Nature Health maps 31 AI healthcare lawsuits, shifts blame frame

Two Georgia Tech researchers went through 31 US lawsuits in which an AI tool was implicated in patient harm, and the pattern they describe in Nature Health is that the physician is rarely the whole story.

"Our findings reflect how patients' medical care relies on a complex web of stakeholders—physicians, state health departments, health insurers and care facilities, among others," write Gennie Mansi and Mark Riedl, drawing on three litigation trackers: DAIL, the Georgetown O'Neill Institute Health Litigation Tracker, and AIAAIC.

The named systems in the analysis include nH Predict, the Babylon Health Triage System, and the Oregon DHS Safety Screening Tool. These sit at insurers, triage vendors and state agencies rather than at the exam room. Patients, the authors write, "have had no option but to seek legal recourse for harms."

The remedy the paper offers is a reframe. "We shift the frame from physician-centred to patient-centred accountability approaches by describing how lawyers and technologists need to recognize and address where AI harms happen," Mansi and Riedl write. Their two paths: change liability structures so they reflect the many stakeholders shaping how AI reaches a patient, and design tools that let patients' lawyers "provide critical legal expertise and practically support recourse for patients."

The analysis does not publish per-case outcomes or the breakdown of defendants by type across the 31 cases.