The FDA is asking for public input on how to regulate generative AI medical devices. A discussion paper issued Tuesday sketches a two-axis risk framework and a doctor-like competency assessment for premarket review. Comments run through October 19 under docket FDA-2026-N-7874 on Regulations.gov.
The paper, released by the agency’s Digital Health Center of Excellence, covers risk assessment, premarket evaluation, postmarket monitoring, foundation models and agentic AI systems. Regulators want input from device makers, clinicians, researchers and the public.
Under the proposed premarket approach, a GenAI device would be tested through non-clinical benchmarking plus clinical confirmation to show it performs as intended before reaching patients, an idea the FDA says is inspired at a high level by how physicians are trained and evaluated. That matters because generative models can change behavior over time in ways conventional locked algorithms do not.
Acting FDA Commissioner Kyle Diamantas said the U.S. must lead in shaping how AI is developed and used safely, and CDRH Director Michelle Tarver said the process aims to keep pace with rapid digital health innovation. The paper is part of the administration’s push to harness AI to speed innovative medical products to market.
Industry watchers are already parsing the details, with some noting the competency-assessment idea could reshape how adaptive and agentic software is cleared. The FDA framed the document as a first step toward a regulatory approach that could serve as a model for other countries.
