AMA Backs Liability Shield for Physicians Declining AI

The American Medical Association's policy seeks to assign responsibility for AI-related harm to vendors, manufacturers, and distributors rather than to physicians.

Updated on Nov. 13, 2026 in Artificial Intelligence

AMA Backs Liability Shield for Physicians Declining AI

Live Poll

Would you trust an AI tool to influence your medical care?

AMA delegates approved a policy on Monday, November 9, at the organization's interim meeting in Orlando, seeking liability protections for physicians who decline to use clinical AI tools or who do not follow their recommendations. The policy assigns responsibility for harm from a defective product to the vendor, manufacturer, or distributor when a physician used the tool as intended and exercised appropriate clinical judgment.

Why it matters

The policy addresses who bears legal responsibility as AI tools enter clinical care, a question AMA President Willie Underwood III framed as a patient-safety matter. The AMA says it plans to pursue legislation at the federal and state levels to establish these protections.

The policy was approved by AMA delegates on November 9, 2026, at the organization's interim meeting in Orlando. The extraction does not provide a vote count, and it does not indicate whether any associated bill has yet been introduced in Congress or any state legislature.

The players

American Medical Association

The American Medical Association is the largest professional association of physicians and medical students in the United States, and its policy positions often shape state and federal health care debates.

Willie Underwood III

Willie Underwood III is the president of the American Medical Association, serving as the organization's public spokesperson on its policy priorities.

The details

The policy calls for legislation requiring AI vendors and manufacturers to carry adequate liability insurance and to disclose known failure rates, product limitations, and performance data. It also opposes contract terms that shift responsibility for AI product failures onto physicians or health systems, and states that AI should not receive physician licensure, professional titles, or certification to practice medicine.

November 9, 2026: AMA delegates approve the policy at the interim meeting in Orlando.

Industry Context

The policy builds on the AMA's earlier framework that allocates responsibility among AI developers, health care organizations, and physicians based on their authority and ability to prevent harm. It moves further by placing liability for defective products on vendors and manufacturers when physicians use tools as intended.

For physicians, the policy would shift the legal risk of declining or overriding an AI recommendation away from individual clinicians if the tool is defective. For patients, it signals that accountability for AI-related harm would be tied to vendors, manufacturers, and the organizations that require specific tools.

The takeaway

Responsibility for AI-related harm in clinical care is becoming a policy question as much as a technical one. Patients and clinicians may want to ask which vendor or organization stands behind a specific AI tool and what performance data it has disclosed.