Kaiser Nurses Say AI and Surveillance Harm Patient Care

{ "title": "Kaiser nurses say AI surveillance and call time pressure put patients at risk", "body": "Advice and triage nurses at Kaiser Permanente say the company’s use of workplace surveillance, including artificial intelligence tools that track their tone of voice and call times, is undermining patient care and putting lives at risk. The complaints come as the California Nurses Association (CNA) begins negotiating a new contract this month for 25,000 nurses, including 1,000 who work in call centers.\n\nSeven current and former Kaiser nurses interviewed by CalMatters described a system where spending more than 15 minutes on a single call can trigger criticism or a performance evaluation meeting. Call time factors into monthly performance scores. Kaiser uses software to predict daily whether nurses are unproductive or failing to answer calls quickly. AI systems also rate nurse empathy and tone of voice.\n\n“I think at some point all of the nurses have been talked to about their average handle time,” said Raquel Alvarez Sanchez, a Kaiser advice nurse in Vallejo who has worked there since 2010 and is a union steward. Last year, she spent more than an hour on a call with a suicidal patient. She said the pressure to keep calls short is constant.\n\n“The only thing I can think of is they’re doing it for profit,” Sanchez said.\n\nKaiser Permanente provides healthcare to more than 9 million people in California and 3 million other Americans. It is the largest private employer in California. Its use of AI to manage workers could set precedents for the healthcare industry.\n\n## Nurses fear reprisal for showing compassion\n\nAn anonymous nurse described a call last year with an elderly woman who had just received a terminal cancer diagnosis. The nurse said she withheld compassion because she feared being reprimanded for a long call.\n\n“I had to ask myself: Am I going to get disciplined for going off script or saying more than what is necessary?” the nurse said.\n\nNurses said they are instructed to stick to a script and give no more than two or three pieces of advice. Calls with interpreters often take 30 minutes or more. About 4 in 10 Californians speak a language other than English, and half of them do not speak English well, according to the state’s environmental health agency.\n\nKaiser spokesperson Vincent Staupe said the company does not use average handle time to assess performance.\n\n“Kaiser Permanente does not use Average Handle Time to assess agent performance or enforce call time metrics. Any tools used in contact center settings support our quality assurance efforts and have human review and oversight,” Staupe said.\n\nHe added that Kaiser uses AI responsibly with human oversight, prioritizing patient safety, privacy and equity. The company declined to share specific information about internal technology systems.\n\n“As a large organization, we do not share specific information about internal technology systems for security and operational reasons,” Staupe said.\n\nThe pressure to keep calls short has been a recurring issue for Kaiser. In 2002, the company faced criticism for paying bonuses to non-nurse call center workers for keeping calls short. That history has left many nurses skeptical of current assurances. The union has tracked how call time metrics have evolved over the years, with the current system relying on software that predicts daily whether nurses are unproductive or failing to answer calls quickly. This algorithmic management, nurses say, ignores the complexity of medical triage.\n\nNurses also described how the pressure to keep calls short affects their ability to provide culturally competent care. For patients who speak limited English, calls often require interpretation services, which can double or triple the time needed. Nurses said they are sometimes forced to rush through these calls, potentially missing critical symptoms or failing to build the trust needed for effective care. The script-based approach, they argue, does not account for the nuances of language barriers or the emotional needs of patients facing serious diagnoses.\n\n## AI empathy tests ended but may return\n\nIn summer 2024, Kaiser began testing an AI tool to assess empathy and tone in the voices of nurses and patients. Nurses circulated and signed a petition demanding patient privacy, transparency and the right to exercise professional judgment. The petition’s tag line: “Trust nurses, not AI.”\n\nCharlotte Capulong has worked in nurse call centers for 22 years and helped organize against the AI tone-of-voice tool.\n\n“I can see now where a lot of the problems,” Capulong said, “and that was intensified when they said we’re going to use AI to evaluate our calls and grade us.”\n\nShe said the AI did not understand the nuance of nursing.\n\n“AI did not understand our job and would grade us wrong all the time,” Capulong said.\n\nThe AI tests ended in November 2024. Union representatives said managers told them the program may be brought back.\n\nMichele Ramos, a patient advocate with Consumer Watchdog, said the focus on call times and AI surveillance reflects a broader problem.\n\n“Kaiser’s been known through the years to manage dollars over managing care, … which is only going to fail patients,” Ramos said. “Kaiser’s been known through the years to manage dollars over managing care, and I think this would be a contributor to that, which is only going to fail patients.”