AMA sets out clinical AI governance framework
The AMA outlines four questions on clinical function, evidence and risk, authority and oversight, and accountability for agentic AI.
The American Medical Association published a governance framework on October 13 for AI agents that assume clinical functions. It addresses what those systems may do and who bears responsibility for their actions and outcomes.
The AMA says clinical authority should depend on evidence and continued performance. It plans to revisit authority and oversight as evidence, performance, risk or technology changes.
Governance questions
Four questions will guide its work: the clinical function involved; evidence of capability and the consequences of failure; permitted actions and oversight conditions; and responsibility for risks and outcomes. The AMA will use these questions in discussions with physicians, developers, organizations deploying AI and regulators.
Scope of the announcement
This is the AMA's approach to governance discussions. The announcement does not grant an AI system permission to practice medicine or announce a product approval.
Sources
- [1]
Images: generated with GPT Image 2.5 (illustrative).