FDA seeks input on generative AI devices

1/ The FDA has opened a public comment process on how it should regulate medical devices that use generative AI. On Aug. 18, the agency released a discussion paper covering risk assessment, premarket review, postmarket monitoring and related issues for GenAI-enabled devices. (fda.gov) 2/ This is not new guidance yet. The FDA says the paper is “intended for discussion purposes only,” does not represent draft or final guidance, and is not meant to set final regulatory expectations for future submissions. (fda.gov) 3/ The process is being led by the Digital Health Center of Excellence inside the FDA’s Center for Devices and Radiological Health. The public docket is FDA-2026-N-7874, and comments are due by Oct. 19, 2026. (fda.gov) 4/ What problem is the agency trying to solve? The FDA says GenAI medical devices can differ from traditional software because they may accept open-ended inputs, perform multiple subtasks and produce variable outputs to similar inputs. That makes them harder to assess with conventional device review alone. (fda.gov) 5/ The FDA is also signaling that it wants a life-cycle approach, not just a one-time approval decision. In the paper, CDRH says it expects a risk-based approach tied to intended use and technological characteristics, and says a total product life cycle framework will likely be important for GenAI-enabled devices. (fda.gov) 6/ A key feature is the agency’s focus on use, not just model type. The FDA press release says the paper outlines a possible two-axis framework for risk assessment, which would be used to inform regulatory expectations for these products. (fda.gov) 7/ The practical implication: the FDA appears to be asking how to judge whether a GenAI device is reliable for a specific medical job, rather than treating all large models or foundation models as one category. That is an inference from the agency’s emphasis on risk, intended use and premarket evaluation, not a final FDA rule. (fda.gov) 8/ The agency is framing this as part of a broader AI push. Acting FDA Commissioner Kyle Diamantas said on Aug. 18 that “Artificial intelligence is transforming medicine,” while CDRH Director Michelle Tarver said patients and clinicians need “a regulatory approach that keeps pace” with digital health innovation. (fda.gov) 9/ Another important detail: the FDA says the paper may inform future policy, but it is not yet deciding whether every approach discussed fits within existing legal authority. The document explicitly says it is not intended to address whether new legal authorities would be necessary. (fda.gov) 10/ That matters for companies building AI diagnostic or therapeutic tools. For now, the FDA is gathering input from manufacturers, clinicians, researchers and the public on what evidence, safeguards and monitoring should be expected before and after these devices reach the market. (fda.gov) 11/ The consultation also fits into a larger FDA effort around AI devices. The agency maintains an AI-enabled medical devices list and says it plans to explore ways to identify and tag devices that incorporate foundation models, including LLM-based functionality, in future updates. (fda.gov) 12/ So the immediate takeaway is procedural, not dispositive: the FDA has started a formal input process on GenAI medical device oversight, with comments due Oct. 19. The next concrete milestone is whether the agency turns those responses into draft guidance or another formal regulatory document. (fda.gov)