Nursing AI Watch: What’s Really Happening With AI in Nursing

Introducing Nursing AI Watch — Nurse.org’s ongoing tracker of how artificial intelligence is actually being used in nursing. We follow the tools, the systems deploying them, the safety questions, the union contracts, and the oversight gap, and we update it as the picture changes.

AI is already in nursing — in the charting, the staffing, the contracts. This is the grounded version, built on what we can actually verify: what’s deployed across 100+ health systems, what’s helping real nurses with their workloads, where it’s going wrong, and what nurses are doing about it.

Want to see more Nurse.org articles in your Google results? Add us as a preferred source.

“AI in nursing” gets used as one big phrase, but it covers four very different things — and the differences matter for your day.

• Ambient documentation is the fastest-growing category. These are AI “scribes” that listen to a patient encounter and draft the note for you. Think of dictation that writes itself, then waits for you to review and sign.
• Clinical support tools can include everything from plug-ins and software integration that prompt care plans, education, safety, or risk detection. For instance, some AI flags a risk or suggests a next step — a sepsis alert, a fall-risk score, a deterioration warning pushed to your screen.
• Virtual nursing uses remote nurses plus AI tools to handle admissions, discharges, and monitoring from a screen in the hospital room, easing some load on the bedside team.
• Administrative AI sits behind the scenes — scheduling, staffing, billing — and doesn’t interact with patients directly.

Danielle K. Miller, DNP, RN, a healthcare consultant and nurse executive from Chicago, Illinois, advises healthcare organizations on AI integration into clinical workflows.

“From what I’m seeing across the industry, healthcare organizations are implementing AI in several ways, including ambient clinical documentation, clinical decision support, predictive analytics for patient deterioration and readmission risk, administrative workflow automation, and patient communication tools,” Miller tells Nurse.org.

The throughline: in almost every case, AI is drafting or suggesting, and a licensed nurse is still accountable for what happens next — which is why understanding how AI and data actually work in practice matters before it lands on your unit.

Widely — and faster than most reporting suggests. Nursing AI Watch has identified 127 distinct AI deployments across 106 health systems.

One important caveat up front: a deployment means we found public evidence that a health system has put an AI tool into use somewhere in its operations. It does not necessarily mean the tool is live system-wide. In our dataset, most deployments are enterprise- or system-scale rollouts, but others are department-level, single-facility, or active pilots. We track the scope we can verify and say so when it’s narrower than the headline.

The more striking finding is concentration. At least 68 of the deployments we’ve documented — more than half — trace to just two vendors. Because this reflects what’s in the public record rather than every contract ever signed, the real concentration may be higher, not lower. 

That concentration shapes everything downstream: when two companies supply most of the AI touching nursing workflows, their design choices, error patterns, and contract terms ripple across the whole field.

Read more about this trend: The Companies Behind Nursing’s AI: The Vendor Landscape

Adoption stories tend to skip this part. The evidence that AI makes mistakes at the bedside is real, documented, and growing.

The patient-safety organization ECRI named AI the #1 health technology hazard on its 2025 list. Peer-reviewed work in JMIR examining ambient AI scribes found that roughly 70% of AI-generated notes contained at least one error — omissions, additions, or details that were never said. And researchers at Columbia Nursing have warned that AI scribes can under-document the concerns of patients with non-standard accents, limited English proficiency, or from marginalized communities — an equity gap nurses see firsthand.

We’ve covered the bedside reality of this — from racial bias in AI charting to a reported case in which an AI system advised loading a dialysis patient with fluids before a nurse caught it, to reports of AI chatbots inventing a disease that doesn’t exist. The lesson isn’t “AI is useless” — it’s that the human review step is doing real safety work, and it can’t be rushed.

Read more about this trend: When AI Gets It Wrong at the Bedside: Errors, Bias, and Potential Nurse Liability

Our research shows most AI decisions are largely being made without nurse input. The nurses who are getting a say have done so mostly through hard-fought contracts, not necessarily because they’re being consulted up front.

In a National Nurses United survey of more than 2,300 RNs, 60% said they did not trust their employer to put patient safety first when implementing AI, and among nurses whose employer uses acuity algorithms, 69% said the results didn’t match their own assessment.

Where nurses are gaining ground is at the bargaining table. Nursing AI Watch tracks seven health systems with ratified union contract language addressing AI, covering tens of thousands of RNs — and in most cases it took organizing to get there.

The pattern is a two-wave, two-state story. In New York, NYSNA contracts ratified in early 2026 at systems including Mount Sinai, Montefiore, and NewYork-Presbyterian included explicit technology protections — language that AI cannot be used to replace nurses, to discipline them, or to drive staffing decisions.

In California, nurses with NNU/CNA secured AI language across the University of California’s medical centers, and Sutter Davis Hospital’s first-ever union contract guarantees nurses a say in how new technology like AI is implemented. Note: the Sutter language is specific to Sutter Davis Hospital, not all of Sutter Health. For more on how nurses are building trust and training around AI tools, see our deeper coverage.

