PolicyAI02 sources

FDA opens a public docket on generative AI medical devices

Photorealistic close-up of a pair of hands turning printed policy documents beside a tablet displaying a medical imaging scan with a software-generated overlay, on a desk in an office.

AI-generated illustrationAI-generated illustration

The FDA’s Center for Devices and Radiological Health published a discussion paper on 18 August 2026 on how generative AI-enabled medical devices might be regulated, and opened docket FDA-2026-N-7874 on Regulations.gov for comment until 19 October 2026.

The agency is explicit that this proposes nothing. The paper is not a draft guidance and does not state CDRH’s regulatory expectations; it asks manufacturers, clinicians and researchers what risk assessment, premarket evaluation and postmarket monitoring should look like for this class of device.

The gap it is aimed at is real. The existing framework was built for models that do one bounded thing and can be locked, validated against a reference standard and re-validated when they change. A model that generates open-ended text or images has no fixed output space to test against, drifts when the underlying foundation model is updated, and can fail in ways that read as fluent.

What the docket produces is not a rule. It is the record a future rule gets built on, and the comment window is where the terms are set — which is why device makers, hospitals and the patient side all have reason to be in it before October.

Sources

  1. [1]FDA Seeks Public Feedback to Inform Regulatory Approach for Generative AI-Enabled Medical DevicesU.S. Food and Drug Administration··Newsroom
  2. [2]Considerations for the Regulation of Generative AI-Enabled Medical Devices: Discussion Paper and Request for FeedbackU.S. Food and Drug Administration··Report