Loading Curie Briefs...
Getting the latest healthcare news for you
Getting the latest healthcare news for you
The FDA and states are tightening the rules around medical AI, but the landscape is still a patchwork. The FDA is exploring competency-based testing for AI devices and seeking public feedback on a new risk framework, while over a dozen states have already passed their own AI-in-healthcare laws. Meanwhile, physicians may still be on the hook when AI gets something wrong.
The FDA and state regulators are actively reshaping the rules around medical AI — and physicians are right in the middle of it. The FDA's Center for Devices and Radiological Health recently released a discussion paper outlining how generative AI-enabled medical devices could be evaluated, proposing a competency-based testing framework inspired by how human clinicians are credentialed. Risk would be assessed based on what the device does and the consequences of a wrong output — think: a tool nudging a patient to change their insulin dose carries far more risk than one displaying a cardiovascular risk score.
But here's the catch: not all AI tools used in clinical settings are FDA-regulated. Many clinical decision-support tools and general-purpose AI (like ChatGPT) fall outside FDA oversight entirely. Even FDA-cleared devices often went through the less rigorous 510(k) pathway. Experts warn physicians not to assume any AI tool has been properly vetted.
Key Takeaways:
Why it matters: With more than 80% of physicians now using AI professionally, the regulatory gap between what's deployed and what's been rigorously tested is a real patient safety concern. Doctors need to know what they're using — and what liability they may be assuming.
