Starting Jan. 1, 2027, California hospitals, clinics and doctors' offices that lean on artificial intelligence to help diagnose or treat patients will have to make sure a licensed human is still making the call. That is one of the main results of the health care AI bills Gov. Gavin Newsom acted on before his Sept. 30 deadline, according to a review by the Oakland-based California Health Care Foundation. Of five key measures, he signed three and vetoed two.
What becomes law
AB 1979, by Assemblymember Mia Bonta, covers what it calls clinical decision support systems: AI tools that predict, classify or recommend things like diagnosis, treatment or the timing of care. Providers must take reasonable steps to ensure a licensed clinician uses independent judgment whenever such output shapes a patient's care. The software cannot perform clinical work reserved for licensed professionals, and it cannot be used to direct unlicensed staff to do it. Scheduling tools and routine messages like appointment reminders are exempt. The law also treats companies that run health care chatbots as health care providers under the state's medical privacy law, so they carry the same confidentiality duties and penalties.
A companion law, SB 503, also takes effect in January. It requires the companies that build these tools and the facilities that use them to make reasonable efforts to spot systems that could give worse care to people based on race, sex, age or other protected traits. Developers must hand over documentation on training data, testing, intended use and known risks, and facilities must keep monitoring results.
The third signed bill, SB 947, limits how employers, including large health systems, can use automated software to discipline or fire workers starting July 2027.
What Newsom rejected
The governor vetoed SB 903 by Sen. Steve Padilla, which would have barred companies from advertising chatbot services as therapy. It also would have limited licensed therapists to using AI for administrative or supporting tasks, and required patients to give clear consent before a session was recorded or transcribed by AI. Newsom wrote that the bill's definitions and scope were broad enough to hurt legitimate clinical tools, the foundation reported.
Padilla, a San Diego Democrat, called the veto disappointing in an Oct. 1 statement. He noted the bill passed the Senate unanimously and the Assembly 71 to 4, and said he will bring it back next session. He said thousands of licensed mental health professionals had urged Newsom to sign it.
Newsom also vetoed AB 2575, which would have protected clinicians from punishment for overriding AI recommendations. He said its evidence standard was impractical and that the labor commissioner lacks the medical expertise to judge standards of care.