\n\nThe AI empathy detection tool is just one of several AI systems Kaiser has deployed. The company uses AI to identify patients at risk of adverse events via electronic health records, to determine when to discharge patients with the Preventus system, and to record therapy sessions with the Abridge system. Kaiser also tested remote monitoring with AI in Bay Area facilities. The breadth of these tools has raised concerns among nurses and therapists about how AI is shaping patient care and worker autonomy.\n\nNurses said the AI empathy tool was particularly troubling because it judged their tone without understanding context. For example, a nurse might need to sound firm when giving urgent medical instructions, but the AI could interpret that as lacking empathy. Conversely, a calm tone during a crisis could be misread as appropriate, even if the nurse was failing to convey urgency. Nurses said these misjudgments led to unfair performance scores and increased anxiety about how they speak to patients.\n\nThe union has documented cases where nurses were penalized for calls that the AI flagged as low empathy, only to have the flag overturned upon human review. But the process of appealing AI decisions is time-consuming and stressful, nurses said. Some said they have stopped using certain phrases or tones that they believe the AI might misinterpret, even if those approaches were effective with patients.\n\n## Call time pressure raises safety concerns\n\nNurses said they now get 30 seconds or less between calls when lines are busy. Previously, they had about 10 minutes between calls.\n\n“The amount of time that Kaiser is giving us to complete a call is sometimes not safe,” said Pa Vue, a Kaiser call center nurse for about 10 years and a union representative.\n\nVue said her evaluation scores were reduced for repeating advice to a patient with possible heart issues. She also saw nurses receive lower scores for going against software recommendations or making appointments without consulting a doctor.\n\n“You aren’t calling Comcast. We’re dealing with life here,” Vue said.\n\nAnnette Bernhardt, director of the UC Berkeley Labor Center Technology and Work Program, warned that algorithmic management can turn people into “fleshy robots.”\n\n“Stress and burnout can lead to more mistakes across a range of areas, and in the healthcare setting that is much higher risk because you’re dealing with people’s lives and their health,” Bernhardt said.\n\nA 2023 academic survey of call centers in four developed countries found that AI monitoring left workers with less time between calls and more emotionally drained. Nearly half said AI made their jobs more stressful. A prior study by Virginia Dolleghast of Cornell University and Sean O’Brady of McMaster University found that performance monitoring leads to higher emotional exhaustion.\n\nThe pressure on nurses has been building for years. In 2019, Kaiser faced criticism from the California Department of Managed Health Care over delayed behavioral health appointments and moving patients to group therapy. That led to a record $50 million fine as part of a settlement in 2022. The company also settled with the U.S. Department of Labor over substance use and mental health services. Nurses say the focus on call times and AI surveillance is part of a pattern where financial considerations override patient safety.\n\nNurses shared specific examples of how call time pressure led to potential safety risks. One nurse described a call where a patient reported chest pain but also mentioned a recent family tragedy. The nurse felt compelled to focus on the medical symptoms quickly and move on, rather than taking time to understand the full context. Another nurse said she had to cut short a call with a diabetic patient who was confused about insulin dosing, because the system was warning that her call time was exceeding the limit. She later worried the patient might have made a dangerous error.\n\nThe reduction in between-call time from 10 minutes to 30 seconds or less means nurses have no time to document calls properly, review patient histories, or decompress after emotionally difficult conversations. Nurses said this increases the risk of errors in documentation and follow-up. They also said the constant pressure leaves them feeling exhausted and less able to focus on the next call.\n\n## Nurses distrust AI in patient care\n\nA 2024 survey by National Nurses United of more than 2,000 nurses found that half said their employer uses algorithmic systems to analyze health records. Two-thirds said their assessments disagreed with computer predictions. Six in 10 said they do not trust their employer to prioritize patient safety with AI.\n\nCathy Kennedy, president of National Nurses United and CNA, said the AI empathy detection is part of steps to limit nurse autonomy.\n\n“It’s an ongoing fight, and it’s a fight well worth having,” Kennedy said.\n\nCNA is attempting to track the number of AI models at Kaiser and advising members to report new technology. The union stopped a pilot program that would have replaced nurses at the bedside of confused patients with cameras.\n\nDebru Carthan, a Kaiser radiologist and SEIU member who serves on the Coalition of Kaiser Permanente Unions committee on AI, worries the AI tone detector could discriminate against nurses from different cultures.