The Food and Drug Administration has been holding meetings with OpenAI about how AI could be used inside the agency, according to sources with knowledge of the discussions cited by WIRED. The talks appear to connect to a wider FDA effort to modernize parts of drug evaluation and make reviews move faster.
The idea is not being presented as a replacement for human review. The source article describes a more limited, still-developing push: using AI to assist certain agency tasks while questions remain about reliability, policy, data quality and how much of the drug approval process can realistically be compressed.
What The FDA And OpenAI Have Discussed
According to the source article, a small OpenAI team has met with the FDA and two associates of Elon Musk's so-called Department of Government Efficiency multiple times in recent weeks. The group has discussed a project called cderGPT, which likely refers to Center for Drug Evaluation and Research GPT.
The Center for Drug Evaluation and Research regulates over-the-counter and prescription drugs in the US. Jeremy Walsh, recently named as the FDA's first-ever AI officer, has led the discussions. So far, no contract has been signed.
OpenAI declined to comment. The article also reports that Walsh has met with Peter Bowman-Davis, an undergraduate on leave from Yale who currently serves as acting chief AI officer at the Department of Health and Human Services, to discuss the FDA's AI ambitions. Politico first reported the appointment of Bowman-Davis, who is part of Andreessen Horowitz's American Dynamism team.
The Push To Speed Drug Evaluation
FDA commissioner Marty Makary has publicly framed AI as part of a larger modernization effort. On Wednesday, he wrote on X: "Why does it take over 10 years for a new drug to come to market?" He also wrote: "Why are we not modernized with AI and other things? We've just completed our first AI-assisted scientific review for a product and that's just the beginning."
Those comments followed an annual meeting of the American Hospital Association earlier this week, where Makary spoke about AI's potential to help with approval of new treatments for diabetes and certain types of cancer. The source article notes that Makary did not specify OpenAI as part of the initiative.
The FDA review process currently takes about a year. The agency already has several mechanisms intended to move promising drugs more quickly, including fast track designation for products designed to treat a serious condition and fill an unmet medical need. Breakthrough therapy designation, created in 2012, can allow priority review for drug candidates that may offer a substantial benefit to patients compared with current treatment options.
Where AI May Help First
Robert Califf, who served as FDA commissioner from 2016 to 2017 and again from 2022 through January, told WIRED that FDA review teams have used AI for several years. He said: "It will be interesting to hear the details of which parts of the review were 'AI assisted' and what that means." He added: "There has always been a quest to shorten review times and a broad consensus that AI could help."
Califf also said that before he left the agency, the FDA was considering ways AI could support internal operations. In his view, final approval reviews are only one piece of a broader opportunity.
Rafael Rosengarten, CEO of Genialis, a precision oncology company, and a cofounder and board member of the Alliance for AI in Healthcare, supports automation for certain drug-review tasks. One practical example he gave is checking whether an application is complete. Faster feedback on missing or incomplete materials could help submitters address issues sooner.
But Rosengarten also emphasized limits. More advanced uses would need to be developed, tested and proved out. He said policy guidance should address what data is used to train AI models and what level of model performance is acceptable.
The Reliability Question
The main concern is not whether AI can process information quickly. It is whether it can do so in a setting where errors matter. Rosengarten put the issue in training terms: "These machines are incredibly adept at learning information, but they have to be trained in a way so they're learning what we want them to learn."
An ex-FDA employee who has tested ChatGPT as a clinical tool raised a separate concern: AI models can fabricate convincing information. That tendency creates uncertainty around how dependable a chatbot would be for reviewers' tasks. The ex-staffer said: "Who knows how robust the platform will be for these reviewers' tasks."
Andrew Powaleny, a spokesperson for the industry group PhRMA, also framed the issue around patient needs and risk. He said: "Ensuring medicines can be reviewed for safety and effectiveness in a timely manner to address patient needs is critical." He added: "While AI is still developing, harnessing it requires a thoughtful and risk-based approach with patients at the center."
What Comes Next
The FDA is already exploring AI internally. In December 2023, the agency advertised a fellowship for a researcher to develop large language models for internal use. The fellowship description said the work could include applications of LLMs for precision medicine, drug development and regulatory science.
OpenAI has also moved toward government-focused products. In January, it announced ChatGPT Gov, a self-hosted version of its chatbot designed to comply with government regulations. The company also said it was working toward FedRAMP moderate and high accreditations for ChatGPT Enterprise, which would allow it to handle sensitive government data.
For now, the clearest picture is cautious experimentation. AI may help the FDA handle narrower tasks, organize internal work and possibly shorten parts of the review process. But the source article makes one limitation plain: final review is only a small part of the long drug-development timeline, and most drugs fail before they ever reach FDA review.