The American Medical Association wants the makers of clinical AI, not doctors, to carry the liability when an AI tool gets it wrong. New policy adopted by the AMA House of Delegates at its 2026 Interim Meeting directs the association to advocate legislation “prohibiting contractual provisions, including end-user license agreements and vendor contracts, that attempt to shift liability for AI tool failures from the manufacturer to the physician or health system end user”, the AMA wrote on 10 November.
Axios headlined its 13 November report “Doctors want liability shield from AI mistakes”. The push comes as AI use spreads: 81 percent of physicians use AI in their practices, up from 38 percent in 2023, according to the AMA’s 2026 physician survey. That is 2.1 times the 2023 rate, by our arithmetic.
Business Pill · Vendor Guarantees Explained
An evergreen Business Engineer explainer on a related question: who carries the risk after the sale. It teaches the concept with a chip-financing example and does not discuss the AMA or medical AI.
The key insight: As we read the policy, the AMA is asking lawmakers to move the cost of a wrong AI output from the doctor’s malpractice cover to the vendor’s balance sheet, and to stop vendors moving it back through contract terms.
The policy directs the AMA to advocate federal and state legislation establishing that when a clinical AI product “produces an erroneous output and a physician has used the product as intended while exercising appropriate clinical judgment”, liability for the resulting harm “should rest with the AI vendor, manufacturer, or distributor, not a medical malpractice claim against the physician”.
The delegates also asked that AI vendors be held to the same product liability standards as makers of other medical devices, that they carry product liability insurance, with that obligation tied to “a condition of FDA marketing authorization”, and that they disclose “known failure rates, limitations, and performance data” for their clinical AI products. That makes six asks on vendor liability, by our count.
The sixth covers mandated tools. When a health system, hospital, payer or other entity requires a specific AI tool and that tool errs, the AMA says “the mandating entity and the AI vendor bear the liability”, not the physician who was required to use it.
Four American Medical Association steps on clinical AI in 2026, from its own releases: the 12 March physician survey, the 14 September AMA-FSMB forum, the 29 October liability framework and the 10 November House of Delegates policy. Six asks is our count.
The Other Side of the Shield
The same policy covers doctors who decline AI. The AMA will advocate legislation that includes protections “against liability for choosing not to use an AI tool or follow an AI recommendation”, according to the 10 November article. It also says AI “cannot be granted physician licensure, professional titles or certification to practice medicine”.
AMA President Willie Underwood III said: “Physicians also need the authority, oversight and liability protections to make independent decisions when AI is part of the care team.” The AMA cites more than 1,450 AI-enabled medical devices authorized by the FDA as of December 2025.
How the AMA Got Here
On 29 October the AMA released a framework for clinical AI liability that splits responsibility among developers, healthcare organizations, physicians and patients, with “accountability aligned with each party’s ability to prevent harm”. On 12 November the AMA and the Federation of State Medical Boards issued a report from a 14 September forum, which found that existing liability frameworks “were not designed for AI and need to be modernized”.
According to the forum’s report, some participants said vendor contracts “often shift liability onto physicians or the health care organization”, while others said health care organizations “may have leverage to negotiate liability allocation”. The FSMB plans draft guidance for public comment in December 2026, ahead of a vote by its House of Delegates in April 2027. In the AMA’s March survey, clear liability frameworks ranked highest among the regulatory actions physicians said would build trust in AI tools.
As we read the six asks, the contract clause ban is the commercial lever: under the policy, liability shifts in end-user license agreements and vendor contracts would be prohibited, not negotiated.
As we read the survey, the timing matters: 81 percent of physicians already use AI, 2.1 times the 2023 rate by our arithmetic, and physicians ranked clear liability frameworks highest among the regulatory actions that would build trust.
Willie Underwood III, AMA President, 10 November 2026
“AI can help physicians analyze information, identify patterns and support clinical decision-making, but patients deserve to know who is responsible for their care”
Three Implications
FOR CLINICAL AI VENDORS The policy asks for product liability insurance tied to FDA marketing authorization and disclosure of known failure rates.
FOR HEALTH SYSTEMS AND PAYERS The policy says an entity that mandates a specific AI tool would share liability with the vendor for its errors.
FOR PHYSICIANS The policy asks for protection from liability for choosing not to use an AI tool or follow an AI recommendation.
The framework puts the shift in one line: “SaaS is charged for access to tools. AGaaS charges for the execution of outcomes.”
As we read it, the AMA policy pushes clinical AI vendors toward the second model on the liability side: a vendor that answers for an erroneous output is accountable for the outcome, not just the access. The framework does not discuss the AMA or healthcare liability.
AMA Physician Survey on Augmented Intelligence: 81 percent of physicians used AI in their practices in 2026, up from 38 percent in 2023 (AMA release, 12 March 2026). 2.1 times is our arithmetic.
What Is Not Established
The policy is a lobbying position, not law. The AMA article names no bill, sponsor or vendor, and we found no bill text tied to it. We read only the headline of the Axios report. The article gives no vendor response.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
The AMA’s delegates want liability for erroneous clinical AI output to rest with the vendor when a doctor uses the tool as intended, and want contract terms that shift that liability to doctors prohibited by law. The ask lands with 81 percent of physicians already using AI, 2.1 times the 2023 rate by our arithmetic.
A note on sourcing. The policy comes from the American Medical Association’s article of 10 November 2026 on policies adopted at its 2026 Interim Meeting; the framework, report and survey figures come from AMA press releases of 29 October, 12 November and 12 March 2026, and the AMA-FSMB proceedings report. We read only the headline of the Axios report. The 2.1 times figure and the count of six asks are ours. We did not contact the AMA for this piece.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
