US medical AI replaces nurses as safety whistleblowers emerge
Two major US hospitals are facing backlash after replacing nurses with artificial intelligence and allegedly suppressing safety warnings, raising fresh liability concerns for European healthcare providers considering similar automation.
Two US hospital systems are at the centre of a growing dispute over medical AI, with one accused of using the technology to replace clinical staff and another facing a lawsuit over suppressed safety flaws. The cases highlight a stark divergence between the promised efficiency of healthcare automation and the reality of its deployment.
At Montefiore hospital in New York, a dozen nurses specialising in utilisation review were laid off this month. The New York State Nurses Association claims the workers, including 39-year veteran Marilyn Shuler, were replaced by software that reads patient charts and disputes insurance coverage. Montefiore denies the union's account, calling it inaccurate and insisting the change only affects a nonclinical paperwork programme.
The dispute has triggered a labour conflict, with the union arguing the layoffs violate a contract recently secured through a strike. “It should also concern every practitioner and patient who cares about… the quality of care they receive,” said Shaiju Kalathil, a Montefiore nurse and union representative. The clash underscores a fundamental tension over whether AI is eliminating administrative burdens or skilled medical jobs.
In Minnesota, the concerns centre on patient safety rather than employment. Traci Tamiko Eto, who joined the Mayo Clinic in 2023 to build AI safeguards, has filed a lawsuit alleging she was demoted and fired for raising alarms about an AI assistant called MAYA. The legal filing claims the development team deleted unfavourable test results and overstated the tool's capabilities.
At one stage during its rollout, the suit alleges MAYA had an error rate as high as 67 percent. While the Mayo Clinic states it remains committed to responsible AI with built-in privacy and compliance, Eto's account suggests a system rushed past standard oversight mechanisms. The allegations have not yet been tested in court.
For European health systems and insurers, these US cases offer a warning about the structural risks of clinical automation. Healthcare is an industry where AI benefits are pitched as clearest, yet where failures are least forgivable. The Minnesota allegations illustrate how internal oversight can buckle under commercial pressure to deploy new tools.
The broader economic pattern is already reaching European firms. Employers from Allianz to Thomson Reuters are actively swapping experienced staff for AI models. Research consistently indicates that relying on algorithms can quietly erode the professional skills of remaining staff, a risk that is magnified when the output directly affects patient health.
The core issue in both hospital cases is the distribution of risk. Institutions reap the financial benefits of automation, while patients and practitioners are left to absorb the consequences of any software failures. As European providers evaluate similar cost-cutting tools, the US experience shows that deploying AI in clinical settings is as much a labour and liability issue as a technological one.
