Nurses Demand Contract Protections as AI Tools Enter Hospitals

Nurses Demand Contract Protections as AI Tools Enter Hospitals

Healthcare workers are pushing back against the unregulated introduction of artificial intelligence in clinical settings, with nurses unions now securing contract provisions that give them a voice in how AI systems are deployed.

At Munson Medical Center in Traverse City, Michigan, the nurses union recently made AI regulations a central demand in contract negotiations. Union president Laura Nilsson told members the union wants technology that benefits both staff and patients, but emphasized concerns about AI replacing nursing judgment or eliminating jobs.

The push for contractual safeguards comes as healthcare organizations adopt AI tools that perform tasks ranging from appointment documentation to proposing diagnoses based on medical records. This winter, a New York City nurses union representing more than 10,000 members successfully negotiated AI protections into their contract, according to Marketplace, which first reported the Michigan negotiations.

Why it matters

The healthcare industry is deploying AI systems faster than evidence about their safety and effectiveness can accumulate. Nurses — who spend more direct time with patients than any other clinical role — are asserting that technology decisions shouldn't be made without input from frontline workers who understand how tools perform in real-world conditions. Their contract demands represent an early test of whether labor agreements can shape how organizations implement emerging technologies.

Clinical judgment versus automated systems

James Walker, a critical care nurse at Munson who serves on the union bargaining committee, illustrated the limitations of automation with a simple example: if a blood pressure cuff isn't positioned correctly on a patient, an AI system would import the inaccurate reading without verification. A nurse physically present would catch the error before it influenced treatment decisions.

Max Topaz, a nursing professor at Columbia University, said healthcare organizations routinely deploy new technology without considering how it integrates into existing clinical workflows. "These technologies have to be vetted and validated, and we need to know what the error rates are," Topaz said.

Recent research underscores those concerns. A study from Stanford and Harvard researchers examined AI performance in healthcare scenarios and found that even the best-performing models from Anthropic and Google produced 12 to 15 severely harmful errors per 100 cases. The weakest models made mistakes in 40 out of 100 cases.

Hospital perspective

Jenn Standfest, Munson Medical Center's chief nursing officer, acknowledged AI's potential to reduce clerical burdens that contribute to burnout. She described the technology's growth in healthcare as potentially positive and said the hospital wants "to get it right."

Under the new agreement, nurses in northern Michigan will have input on AI implementation decisions if the hospital purchases new software systems.

These details were first reported by Marketplace.