\n\n“Whenever there are other unions in discussion about artificial intelligence we are in solidarity with them,” Carthan said.\n\nAn anonymous nurse said she used humor to help patients heal but no longer feels comfortable because calls are recorded.\n\n“That really takes away from the whole point of being a nurse and what patients come to know from nurses,” she said.\n\nKaiser therapists are also concerned about the use of therapy session transcripts to train AI models and the possibility that AI could take their jobs. The National Union of Healthcare Workers (NUHW) represents 2,400 Kaiser mental health workers in contract negotiations in Northern California. NUHW spokesperson Matt Artz said contract negotiations are ongoing. The union has raised concerns about the Abridge system, which records therapy sessions, and whether patient privacy is being protected.\n\nA 2024 public records request by CalMatters to the California Department of Managed Health Care found no patient complaints about call times. But nurses say patients may not know about the internal metrics driving their care. The lack of complaints does not mean there are no problems, they argue, because patients are not aware of the pressures nurses face.\n\nThe survey also revealed that nurses are deeply skeptical of AI’s ability to replace human judgment in clinical settings. Two-thirds of respondents said their own clinical assessments disagreed with computer predictions, suggesting that AI tools are not yet reliable enough to guide care decisions. Nurses said they fear that over-reliance on AI could lead to missed diagnoses or inappropriate treatment recommendations. They also worry that AI systems are trained on data that may not reflect the diversity of Kaiser’s patient population, leading to biased outcomes.\n\n## California lawmakers consider AI workplace bills\n\nCalifornia lawmakers are considering several bills regulating AI in the workplace. One bill would protect doctors and nurses from retaliation for overriding automated care recommendations. Another would require employers to inform workers before using automated systems for promotion, discipline or performance evaluation.\n\nLast year, Gov. Gavin Newsom vetoed Senate Bill 7, and Assembly Bill 1018 failed to pass for the third consecutive year. A new version of SB 7 has been reintroduced as SB 947.\n\nAnother bill would prohibit employers from using AI to predict the emotional state of employees. CNA and the California Labor Federation support about half a dozen AI workplace bills.\n\nLabor leaders from Democratic primary states held a press conference in Sacramento earlier this year. They said if Newsom wants to be president, he must pass AI worker protection laws.\n\nKaiser declined to share a comprehensive list of AI systems in use. CNA said Kaiser sometimes does not notify nurses of new technology deployments despite contract provisions. The union is pushing for transparency in the current contract negotiations, which began this month and cover 25,000 nurses, including 1,000 call center workers.\n\nThe push for AI regulation comes as Kaiser continues to expand its use of technology. In 2026, the company is expected to roll out new AI systems for patient triage and scheduling, according to internal documents reviewed by CalMatters. Nurses say they fear these systems will further erode their professional judgment and patient trust.\n\nThe broader context of AI in healthcare is evolving rapidly. A 2023 study by the University of California, San Francisco found that AI tools in call centers can reduce call times by up to 20%, but at the cost of increased worker stress and lower patient satisfaction. The study, which surveyed 1,500 nurses across the country, found that those working under AI monitoring reported higher rates of burnout and turnover intention.\n\nKaiser’s use of AI is not limited to call centers. The company has deployed AI in radiology, where it helps analyze medical images, and in emergency departments, where it predicts patient flow. But nurses say the most intrusive applications are those that monitor their performance and behavior.\n\n“We are not robots,” said Sanchez. “We are human beings who care about our patients. And when you take away our ability to show that care, you are putting lives at risk.”\n\nThe contract negotiations are expected to continue through the spring. CNA has said it will push for limits on AI surveillance and for more nurse input into technology decisions. Kaiser has said it is committed to using AI responsibly and with human oversight.\n\nThe outcome of these negotiations could have implications beyond California. As one of the largest healthcare providers in the country, Kaiser’s policies on AI and worker surveillance could set a precedent for the industry. Other healthcare systems, including Dignity Health and Sutter Health, are also experimenting with AI tools for call centers and patient monitoring.\n\nFor now, nurses say they will continue to do their jobs as best they can, despite the pressure.\n\n“We are here for the patients,” said Vue. “And we will not let a machine tell us how to care for them.”\n\n## Related on Neura Market\n\n- California Labor and Employment Law\n- Healthcare AI Regulation\n- Workplace Surveillance Technology" }