The American Medical Association (AMA) this week urged federal and state lawmakers to create legal protections for physicians when artificial intelligence (AI) products injure patients, provided the physicians used the tools as intended. The AMA is also pressing for laws that require AI companies to carry adequate liability insurance and to disclose limitations and performance data for their products.
Why this matters
Widespread deployment of increasingly sophisticated AI tools in hospitals and clinics is blurring lines around negligence in health care. As autonomous AI assumes more important clinical tasks with less direct oversight, it becomes harder to determine who should bear responsibility when an error harms a patient.
AMA's position
The AMA’s policy states that manufacturers and vendors — and potentially health systems that mandate use of specific AI products — should be held responsible if required use of an AI system results in harmful errors. The organization argues that patient harms caused by AI should not automatically be routed through medical malpractice claims against treating physicians when the tools were used appropriately.
Practical risks highlighted
The AMA points to scenarios such as triage errors, where AI systems that process clinical histories, vital signs and imaging might misjudge the severity of a case without the attending physician’s knowledge. It also raises the risk that AI could offer plausible treatment recommendations based on incomplete lab results or outdated diagnoses, situations in which a physician may be unable to verify the system’s inputs.
Legal background and precedent
There have already been lawsuits tied to algorithm-driven medical decisions, including a wrongful-death case in Alabama involving a screening tool for heart abnormalities. In that case the software developer was not held liable because its licensing agreement left final medical decisions to physicians.
Martin Hilbert, a professor at the University of California, Davis who studies digital information and algorithms, noted that courts are increasingly concluding that companies that design, train and deploy AI systems bear responsibility for what those systems do. He emphasized that accountability depends on factors such as who shares responsibility when multiple parties build a system, how to prove fault within an opaque model, and how to address damages that surface only after deployment. "Different parties are responsible for different cases, and you can't just hold the end user responsible," he said.
Regulators and industry responses
The Food and Drug Administration (FDA) is considering whether a new regulatory framework is needed for AI-enabled medical devices. The agency’s jurisdiction, however, covers only software that meets the definition of a medical device, not all AI used in health care.
Hospitals generally support placing responsibility on developers and vendors for monitoring system performance, citing the "black box" character of many models that can make flaws hard to identify. AdvaMed, the trade group representing device makers, is urging the FDA to continue using its current approaches for evaluating generative AI-enabled medical devices and to avoid post-market monitoring requirements that would impose added burdens irrespective of a product's actual risk.
Takeaway
The AMA’s push highlights growing concerns about accountability as clinical AI tools proliferate. The core issue is how to apportion responsibility among manufacturers, vendors, health systems and individual clinicians to protect patient safety while enabling beneficial AI use.
