October 29, 2026 by matray

The American Medical Association has proposed a framework for determining liability when artificial intelligence used in clinical care causes patient harm, calling for responsibility to be allocated among physicians, technology developers and healthcare organizations according to their ability to prevent or mitigate risks.

The framework, announced Oct. 29 by the AMA's Center for Digital Health and AI, addresses concerns that existing medical liability standards were not designed for technologies in which multiple parties exercise varying degrees of control over clinical decisions and patient outcomes.

The AMA warned that physicians could face liability for AI-related risks they cannot reasonably identify or prevent, while patients harmed by the technology could encounter difficulties obtaining legal remedies.

“Patients must have meaningful avenues for recourse when clinical AI causes harm, and physicians must be able to use these technologies with confidence that responsibility reflects what they can reasonably know and control,” said AMA President Willie Underwood III, MD.

The proposed approach would establish distinct but complementary responsibilities for the parties involved in developing, deploying and using clinical AI systems.

Under the framework, developers and manufacturers would be responsible for demonstrating and maintaining the safety and effectiveness of their products, including addressing risks within their control.

Hospitals, health systems and other healthcare organizations would be accountable for selecting, implementing, governing and monitoring AI technologies in clinical settings.

Physicians, meanwhile, would remain responsible for exercising clinical judgment, but their liability would be tied to the degree of oversight they can meaningfully provide, considering an AI system's intended use, known limitations and the information available when making treatment decisions.

The AMA also emphasized that patients must retain meaningful opportunities to pursue remedies when injuries result from defective AI products or other failures for which a party may be held legally responsible.

The framework recognizes that multiple parties could share liability for an AI-related patient injury, rather than placing responsibility exclusively on the treating physician or the technology developer.

A central principle is that requiring physician oversight of AI does not automatically relieve developers or healthcare organizations of their own responsibilities. The AMA said clinical review cannot compensate for inadequate evidence supporting an AI system's safety, deficiencies in its design or circumstances that prevent physicians from exercising meaningful oversight.

“As AI assumes more clinical functions, developers, healthcare organizations and physicians each have important responsibilities,” Underwood said. “We need clear expectations that protect patients, support sound clinical judgment, and ensure accountability remains with those best positioned to prevent harm.”

The proposal reflects growing concern over how traditional medical malpractice and product liability principles should apply as AI technologies become more deeply integrated into patient care.

The AMA's framework outlines recommended principles for assigning responsibility rather than establishing binding legal standards. Its central recommendation is that liability should follow each party's authority over an AI system and its practical ability to identify, prevent or mitigate harm.