California’s new AI health care law leaves key compliance questions open

California’s AB 1979 takes effect Jan. 1, 2027, and requires health care practices to preserve physicians’ independent judgment when AI informs care. The law permits AI-assisted decision support, but it does not spell out how much review is enough, leaving batch approvals and workflow design as major compliance questions.

Why it matters: - California health care practices will need to prove that physicians still make independent clinical decisions when AI tools inform patient care. - The law gives practices flexibility, but it also creates risk for high-volume workflows that rely on quick sign-offs. - Batch approvals, telehealth triage and AI-drafted patient communications now need closer legal and operational review.

What happened: - Gov. Gavin Newsom signed AB 1979 on Sept. 30, 2026. - Assemblymember Mia Bonta, D-Oakland, authored the bill. - The California Nurses Association sponsored the legislation. - The law adds Section 22758.5 to the California Business and Professions Code. - AB 1979 takes effect Jan. 1, 2027. - The law applies to health facilities, clinics, physician offices and group practice offices. - Newsom also signed SB 503, which targets bias in clinical decision support tools. - Newsom vetoed AB 2575 and SB 903.

The details: - Covered practices must take reasonable steps to ensure that a licensed provider retains the ability to exercise independent professional judgment whenever care is informed by a clinical decision support system. - The law defines a clinical decision support system as an AI system that produces a prediction, classification, recommendation, evaluation or analysis to aid decisions about timing of care, diagnosis or treatment. - Practices may not use AI to independently perform a clinical function that state law requires a licensed person to perform. - Practices also may not use AI to direct, guide, supervise or instruct unlicensed personnel in those licensed functions. - The law exempts documentation and communication tools that do not require professional judgment, including automated record-update notices, appointment reminders and patient search tools. - The law also excludes trainees using a tool in a supervised course of study. - Violations by physicians fall under the Medical Board of California or the Osteopathic Medical Board of California. - A licensing board may seek an injunction or restraining order when a violation amounts to unlicensed practice. - AB 1979 allows AI-assisted decision support when professional judgment remains with licensed health care providers. - Earlier drafts went further, but lawmakers removed a ban on clinical decisions based solely on AI output. - Lawmakers also revised language that would have required a licensee to exercise independent judgment when reviewing and approving AI clinical decisions. - As enacted, the duty falls on the practice to take reasonable steps so the clinician retains the ability to exercise that judgment. - The law does not expressly address batch approvals. - The law does not set a minimum review time per patient. - The law does not automatically make existing AI-assisted workflows unlawful on Jan. 1, 2027.

Between the lines: - The biggest compliance risk is not whether a physician signs off, but whether the physician had a real chance to review the underlying information and reach a different conclusion. - A signature alone may not prove meaningful oversight. - In a telehealth workflow with hundreds of low-acuity cases, grouped approval could become a rubber stamp if physicians do not actually evaluate each case. - The law places obligations on the practice itself, so a medical group cannot assume the vendor owns compliance. - A too-strict interpretation could slow care, raise costs and limit lower-cost service models. - The authors argue that training matters because physicians need to know when AI output should be questioned rather than trusted automatically. - A 2026 American Medical Association survey found that 72% of physicians who answered the relevant question reported using at least one AI use case in practice, up from 38% in 2023. - Another 9% were unsure which AI tools their practice offered. - Eighty-eight percent expressed at least some concern about AI-related skill loss.

What's next: - Practices should sort administrative tools from tools that influence diagnosis, treatment or other licensed functions. - Practices should map what information physicians see, how much time the workflow allows and how easy it is to reject AI recommendations. - Practices should set written rules for grouped review if physicians approve cases in batches. - Practices should document overrides when physicians reject AI recommendations. - Practices should train physicians on each tool’s limits and when to disregard output. - Practices should confirm that AI-drafted clinical messages include the AB 3030 disclaimer or receive documented clinician review. - Practices should check that consumer health chatbots comply with the Confidentiality of Medical Information Act. - Until regulators issue more guidance, records showing how physicians review, challenge and document AI output may be the best evidence that professional judgment stayed central.

The bottom line: - AB 1979 permits AI in health care, but it puts the burden on practices to show that licensed clinicians still control the clinical decision. - How much physician involvement is enough remains unresolved.