AMA urges risk-based FDA oversight of generative AI devices

The American Medical Association (AMA) is urging the U.S. Food and Drug Administration (FDA) to modernize oversight of generative AI medical devices while maintaining requirements for safety and effectiveness. In feedback to the agency, the AMA recommended evaluating performance in real-world clinical settings, strengthening monitoring after deployment and involving physicians in defining clinical performance standards.

The proposed approach would tailor oversight to a device’s intended use, potential for harm and performance throughout its lifecycle. The AMA emphasized that regulatory flexibility must not weaken the evidence needed to establish safety and effectiveness. Physician supervision should not substitute for rigorous proof that a device works reliably in its intended clinical setting, and manufacturers must remain accountable for their products.

The AMA offered to work with the FDA and other stakeholders on best practices for clinician adjudication, ongoing performance reviews and identifying clinically consequential AI failures. It supports a least-burdensome, risk-based framework that combines meaningful clinical evidence, appropriate safeguards and continued monitoring as technologies evolve, aiming to protect patients while allowing responsible medical device innovation.