Read more about this trend: When AI Lands in the Contract: Nurses Bargaining Over AI

>>Listen to The Latest Nurse News Podcast

There are significant drawbacks to excluding nurses from the base level of AI development and integration, Miller warns. 

“The organizations that are seeing the greatest success are those that involve nurses from the beginning of the design and implementation process,” she points out. “Nurses understand workflow, patient safety, and the realities of care delivery, making their input essential to developing tools that are actually useful.”

Miller adds that her experience has shown her that AI has “tremendous potential” when implemented correctly—meaning with strong clinical oversight that starts at the base level. 

Adoption is outrunning oversight, and the people closest to patients are rarely the ones governing it. Our findings show that of the 106 systems deploying AI, at least 26 have a named AI governance body — but of the governance structures we could document, we could only confirm six named nurse leaders publicly shaping AI governance. Just 16 of 106 publish a patient-facing AI disclosure policy, and only 12 of 106 have a confirmed Chief Nursing Informatics Officer. 

These are floor numbers: where a system doesn’t disclose a governance body or a CNIO, we count it as unconfirmed rather than assume one exists.

So AI is drafting nursing notes and flagging nursing risks, while the nurse perspective is largely absent from the committees deciding how that AI is validated and deployed. That’s not a hypothetical risk — it’s the current structure, as far as the public record shows. The ANA’s AI guardrails for nursing lay out what nurse-led oversight should look like.

Miller again cautions about the lack of human oversight in AI nursing implementations, let alone the lack of a human nurse in that oversight.

“When thoughtfully implemented, AI can reduce documentation burden, surface clinically relevant information more quickly, identify patterns that might otherwise be missed, and allow clinicians to spend more meaningful time with patients,” she says. “However, AI is only as good as the data it is trained on and the governance surrounding its use. Human oversight remains essential.”

Read more about this trend: Adoption Is Outrunning Oversight: Nursing’s AI Governance Gap

The statehouses are moving faster than Washington. Nursing AI Watch tracks 67 AI-related measures across 39 jurisdictions (38 states plus the federal government), 18 of them already enacted.

A handful touch nursing directly. California’s AB 489 and Nevada’s AB 406 are enacted laws addressing AI and clinical scope, and Arizona’s HB 4080 has been introduced. A growing cluster of bills targets AI “impersonating” a licensed provider or giving medical advice without disclosure.

The federal picture, by contrast, is thinner and notably quiet on nurses — seven federal measures touch health AI, but none are nursing-specific — a gap we covered when the White House AI policy left out nurses. That vacuum is part of why the action has shifted to state legislatures, including a New York bill to limit AI medical advice, and to union contracts.

Read more about this trend: The Statehouse AI Tracker: AI Laws Affecting Nursing by State

This is the question that drives the most fear, so it deserves the most honesty.

Of the WARN-notice layoff filings Nursing AI Watch has reviewed, AI correlation is unconfirmed; where roles have been cut, the stated reasons have usually been financial pressure and outsourcing, not automation. The clearest exception on our radar: at Montefiore, NYSNA says 12 utilization-review nurses had duties shifted to an AI system, and the union has filed a class-action grievance — a claim we attribute to NYSNA and are still verifying. We’re not going to declare AI is replacing nurses until the evidence supports it.

What we can say is that the mix of AI capability, hospital budget strain, and virtual-nursing models is exactly the kind of pressure worth tracking. Nursing AI Watch is monitoring WARN filings, and a dedicated workforce analysis will follow once the data can support a defensible conclusion. In the meantime, we’ve looked at which nursing jobs are most AI-resistant and why even Google DeepMind’s CEO says AI won’t replace nurses.

AI is already woven into nursing work, drafting the notes, scoring the risks, and guiding clinical decisions. The honest picture isn’t hype or doom; it’s a fast-moving rollout that’s running ahead of the safeguards, the oversight, and the nurse voice that should be shaping it.

The tools will keep developing and spreading. The open question is whether nurses get a real seat at the table while it happens, and that’s exactly what Nursing AI Watch will keep tracking. As Miller succinctly sums up:

“I don’t believe AI should or could ever replace nursing judgment. Instead, it should augment it. “

Related Nursing AI Watch Analysis:

🤔 Are you seeing AI show up in your unit yet — in the charting, the alerts, the staffing? What’s working, and what worries you? Tell us in the comments below.

About the data: Nurse.org’s Nursing AI Watch is Nurse.org’s ongoing tracking project covering AI in nursing across 106 large health systems. Every figure in this article comes from a structured dataset that we source-verify and quality-check before publishing, and refresh each reporting cycle. We count only what we can document from the public record — company announcements, peer-reviewed research, union and regulatory filings, and official disclosures — which means our numbers are conservative floors, not exhaustive totals. When we can’t confirm something, we mark it unconfirmed rather than assume it. When the evidence supports only a narrow claim, we scope the claim to match.

Nurse.org Analysis