Kaiser Permanente advice nurses who triage patients by phone say artificial intelligence is being used to grade their empathy, score their tone of voice, and cap their calls at 15 minutes — pressure they describe as forcing them to withhold compassion from suicidal, terminally ill, and newly diagnosed patients. Seven current and former nurses told CalMatters that the surveillance is now a central issue in contract talks covering 25,000 California Nurses Association members, including 1,000 call-centre nurses.
🔍 THE BOTTOM LINE
Kaiser is the largest private employer in California and provides healthcare to more than 12 million Americans. What it does with AI-managed nurses sets the precedent for every hospital system watching from the sidelines. A nurse who stopped herself from comforting a terminal cancer patient because an AI might dock her score is not a hypothetical — it is an on-the-record account from the people who answer the phone when you call your health plan in a crisis. California lawmakers have tried and failed three years running to pass a bill requiring employers to tell workers when an automated system is grading them. The contract Kaiser negotiates this summer may do more than the legislature has.
A 15-Minute Cap and a Suicidal Caller
The most detailed account in the CalMatters report comes from Raquel Alvarez Sanchez, a Kaiser advice nurse in Vallejo and a union steward. In 2025 she took a call from a suicidal man that lasted more than an hour because she had to stay on the line until police arrived. She knew the call would throw off her average handle time for weeks and likely trigger a meeting with management. She stayed on anyway.
Another nurse, speaking anonymously for fear of retaliation, described a call with an elderly woman who had just received a terminal cancer diagnosis. The nurse initially thought the patient was suicidal, then realised she was in shock and needed someone to talk to. The nurse wanted to offer comfort. She stopped herself, because an extended call would hurt her monthly performance score and could lead to a reprimand. “I had to ask myself: Am I going to get disciplined for going off script or saying more than what is necessary?” she told CalMatters.
The nurses say calls involving interpreters routinely take 30 minutes or more — a problem in a state where roughly four in ten residents speak a language other than English at home and half of those do not speak English well. Nurses are instructed to stick to a script and give no more than two to three pieces of advice per call, according to Charlotte Capulong, a 22-year call-centre veteran who helped organise the petition against the AI tone tool. “You aren’t calling Comcast. We’re dealing with life here,” she said.
The Empathy AI That Graded Nurses Wrong
In summer 2024 Kaiser began testing an AI tool that attempted to assess empathy and tone in the voices of both nurses and patients, according to nurses who spoke with CalMatters. Nurses circulated a petition — “Trust nurses, not AI” — demanding transparency, patient privacy, and the right to exercise professional judgement. The AI tests ended in November 2024, but union reps were told managers could bring the program back.
One nurse said the AI “did not understand our job and would grade us wrong all the time.” Another said the existing surveillance was already harassing, “and that was intensified when they said we’re going to use AI to evaluate our calls and grade us.” A Kaiser spokesperson declined to answer questions about the empathy AI, about whether patients were informed their voices were being analysed, or about which systems are currently in use. The company did say it does not use “Average Handle Time” to assess performance — a claim nurses say contradicts their lived experience of being called into meetings over long calls.
What Kaiser Is Actually Doing With AI
The call-centre fight is one corner of a much larger AI deployment. According to CalMatters, Kaiser is already using or testing AI to: identify hospital patients at risk of adverse events from electronic health record data; run a system called Preventus that helps decide when to discharge patients; give doctors and therapists the Abridge tool to record visits and transcribe speech to text; and pilot remote AI monitoring of patients who need extra care at Bay Area facilities.
Therapists at Kaiser are separately bargaining over the use of therapy session transcripts to train AI models — a use case Kaiser has repeatedly refused to rule out, according to KQED. 2,400 mental health workers in Northern California are in those negotiations now.
The California Law That Keeps Failing
The nurses’ complaints are not happening in a regulatory vacuum — they are happening in spite of one. California lawmakers have tried three years running to pass Assembly Bill 1018, which would require employers to inform workers before using automated systems to evaluate, promote, or discipline them. It failed again in 2025, facing opposition from Kaiser Permanente itself. A companion bill, Senate Bill 7, reached the governor’s desk and was vetoed by Gavin Newsom.
This year lawmakers reintroduced the framework as Senate Bill 947, with a separate bill — AB 2575 — that would protect doctors and nurses from retaliation when they override an automated recommendation, and require healthcare employers to give staff an annual inventory of every AI system in use. Kaiser declined to share such an inventory with CalMatters. The California Labor Federation is backing roughly half a dozen AI-in-the-workplace bills in total, framing AI regulation as a central issue for the next presidential cycle.
