The FDA is drafting a first-ever policy roadmap for regulating medical devices powered by generative AI, according to STAT News reporter Mario Aguilar. “Our goal is formal policy guidance,” said Rick Abramson, director of the FDA’s Digital Health Center of Excellence, describing the discussion paper the agency recently released.
A credentialing model for algorithms
The FDA is weighing a “competency-based approach” to evaluating these tools, modeled on how doctors get credentialed. The framework would cover products like AI medical chatbots, prescription-renewal tools, and diagnostic aids such as imaging or treatment-recommendation software. “You can’t regulate generative AI the way you regulate pills and cardiac pacemakers,” said David Blumenthal, a professor at the Harvard T.H. Chan School of Public Health. The plan pairs benchmarking with ongoing postmarket monitoring, since no lab test can anticipate every diagnosis an AI model might someday face.
The document isn’t a rule — it’s an invitation for public comment, so real requirements are still months away. Until then, patients and clinicians are left navigating AI tools without a clear rulebook for what counts as safe.
Related on the SLP
|
Related on the SLP
AI medical advice: Not ready for prime time
|
