The US FDA is utilizing AI in a bid to enhance efficiency. Photo credit: FDA
On May 8, 2025, US Food and Drug Administration (FDA) commissioner Marty Makary announced that the agency plans to deploy artificial intelligence tools across all FDA centers immediately. Under the aggressive timeline, all FDA offices are targeted to run on a unified, secure system tied to internal data platforms by June 30, 2025. The internal rollout, which aims to enhance efficiency in evaluating drugs, foods, medical devices, and diagnostic tests by reducing time spent on tedious, repetitive tasks, follows the successful completion of an 18-month generative AI pilot for scientific reviewers.
The technology, Makary said in an interview, could slash tasks in the review process for new therapies “from days to just minutes.” It is hoped to contribute to compressing at least a small part of the notoriously long drug-development timeline.
Set to be implemented across all FDA centers, including the Center for Tobacco Products (CTP), the AI tools might potentially accelerate CTP’s premarket tobacco product application (PMTA) process, which has been criticized as being inefficient and lacking transparency.
While CTP has not yet issued a public statement specifically addressing the agency’s AI rollout announcement, the center has previously demonstrated an interest in integrating advanced computational methods into its regulatory processes—for instance, by exploring the use of a nonclinical computational toxicology program to assess potential hazards associated with ingredients in tobacco products. In addition, CTP’s five-year strategic plan, released in December 2023 and no longer available on FDA’s website, outlined goals that aligned with the broader agency initiative to modernize business processes and enhance information management.
According to Robert Califf, who served as FDA commissioner from 2016 to 2017 and again from 2022 until January this year, the agency’s review teams have been using AI for several years, although there had never been a concerted effort. While being “enthusiastic” about FDA’s plan, he shared other experts’ caution about the tight timeline for implementation.
Although the pharmaceutical industry is expected to welcome any solution that might get their drugs to market faster and cut costs in the process, one of the key questions remains how the vast amount of proprietary data companies submit will be secured and whether sufficient guardrails are in place. Critics also warn that important decisions could get automated, thus putting Americans at risk. Furthermore, it is not clear at this point which models are being used to train the AI and what inputs are provided for specialized fine-tuning.
According to press reports, FDA has recently been holding talks with OpenAI to discuss its use of AI. Still under Califf, the agency in January 2025 also issued draft guidance to ensure the credibility of AI models used in regulatory decisions, which emphasizes the need for robust principles, standards, and best practices to manage risks associated with AI, including biases and cybersecurity risks.
Whether this guidance stays relevant under Trump’s overhaul of federal AI policy—with Elon Musk leading the charge under an “AI first” strategy—remains to be seen. FDA’s AI initiative is led by Jeremy Walsh, the agency’s newly appointed, first-ever AI officer and previously chief technologist at Booz Allen Hamilton, alongside long-time FDA data leader Sridhar Mantha.