Nurses' unions win AI guardrails in Michigan, NYC contracts
TL;DR
- Nurses' unions in Michigan and New York City are winning contract language that gives them oversight of how AI tools are deployed in their hospitals.
- A New York City nurses' union with more than 10,000 members secured AI protections in its contract this past winter.
- Traverse City Munson Nurses Association president Laura Nilsson says the union does not want technology to replace nursing judgment.
There is a quiet but interesting thing happening in how clinical AI is actually getting governed inside US hospitals right now, and it is not happening at the federal level. It is happening at the bargaining table.
Marketplace reported that nurses' unions in Michigan and New York City are winning contract language that gives them oversight of how AI tools are deployed in their hospitals. In Traverse City, Laura Nilsson, president of the Traverse City Munson Nurses Association, stood at the front of a room listing the union's demands, which included regulations around artificial intelligence alongside pay and staffing. "We understand technology is coming. We generally want it to be good for us and our patients, but right now, there [are] a lot of unknowns. What we don't want technology to do is replace nursing judgment," she told Marketplace. This past winter, a nurses' union in New York City with more than 10,000 members got similar AI protections in its contract.
The concern that comes through in the reporting is not abstract. AI tools in hospitals now do everything from taking notes on appointments to proposing real diagnoses based on medical records, and nurses are the ones standing at the bedside when the model's read disagrees with what they can see. A critical care nurse named James Walker gave Marketplace a small but telling example: if the blood pressure cuff is not on the patient appropriately, the bad reading still flows into the system, and a downstream decision can ride on it. A bedside nurse can catch that. A workflow that routes around the bedside nurse cannot.
The honest caveat is that the reporting is light on specifics about what the contract language actually requires. "Oversight" can mean review rights, veto rights, or just a seat in a meeting, and the article does not say which. It also does not name the AI vendors or the specific tools at issue. Take this as a direction-of-travel story rather than a settled regulatory picture.
The forward-looking thing to watch is whether the Michigan and New York language gets copied. If it does, the practical floor for clinical AI deployment in a lot of American hospitals could end up being set in nursing contracts rather than in agency guidance. That changes who the vendors actually have to convince.
Shared on Bluesky by 2 AI experts
-
“In a recent study looking at the use of AI in healthcare, researchers from Stanford and Harvard found even top-performing AI models from Anthropic and Google produced 12 to 15 severely harmful errors per 100 cases. The …
View on Bluesky →
Originally reported by marketplace.org
Read the original article →Original headline: Why nurses' unions are fighting for AI guardrails