BOSTON — Rick Abramson, director of the Food and Drug Administration’s Digital Health Center of Excellence, said to expect draft guidance on generative artificial intelligence in the coming months at AdvaMed’s The MedTech Conference on Tuesday.
The comment period closed on Monday for an FDA discussion paper on regulatory approaches to generative AI, including a greater emphasis on postmarket monitoring. As of Tuesday, the agency had received more than 500 submissions.
“This is not just an academic paper. It’s not just a thought exercise. We intend to draft policy, and it’ll come soon,” Abramson said. “This is a priority at the highest levels of the administration. … I’m talking about the level of the secretary and the White House.”
Abramson didn’t commit to a specific timeline, as the agency needs time to wade through the comments, but he said to expect draft guidance in a matter of months.
On the device center’s list of priority guidances for fiscal 2027, a draft on generative AI-enabled devices for mental disorders appears on the FDA’s “A list,” while another draft on risk assessment and evidence generation for generative AI-enabled devices is under construction.
The former is the direct output of digital health advisory committee meetings last year discussing mental health chatbots, Abramson said, while the latter is more of an “umbrella guidance.”
Another priority for the agency is finalizing the AI lifecycle management guidance that was drafted in early 2025, Abramson said. That policy is designed for more “static” AI devices.
“This is very, very complex and very important policy,” Abramson said of the generative AI discussion paper. “We’re going to be setting precedent that lasts for years, and so we want to do it right.”
MHRA policy update
Jessica Paulsen, deputy director of AI and software for the Medicines and Healthcare Products Regulatory Agency, also provided updates on policies to watch at the U.K. regulator. Paulsen joined the MHRA earlier this year after working at the FDA for more than 15 years, as the agency looks to work more closely with its U.S. counterpart.
Earlier this month, the MHRA moved to adopt 44 AI recommendations made by an independent advisory group called the National Commission into the Regulation of AI in Healthcare. Some of the recommendations carried similar themes to the U.S. discussion paper, such as the need for performance monitoring of devices that use generative AI.
“I do think we are currently in a bit of a state where we may be over-relying on single point-in-time assessments that may not be representative of how some of these AI products are performing in real world settings,” Paulsen said, noting the models are context dependent and can change.
The MHRA plans to draft guidance by the end of the year on a pre-determined change control framework to manage changes to medical devices after they are in the market, with some similarities to the FDA’s framework.
Early next year, the MHRA will also consult on how to qualify and classify AI-enabled devices, and the regulator plans to share an implementation roadmap for the 44 recommendations by spring.
