Abridged)

Citation numbers match the full paper. Omitted numbers are cited only in sections not reproduced here.

3. FDA. Artificial Intelligence for Drug Development (CDER landing page). https://www.fda.gov/about-fda/center-drug-evaluation-and-research-cder/artificial-intelligence-drug-development

4. FDA. Using Artificial Intelligence and Machine Learning in the Development of Drug and Biological Products: Discussion Paper and Request for Feedback. May 2023; revised February 2025.

5. FDA. Artificial Intelligence and Medical Products: How CBER, CDER, CDRH, and OCP are Working Together. March 2024; revised February 2025.

6. FDA. Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drug and Biological Products. Draft Guidance. January 7, 2025. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-artificial-intelligence-support-regulatory-decision-making-drug-and-biological

7. FDA. Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations. Draft Guidance. January 7, 2025. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/artificial-intelligence-enabled-device-software-functions-lifecycle-management-and-marketing

8. FDA. FDA Launches Agency-Wide AI Tool to Optimize Performance for the American People. Press release. June 2, 2025. https://www.fda.gov/news-events/press-announcements/fda-launches-agency-wide-ai-tool-optimize-performance-american-people

9. Al-Faruque F. FDA's Elsa AI tool gets mixed response from some staff. Regulatory Focus (RAPS). June 4, 2025. https://www.raps.org/news-and-articles/news-articles/2025/6/fda-s-elsa-ai-tool-gets-mixed-response-from-some-s

10. FDA. Warning Letter to Purolea Cosmetics Lab. MARCS-CMS 722591. April 2, 2026. https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/purolea-cosmetics-lab-722591-04022026

11. Kwiecinski G. An Analysis of FDA Warning Letter Citations from 2019–2023. Journal of Pharmaceutical Innovation. 2024;19:78. https://doi.org/10.1007/s12247-024-09879-x

12. Kwiecinski G, Yuan K. The Foreign Inspection Gap: FDA GMP Oversight of U.S. Drug Imports, 2014–2024. Journal of Pharmaceutical Innovation. 2026. https://doi.org/10.1007/s12247-026-10406-3

13. Aglave G, Yerram S, Patnam JD, et al. Proactive approaches to CGMP compliance: insights and key takeaways from USFDA warning letters to pharmaceutical industries between 2019 and 2024. Journal of Pharmaceutical Innovation. 2024;19(4). https://doi.org/10.1007/s12247-024-09857-3