Skip to content
AI.info

The Pulse

AMA and state medical boards say AI oversight and liability frameworks need modernizing

A joint report lists establishing how responsibility for AI-related patient harm should be assigned among its priorities. A medical-board workgroup plans to seek public comment on draft guidance in December.

AMA and state medical boards say AI oversight and liability frameworks need modernizing

Establishing how responsibility for AI-related patient harm should be assigned across physicians, developers and other parties is a priority in a new report from the American Medical Association and the Federation of State Medical Boards. The groups say existing oversight and liability systems need to be modernized as AI takes on clinical functions.

The report, released Nov. 12, summarizes a Sept. 14 forum in Washington, D.C., on accountability in medicine. More than 200 physicians, medical-board representatives, patients, policymakers and technology innovators attended, according to the AMA.

Drawing the clinical boundary

The report says existing regulatory, licensing and liability frameworks were not designed for AI, while patients and medical professionals are adopting the technology faster than the health system is adapting. It identifies a question for oversight: when does AI serve as a clinical tool, and when does it constitute the practice of medicine?

The forum focused on autonomous and semiautonomous tools and their implications for clinical responsibility, medical-board oversight and the physician-patient relationship. Participants brought differing views, including on how quickly AI will change care teams and physicians’ roles.

Humayun J. Chaudhry, the federation’s president and CEO, said patients need clarity as AI begins performing more functions previously reserved for licensed physicians. “Patients deserve clarity about who is responsible for their care and who is accountable when something goes wrong,” he said.

Who would be responsible?

Among the report’s priorities is clarification of the boundary between clinical technology and medical practice. It also calls for clearer roles for state medical boards, federal regulators, developers, physicians and other healthcare stakeholders when AI-related harm occurs.

An AMA issue brief on liability for clinical AI, published Oct. 29, sets out an approach under which responsibility would reflect each party’s authority, access to information and ability to prevent or mitigate harm. Under that approach, physicians would be responsible for clinical judgment they can meaningfully exercise, developers for the systems and risks within their control, and deploying organizations for where and how AI is used.

Patients should retain meaningful avenues for accountability and appropriate remedies when harm occurs, the AMA says.

Training and next steps

The joint report also identifies a need for best practices covering clinical use, evidence for adoption, deployment and performance monitoring. Another priority is adapting medical education, clinical training, continuing professional development and team-based care to AI use.

The groups want care delivery to remain transparent and equitable while preserving medical ethics and the physician-patient relationship. The AMA Center for Digital Health and AI and the federation’s AI workgroup will continue working with stakeholders on the priorities.

The federation’s workgroup plans to release draft guidance and recommendations for public comment in December 2026. A formal adoption vote by its House of Delegates is scheduled for April 2027.

Sources

Related stories

Explore

More articles