FDA stresses human accountability as AI expands in drug review processes

FDA expands scrutiny of AI use in drug manufacturing and regulatory processes, emphasizing companies remain accountable for AI-generated outputs.

Objective Facts

On April 2, 2026, the FDA issued its first AI-related warning letter to Purolea Cosmetics Lab, a Livonia, Michigan contract manufacturer, citing overreliance on artificial intelligence in a Current Good Manufacturing Practice context. The firm used AI agents to generate drug product specifications, procedures, and master production or control records, but its quality unit failed to review or verify those AI-generated outputs before incorporation into manufacturing. The warning letter signals FDA expectations around AI are no longer theoretical but being enforced under existing CGMP frameworks; most critically, the firm failed to independently verify outputs and was unaware that process validation was required because the AI system never told them. FDA's posture is that AI may inform work but does not own accountability—someone must remain responsible for decisions and be able to explain why the decision was appropriate despite uncertainty, model limitations, or data gaps.

Left-Leaning Perspective

Rep. Jake Auchincloss, a Massachusetts Democrat on a House health subcommittee, stated details about the FDA's voucher program have been 'shrouded in secrecy,' calling for transparency as 'drug approvals have been made almost wholly and in an unprecedented manner by the FDA's political leadership.' Sen. Bernie Sanders of Vermont and Rep. Frank Pallone of New Jersey, with Pallone as the top Democrat on the House Energy and Commerce Committee, sent a letter seeking answers to 15 questions about the voucher program, to which the FDA did not respond. Critics including those at Harvard warned in the New England Journal of Medicine that the expedited review program could strain FDA staff already depleted from layoffs, 'potentially leading to safety issues,' and that it 'could undermine trust in the FDA's approval decisions' due to 'broad and opaque' criteria and being 'ripe for both abuse and legal challenge.' Makary's program faces 'intensifying scrutiny from Congress and the medical establishment over whether it's putting politics over science,' with concerns about legal challenges and rushed reviews as departures mount at the drug center. Dr. Reshma Ramachandran of Yale accused the FDA of 'parroting Big Pharma's wishlist,' arguing the AI push reads 'straight out of PhRMA's playbook.' The left's coverage emphasizes AI acceleration risks and the prioritization of speed over rigorous safety review, downplaying the positive potential for faster access to medications for serious diseases.

Right-Leaning Perspective

Former FDA Commissioner Scott Gottlieb, who led the agency from 2017-2019 and is now a senior fellow at the conservative American Enterprise Institute, expressed concern in April 2026 that agency changes could bring back days of 'drug lag,' arguing 'through a generation of congressional actions, investments in expertise and hiring, and careful policymaking, we built the FDA into the most efficient, forward-leaning drug regulatory agency in the world.' Former Republican Senator Rick Santorum at a March 2026 CNBC panel pointed toward FDA leadership of Vinay Prasad, saying 'You have someone who's running one of the departments who's a long critic of this' accelerated approval, and 'now you're seeing the results' in drops in accelerated approvals and rejections of rare disease drugs. Gottlieb said FDA is 'falling short of President Trump's goals' with 'significant staff reductions' and loss of 'major departures' from the Center for Drug Evaluation and Research, arguing 'when you lose the folks who have been doing this a long time... it starts to impact review decisions,' with part driven by 'the orientation of some of the people currently leading these centers.' Gottlieb stated 'recent changes to policies related to the regulation of AI have added new uncertainties' and called for the FDA to revert to exempting more clinical decision support software from premarket review if 'designed to augment the information available to clinicians and do not provide autonomous diagnoses or treatment decisions.' Right-leaning sources frame the accountability emphasis as regulatory caution that slows innovation, downplaying legitimate safety concerns about AI oversight.

Deep Dive

Under FDA Commissioner Marty Makary in the Trump administration, the agency has undergone significant change, including shifting approval standards, embedding AI in review workflows, and offering expedited reviews for medicines supporting 'national interests,' all reflecting an agenda centered on accelerating drug development. Simultaneously, on January 14, 2026, the FDA and EMA jointly released 10 Guiding Principles of Good AI Practice covering the entire product lifecycle, emphasizing human-centric design, a risk-based approach, multidisciplinary expertise, and clear documentation. The April 2026 Purolea warning letter reveals a critical tension: the failure reflected three stacked failures—a management decision to substitute AI for regulatory expertise, a Quality Unit without literacy to detect omissions, and an accountability structure that approved output anyway. This echoes broader organizational failures seen in other sectors when adoption outpaces literacy. However, both left and right miss important aspects of the underlying debate. The left focuses on political pressure for speed without acknowledging legitimate gains from expedited pathways for serious diseases; the right emphasizes lost expertise without addressing genuine risks of over-reliance on algorithmic outputs that can hallucinate or omit critical information. Researcher Aaron Kesselheim argues 'it doesn't make any sense for the FDA to voluntarily take power away from itself to make decisions about what kind of testing is necessary,' suggesting FDA already had flexibility in oversight—the question is how to use it responsibly. The real issue is not whether to use AI, but how to ensure humans remain meaningfully accountable—a position both sides nominally support but implement very differently. The FDA received extensive feedback on its draft guidance from industry, academia, patient groups, and software vendors, with final guidance expected in Q2 2026. Watch for whether final guidance clarifies accountability for novel AI types (generative AI, large language models) that evolved after the draft was written, and whether it provides sufficient flexibility for legitimate innovation while preventing the failures revealed in the Purolea case.

