Kaiser Permanente advice and triage nurses say call monitoring and automated performance tools pressure them to prioritize speed over professional judgment, raising broader questions for physicians about clinical autonomy, patient safety, and workplace AI governance.
Seven current and former nurses told media outlets that managers routinely question calls lasting longer than 15 minutes and that call duration can affect monthly performance scores. The nurses also described software that predicts whether employees appear unproductive or respond too slowly.
Kaiser disputes those accounts. The organization says it does not use average handle time to assess performance or enforce call-time requirements. Kaiser also says its contact-center tools support quality assurance and include human review and oversight.
The nurses said productivity pressure becomes particularly difficult when callers have multiple symptoms, need interpreters, express suicidal thoughts or require support after receiving life-changing diagnoses. Kaiser call-center nurse and union representative Pa Vue told the news organizations that she had observed nurses receive lower performance scores after they departed from software recommendations based on their professional judgment.
The dispute also involves an AI experiment that Kaiser nurses say began during summer 2024. According to the nurses, the tool attempted to assess empathy and vocal tone during calls involving nurses and patients. The test ended in November 2024, but union representatives said managers might bring it back. Kaiser declined to answer detailed questions about the tool, including whether it evaluated nurse performance and whether patients knew that AI was assessing empathy and tone.
The investigation did not establish that call monitoring or AI evaluation caused adverse patient outcomes. Advice nurses generally lose contact with patients after calls end, limiting their ability to identify downstream effects. News stories also reported that a 2024 public-records request found no complaints to the California Department of Managed Health Care involving Kaiser call times. The absence of complaints, however, does not establish whether the systems affected the quality of care.
California lawmakers are considering protections that could apply more broadly to physicians and other healthcare professionals. Pending AB 2575 would prohibit specified retaliation against direct-care professionals who override clinical decision-support outputs when their judgment requires an override to meet the standard of care or comply with the law. It would also require healthcare organizations to provide clinicians with access to certain information about their clinical decision-support systems.
Pending AB 1883 would restrict employers from using AI-enabled workplace-surveillance tools to recognize workers’ emotional states, subject to specified exceptions. Both bills remained active in the Senate Appropriations Committee as of July 19, with hearings scheduled for August 3.
The Kaiser controversy illustrates a governance problem that physicians will increasingly encounter: Health systems may introduce tools that do more than support clinical decisions. Those systems may also measure productivity, recommend actions, or evaluate communication.
Medical staff and physician organizations will need clear policies defining what these tools measure, how employers use the results, whether clinicians can challenge automated assessments, and who remains accountable when efficiency metrics conflict with patient needs.
To protect clinical autonomy and patient safety, physicians should ask whether automated systems evaluate their productivity, communication style, or adherence to clinical recommendations. They must also seek written protections for any clinical decisions that override software-generated suggestions and require their active participation in the selection, testing, and monitoring of workplace AI systems. Finally, physicians should clearly distinguish between patient-safety measures and productivity metrics, ensuring the latter does not discourage the time necessary for complex cases.
Kaiser declined reporters’ request for a list of the AI systems it uses. The available reporting does not establish whether call monitoring has affected patient outcomes or whether Kaiser plans to revive AI-based evaluation of empathy and tone.