The National Pattern
The Kaiser fight is the sharpest edge of a broader trend. A 2024 National Nurses United survey of more than 2,000 nurses found half worked for employers using algorithmic systems to analyse health records — predicting patient fragility, care hours needed, discharge timing. Two-thirds said their own clinical judgement had at some point disagreed with the computer’s prediction. Six in ten did not trust their employer to prioritise patient safety when deploying AI.
Kaiser nurses went on strike against AI for one day in March and picketed against AI last autumn. The CNA is now bargaining for 25,000 nurses. Whatever it wins on AI — notification rights, override protections, an inventory requirement, a ban on empathy scoring — will be read closely by every hospital system in the country.
NZ Angle
New Zealand has no equivalent of AB 1018 or SB 947. The AI Workplace Regulation Gap report we covered in May found NZ’s framework makes Australia’s look strict, and Australia’s is the one that keeps failing to pass. District Health Boards here have been rolling out AI triage and documentation tools on the same “efficiency” logic Kaiser used — and NZ nurses, like their California counterparts, have no statutory right to be told when an automated system is scoring their work. If a Kiwi nurse were to describe the same experience the Kaiser nurses described to CalMatters — withholding empathy from a suicidal caller because an AI might dock her score — there is no California-style bill moving through Parliament to catch it. The California Nurses Association’s bargain this summer is the closest thing to a template English-speaking healthcare unions have. New Zealand’s broader workplace surveillance gap is the same gap, one jurisdiction further from being closed.
❓ FAQ
Does Kaiser admit to using AI to grade nurses’ empathy? Kaiser declined to confirm or deny the specifics. Nurses say the empathy-and-tone AI was tested from summer 2024 and paused in November 2024, with managers reserving the right to bring it back. Kaiser says it uses AI “responsibly, with human oversight” but would not share an inventory of systems in use.
Is there any evidence that AI call-time caps hurt patients? There is no direct causal study. A 2024 public records request by CalMatters found no patient complaints against Kaiser specifically tied to call times. Nurses insist the risk is real and point to the structural pressure: a nurse who cuts a call short to protect a performance score is a nurse who may miss a cue. Consumer Watchdog patient advocate Michele Ramos said the time pressures fit Kaiser’s broader pattern of “managing dollars over managing care.”
Could this happen in New Zealand? Yes, and there is no statute preventing it. NZ has no equivalent of California’s proposed AB 2575 (override protection) or AB 1018 (notification requirement). DHBs are deploying AI triage and documentation tools on the same efficiency rationale Kaiser used. The NZ Nurses Organisation has not yet bargained an AI-specific clause equivalent to what the CNA is pursuing this summer.
What is the California Nurses Association actually asking for? CNA reps declined to discuss specific bargaining provisions ahead of talks. Public positions include notification of new AI deployments, the right to professional override of automated recommendations, transparency on what systems are in use, and — implicitly — an end to empathy-and-tone scoring of nurse-patient calls.
Why does a US healthcare labour dispute matter to a NZ reader? Two reasons. First, Kaiser is the largest non-governmental healthcare system in the US and the precedent it sets on AI-managed clinical staff migrates fast. Second, the regulatory gap the nurses are exposing — no notification, no inventory, no override protection — is wider in NZ than it is in California. What California fails to pass this year, NZ has not even drafted.
🔍 THE BOTTOM LINE
A nurse who weighs whether comforting a dying patient will hurt her performance score is not a worker with a tool. She is a worker being managed by a tool. Kaiser is betting that the public will not notice the difference as long as the call times come down. The California Nurses Association’s 25,000-member bargain this summer is the first real test of whether a healthcare workforce can write the rules before the AI does. California has tried and failed to legislate those rules three years running. The union contract may be the only enforceable version that actually ships.
📰 Sources
- CalMatters — Kaiser nurses say AI, workplace surveillance are making their jobs and patient care worse
- Associated Press — Kaiser mental health therapists strike over AI
- San Francisco Examiner — California Kaiser nurses picket over AI use
- SF Standard — Kaiser nurses protest AI in San Francisco
- National Nurses United — 2024 survey on AI and patient safety
- KQED — Will AI replace your therapist? Kaiser won’t say no
- CalMatters — AI regulation on hold again (AB 1018 failure)
- NEJM — Identifying patients at risk of adverse events
