As generative AI moves toward performing increasingly consequential clinical functions, the American Medical Association is pushing the FDA to address a foundational question: who is responsible for keeping these tools safe once they are in use? The AMA's call reflects growing concern that current regulatory assumptions may not fit AI systems that can change or behave unpredictably after deployment.
The AMA is specifically recommending a risk-based oversight approach, meaning regulatory scrutiny would be scaled to the level of patient harm a given tool could cause. This is intended to guard against unsafe uses without smothering development. The group frames the request as a balance between patient protection and responsible innovation.
By raising the accountability question now, the AMA is signalling that the clinical use of generative AI has moved beyond hypothetical. Regulators need a clear framework for monitoring and ensuring safety over time, not just at the point of approval.