AMA seeks liability protections for physicians using clinical AI
The American Medical Association is calling for federal and state laws that would shield physicians when AI products harm patients despite being used as intended. Its policy also calls for liability requirements and disclosures for AI companies.
Key takeaways · 4
- 01
The AMA is seeking state and federal legislation to protect physicians when AI causes harm despite intended use.
- 02
The policy would place responsibility on an organization that mandates a specific AI tool and its vendor if the tool causes harm.
- 03
The AMA wants AI companies to carry adequate liability insurance and disclose product limitations and performance data.
- 04
The policy is not law and does not change malpractice rules in any state.
What the policy would change
The AMA is asking lawmakers to shield physicians from liability when an AI product harms a patient despite being used as intended.[1] It also wants physicians protected when they decline to use an AI tool or reject its recommendation.[3] Under the policy, when an organization requires use of a specific AI product and that tool causes harm, the organization and its vendor should bear liability.[1][2] The AMA’s proposed framework distinguishes the responsibilities of physicians, developers and deployers: physicians for clinical judgment they can meaningfully exercise, developers for what they build, and deployers for where and how AI is used.[4]
Disclosures and safeguards
The AMA is pressing for laws requiring AI companies to carry adequate liability insurance and disclose product limitations and performance data.[1] It also wants companies to avoid contract terms that shift liability to physicians or health systems.[3] For clinical AI used in diagnosis, prescribing and care management, the AMA says safety and efficacy evidence should be transparent and auditable.[2] The policy also says AI should not receive physician licensure, professional titles or certification to practice medicine.[2] These proposals concern both information users need to assess a tool and how responsibility is allocated when its use causes harm.[3][2]
Policy is not yet law
The House of Delegates approved the policy on Nov. 9 at the AMA’s Interim Meeting in Orlando, Florida.[2] The AMA had released a clinical AI liability framework in October, before delegates adopted the policy.[2] The adopted position is advocacy, not legislation, and it does not change malpractice rules in any state.[3] The AMA says more than one-third of respondents selected clear liability frameworks as the regulatory action most likely to increase their trust in AI and likelihood of adopting it.[4]
Why accountability is contested
The AMA says AI systems can misjudge case severity without the attending physician’s knowledge, or make plausible treatment recommendations based on incomplete laboratory results or outdated diagnoses that a physician may be unable to verify.[1] The FDA oversees AI software that meets the definition of a medical device, but not all AI used in health care.[1] According to the AMA, the FDA had authorized more than 1,450 AI-enabled medical devices as of December 2025.[2] Hospitals generally support placing responsibility for monitoring AI performance on developers and vendors, citing the difficulty of identifying flaws in opaque systems.[1] AdvaMed, meanwhile, wants the FDA to avoid post-market monitoring requirements that add burdens regardless of a product’s actual risk.[1]
For health systems and clinicians, the proposal puts responsibility for AI-related harm, disclosure and monitoring squarely on the agenda. Professionals evaluating clinical AI can use the policy’s proposed division of accountability as a prompt to clarify who makes decisions, monitors performance and bears risk in their own arrangements.
Why it matters
AI training for your team
kju.ai is built for teams: daily sessions, team analytics, and rollout support for your organization.
How this developed
13 November 2026
AMA seeks liability protections for physicians using clinical AI
Sources
- Doctors want liability shield from AI mistakesaxios.com
- AMA says AI can’t be licensed as a physician, pushes vendor liabilitybeckershospitalreview.com
- AMA Seeks Legal Protection for Doctors Who Reject AI Advice and Wants Vendors Liable for Faulty Toolsmedicaldaily.com
- Issue Brief: Liability for Clinical AI | AMAama-assn.org
