Generative AI Medical Devices Draw FDA Request for Feedback

The FDA has opened a request for feedback on generative AI medical devices, inviting responses from manufacturers, clinicians, researchers, and the public under docket FDA-2026-N-7874. Submissions are due by October 19, 2026, and the request accompanies a discussion paper released by the Digital Health Center of Excellence within the Center for Devices and Radiological Health. 

The paper, titled Considerations for the Regulation of Generative AI-Enabled Medical Devices, does not create obligations for any company currently developing or marketing a product in this category. It does, however, set out the questions the agency is working through, and those questions offer regulatory affairs teams a rare look at where FDA’s thinking on generative AI medical devices remains unsettled. 

What The Discussion Paper Is And Is Not 

FDA has been direct about the limits of the document, and the framing deserves attention because early coverage of pre-rulemaking materials often overstates their weight. 

According to the agency, the paper is intended for discussion purposes only and represents neither draft nor final guidance. It does not propose or implement policy changes regarding how CDRH intends to regulate generative AI-enabled devices. Nor does it communicate the centre’s proposed or final regulatory expectations, including expectations for supporting evidence in future marketing submissions. 

One further boundary is worth registering. FDA states that the paper does not address whether the approaches it discusses fall within the agency’s existing legal authorities, or whether new authorities would be required to implement them. That caveat matters for anyone reading the document as a forecast of near-term rulemaking, because it signals that some of the ideas under discussion may require legislative action rather than agency guidance alone. 

The stated purpose is narrower and more useful than a forecast. FDA wants early input from groups and individuals outside the agency, and it wants to advance a broader conversation among stakeholders on how this category should be regulated. 

The Four Topic Areas Under Review 

The discussion paper is organized around four areas, each accompanied by its own set of questions: 

• Considerations for the assessment of risk for generative AI-enabled medical devices 
• Considerations for premarket evaluation of generative AI-enabled medical devices 
• Postmarket monitoring for generative AI-enabled medical devices 
• Other topics relevant to the regulation of generative AI-enabled medical devices 

Underpinning all four is the agency’s position that these devices hold significant promise for patient care and the broader health ecosystem, while potentially introducing risks that differ from those presented by traditional software and by earlier AI-enabled medical devices. That distinction between generative systems and the AI-enabled devices FDA has already authorized runs through the document. 

The full paper is available as a PDF download of roughly 8.8 MB. Regulatory professionals with generative AI on the product roadmap should read it in full rather than relying on secondary summaries, since the value sits in the specific wording of the questions. 

Who FDA Wants To Hear From 

The request is open to device manufacturers, clinicians, researchers, members of the public, and other interested parties. Feedback goes through Regulations.gov under docket number FDA-2026-N-7874. 

A detail in FDA’s instructions may change the calculation for smaller regulatory teams. Respondents are not required to answer every question in the paper. The agency states that interested parties may address only those questions, combinations of questions, or topics relevant to their expertise, experience, or organizational capacity, and that partial responses are appropriate where they make sense. 

Practically, that means a focused submission drawing on two or three areas of genuine operating experience is a legitimate contribution. Companies frequently assume that participating in a federal docket requires a comprehensive position paper prepared with outside counsel, and this request explicitly says otherwise. 

What This Means For Your Business 

Because the paper carries no compliance requirements, the decision facing most companies is about time and attention rather than obligation. 

Reading the questions against your own product architecture is the most direct benefit available. Some organizations have generative components operating inside a device software function without having classified them that way internally, and the paper’s risk assessment questions provide a structured way to surface that. Quality Smart Solutions works with device companies on exactly this kind of internal mapping exercise, particularly where a product sits near the boundary between a regulated device software function and a general-purpose tool. 

Drafting a response produces a second benefit that outlasts the docket. Articulating how a system behaves, how the company evaluates its performance, and how that performance is monitored after release forces a level of documentation that tends to be useful in later submissions, whatever form FDA’s eventual guidance takes. 

For companies already holding authorizations for AI-enabled devices, the postmarket monitoring questions are the ones to read closely. FDA has not indicated any change to existing obligations, and none should be inferred from the paper. Still, the questions the agency has chosen to ask indicate which aspects of ongoing device performance are receiving attention.