Georgia Tech: 31 AI healthcare suits push liability rethink

Thirty-one US legal cases so far show an AI tool tangled up in patient care, and a new Nature Health paper from Gennie Mansi and Mark Riedl of Georgia Tech works through them to argue that the standard reflex, pinning the outcome on the physician, misses most of what actually went wrong.

"The medical, legal and technology communities often centre physicians' role in preventing and mitigating harms from medical artificial intelligence (AI) tools," the authors write in the opening line of their abstract. Their case for widening the frame is straightforward: patient care rides on "a complex web of stakeholders," spanning physicians, state health departments, health insurers and care facilities, and AI tools now sit inside that web at multiple points.

The consequence, in the paper's blunter phrasing: "patients have had no option but to seek legal recourse for harms."

Mansi and Riedl propose two moves. First, restructure liability so it reflects the range of actors shaping how an AI tool touches any given decision. Second, design tools that help "advocates, such as patients' lawyers," do the forensic legwork of building a case, rather than aiming yet more decision-support at the clinician.

The abstract stops short of naming the 31 cases or the specific doctrines the authors want revised. The paper, published 21 September 2026 in Nature Health, reads as a call for the health-AI conversation to broaden the list of defendants beyond the doctor in the room.