OBJ SPEAKING

Create StoryTimelinesVoter ToolsRegional AnalysisPolicy GuideAll StoriesCommunity PicksUSWorldPoliticsBusinessHealthEntertainmentTechnologyAbout

FDA stresses human accountability as AI expands in drug review processes

FDA expands scrutiny of AI use in drug manufacturing and regulatory processes, emphasizing companies remain accountable for AI-generated outputs.

May 26, 2026
FDA stresses human accountability as AI expands in drug review processesVia Wikimedia (contextual reference image) · Subscribe to support objective journalism and fund real-time news imagery
What's Going On

On April 2, 2026, the FDA issued its first AI-related warning letter to Purolea Cosmetics Lab, a Livonia, Michigan contract manufacturer, citing overreliance on artificial intelligence in a Current Good Manufacturing Practice context. The firm used AI agents to generate drug product specifications, procedures, and master production or control records, but its quality unit failed to review or verify those AI-generated outputs before incorporation into manufacturing. The warning letter signals FDA expectations around AI are no longer theoretical but being enforced under existing CGMP frameworks; most critically, the firm failed to independently verify outputs and was unaware that process validation was required because the AI system never told them. FDA's posture is that AI may inform work but does not own accountability—someone must remain responsible for decisions and be able to explain why the decision was appropriate despite uncertainty, model limitations, or data gaps.

Left says: Democratic critics worry drug decision-making is being taken away from agency scientists, with concerns that political leadership rather than career staffers is driving accelerated approval decisions.
Right says: Conservative commentators argue the FDA is falling short of speeding drug approval due to leadership ideology and loss of experienced senior review staff, with former FDA chief warning that losing key personnel impacts review decisions and slows access to innovation.
✓ Common Ground
Both industry and regulators across perspectives acknowledge that someone must remain responsible for AI-assisted decisions and be able to explain why decisions were appropriate, and that regulatory judgment and final decision-making must remain with trained experts.
There is alignment that AI-assisted content must be reviewed by competent personnel to ensure accuracy and CGMP compliance, and that Quality Unit responsibility is not reduced when AI is involved, with the message being that AI use requires proper governance, literacy, and expertise.
Both conservative and progressive sources agree the FDA is not discouraging AI adoption but requiring it be implemented within a controlled, risk-based framework, and that benefits are sustainable only when paired with structured governance, validation, and monitoring.
Even critics of accelerated approval acknowledge 'there's no easy way for the agency to strike the perfect balance between sufficient speed and ample information,' particularly for serious or life-threatening diseases.
Objective Deep Dive

Under FDA Commissioner Marty Makary in the Trump administration, the agency has undergone significant change, including shifting approval standards, embedding AI in review workflows, and offering expedited reviews for medicines supporting 'national interests,' all reflecting an agenda centered on accelerating drug development. Simultaneously, on January 14, 2026, the FDA and EMA jointly released 10 Guiding Principles of Good AI Practice covering the entire product lifecycle, emphasizing human-centric design, a risk-based approach, multidisciplinary expertise, and clear documentation.

The April 2026 Purolea warning letter reveals a critical tension: the failure reflected three stacked failures—a management decision to substitute AI for regulatory expertise, a Quality Unit without literacy to detect omissions, and an accountability structure that approved output anyway. This echoes broader organizational failures seen in other sectors when adoption outpaces literacy. However, both left and right miss important aspects of the underlying debate. The left focuses on political pressure for speed without acknowledging legitimate gains from expedited pathways for serious diseases; the right emphasizes lost expertise without addressing genuine risks of over-reliance on algorithmic outputs that can hallucinate or omit critical information. Researcher Aaron Kesselheim argues 'it doesn't make any sense for the FDA to voluntarily take power away from itself to make decisions about what kind of testing is necessary,' suggesting FDA already had flexibility in oversight—the question is how to use it responsibly. The real issue is not whether to use AI, but how to ensure humans remain meaningfully accountable—a position both sides nominally support but implement very differently.

The FDA received extensive feedback on its draft guidance from industry, academia, patient groups, and software vendors, with final guidance expected in Q2 2026. Watch for whether final guidance clarifies accountability for novel AI types (generative AI, large language models) that evolved after the draft was written, and whether it provides sufficient flexibility for legitimate innovation while preventing the failures revealed in the Purolea case.

◈ Tone Comparison

Democratic coverage uses phrases like 'intensifying scrutiny,' 'putting politics over science,' and emphasizes concerns about 'legal challenges' and 'rushed reviews,' while conservative outlets like National Review characterize the FDA's accountability stance as regulatory 'excess' and complaint of 'talks a big game on deregulation, but it's not paired with real action,' reflecting fundamentally different views on whether the issue is too much or too little oversight.