About 2,400 mental health professionals at Kaiser Permanente walked off the job this week in the first major labor action specifically targeting AI in healthcare. Their complaint: Kaiser replaced trained clinicians with AI-driven questionnaires and phone operators for patient screening, and high-risk patients are slipping through the cracks.
The one-day strike on March 18 brought picket lines to Kaiser facilities across Northern California, from Oakland to Santa Rosa to Sacramento. More than 23,000 nurses joined in support.
What Changed
The issue centers on how Kaiser decides which patients need immediate mental health care.
According to union representatives, what used to be a 10-to-15-minute screening by a licensed clinician is now handled by “unlicensed lay operators following a script, or e-visits” - automated questionnaires that make recommendations about next steps in care.
One therapist told KQED her screening team in the Walnut Creek area was reduced by two-thirds, with AI and phone operators filling the gap.
The union says this shift began in 2024, and the consequences have been dangerous.
Patients Falling Through
A clinician reported that patients screened by telephone operators were arriving at appointments already self-harming - situations that should have triggered immediate clinical risk assessment but didn’t.
Another therapist described the timeline: “By the time triage sends them, it’s been a month… and that’s really dangerous” for patients in acute distress.
The union considers the replacement of licensed clinicians with unqualified operators an unfair labor practice that violates California state law.
Kaiser’s Position
Kaiser denied that AI makes care decisions or replaces therapists.
“AI can be helpful when it supports clinicians - by reducing administrative work or improving efficiency - but it does not replace clinical judgment or human assessment,” a spokesperson told reporters.
The company pointed to nearly $2 billion invested in mental health services since 2020 and a doubled workforce over ten years.
A Kaiser spokesperson also accused the union of “pushing a false narrative” about AI replacing care teams.
The therapists have been working without a contract since September. Both sides describe negotiations as “far apart.”
Why This Matters
This isn’t a theoretical debate about AI in healthcare. It’s a live test case.
Dr. John Torous, digital psychiatry director at Beth Israel Deaconess Medical Center, warned that many AI tools in mental health are “exciting, but they’re not well-tested.”
Vaile Wright, senior director at the American Psychological Association, noted that she hasn’t seen mental health jobs actually replaced by AI yet - “in part because there are no AI digital solutions that can replace human-driven psychotherapy or care.”
But the Kaiser situation suggests a different pattern: AI isn’t replacing therapy sessions. It’s replacing the screening that determines who gets access to therapy in the first place.
If an AI questionnaire or untrained operator misses warning signs in a patient seeking help, that patient may never reach a therapist at all.
The Broader Picture
Healthcare AI has largely been celebrated for efficiency gains - automating documentation, improving billing, analyzing imaging. Those applications keep humans in the loop for clinical decisions.
But the Kaiser strike highlights a murkier territory: using AI for gatekeeping functions where mistakes can be invisible until they’re catastrophic.
The therapists aren’t arguing that AI should be banned from healthcare. They’re arguing that AI shouldn’t make the first call on whether a patient in crisis gets immediate help or waits a month.
Kaiser says AI supports clinicians. The clinicians say AI replaced them at the point where clinical judgment matters most.
What Happens Next
The strike was a one-day action, following a larger four-week walkout by 30,000 Kaiser workers across California and Hawaii in February over wages and benefits.
Contract negotiations continue. The union is seeking explicit language that Kaiser won’t use AI to replace therapists - language the company has reportedly resisted.
Meanwhile, the broader healthcare industry is watching. If AI screening tools can cut costs at scale, expect more health systems to adopt them. If those tools miss patients in crisis, expect more strikes - or worse.
The question isn’t whether AI will transform healthcare. It’s who decides where the human handoff happens, and what we lose when we get that boundary wrong.