ATA Action urges FDA to build risk-based framework for AI-enabled medical devices

WASHINGTON – ATA Action, the advocacy-focused affiliate of the American Telemedicine Association (ATA), has submitted comments in response to the U.S. Food and Drug Administration's discussion paper on AI-enabled medical devices, saying manufacturers should remain responsible for the safety and performances of the devices they sell.

The organization says the FDA should avoid shifting monitoring duties to the small, rural and safety-net practices that lack the staff to take them on and should instead give clinicians simple ways to report problems.

“AI has enormous potential to expand access to high-quality virtual care, but only if the regulatory framework is clear, proportionate and built to last,” said Joe Nye, head of federal government relations for ATA Action. “FDA’s discussion paper is a strong start. We look forward to working with the agency to turn it into guidance that protects patients, gives developers and clinicians predictability, and gets safe, effective tools to the people who need them, especially in rural and underserved communities.”

ATA Action's comments draw from the ATA's Policy Principles on Artificial Intelligence.

The organization also recommends:

• Provide clear disclosure that is reasonable and proportionate. FDA labeling should explain in plain language what a device does, who it is for and its known limitations; make clear the user is interacting with AI; and offer a clear path to a human clinician or emergency service, as appropriate.
• Regulate by function and risk, not by technology. Oversight should rest on what a tool does and risk, not on the fact that it uses generative AI. The FDA’s two-axis framework for AI systems that serve multiple clinical functions within a single interaction supports unified, harmonized, and risk-based regulation, with federal leadership to promote national consistency and without general prohibitions or broad pre-authorization schemes.
• Create a voluntary federal-state clinical AI sandbox. As states adopt differing approaches to clinical AI, ATA Action proposed a voluntary federal-state AI sandbox, built on the TEMPO pilot and developed with the Department of Health & Human Services (HHS) and participating states, and work alongside FDA’s existing authorization pathways. It would generate real-world evidence and phased clinician oversight, while FDA exercises conditional, time-limited enforcement discretion. This would give real-world evidence and direct insight into how these tools perform once deployed, and can inform a durable, nationally consistent framework.
• Scale validation to risk. ATA Action supports pairing benchmark testing with clinical confirmation scaled to risk, agrees that a prospective clinical study should not be required in every case, and strongly supports FDA’s openness to accepting more uncertainty at authorization when paired with strong post market monitoring and flexible predetermined change control plans.
• Evaluate performance across patient subgroups. Manufacturers should test devices with people who reflect intended users.
• Weigh the real-world alternative. In evaluating risk, FDA should consider the care a patient would realistically receive without an AI-supported service. In many rural communities and areas with clinician shortages, that alternative is a long wait or no care at all. ATA Action also encourages FDA to keep coordinating with the Centers for Medicare & Medicaid (CMS) to ensure safe and effective devices reach patients who need them.
• Rely on existing privacy and security rules. FDA should rely on established privacy and cybersecurity requirements rather than create new ones specific to generative AI and should protect the proprietary and confidential information developers share with the agency.

Read the ATA's full comment letter here.