FDA offers staff ‘agentic AI’ to support premarket reviews, administrative tasks

The Food and Drug Administration on Monday announced plans to offer its employees a broader set of artificial intelligence tools to use in premarket reviews and for other purposes amid persistent concerns that the technology can behave unpredictably.

The agency touts in a release that staff will now be able to use “agentic AI capabilities” to assist with “meeting management, pre-market reviews, review validation, post-market surveillance, inspections and compliance and administrative functions.” Agentic AI broadly refers to systems that can complete multistep tasks autonomously. The announcement says that the agency is employing “guidelines — including human oversight — to ensure reliable outcomes.” 

Asked for more details about these guidelines, Ben Nichols, an FDA spokesperson, wrote in an email that the “agentic AI tools are exploratory” and that the AI does not make regulatory decisions or replace human judgement. “All outputs from AI are reviewed and validated” by FDA staff “before being incorporated into any official regulatory action, ensuring that the AI remains a support tool rather than a decision maker,” he wrote. (The FDA’s drug center alone has lost over 1,000 staff...