Estimated reading time: 5 minutes · Last updated:
Assembly Bill 1979, signed by Gov. Gavin Newsom and taking effect 1 January, prohibits hospitals and doctors’ offices from letting artificial intelligence independently perform medical tasks that require a professional license, such as prescribing medications or ordering tests. The law requires healthcare facilities to take reasonable steps so licensed professionals can exercise independent judgment when AI tools guide care. Assemblymember Mia Bonta carried the bill, sponsored by the California Nurses Association. The measure keeps clinicians able to use AI to inform decisions but bars AI-alone decision‑making in functions reserved for licensed practitioners.
AI is rapidly integrating into our health care system and reshaping our personal experience with it,
Assemblymember Mia Bonta
Key takeaways
- New law: Assembly Bill 1979, signed by Gov. Gavin Newsom, takes effect 1 January and bars AI from independently performing licensed medical tasks.
- Legislative sponsor: Assemblymember Mia Bonta carried AB 1979, which was sponsored by the California Nurses Association.
- Hospital stance: The California Hospital Association and the California Medical Association, representing 50,000 doctors, were neutral on the bill.
- Kaiser research: Internal Kaiser research published in 2020 and cited by Kaiser found a predictive algorithm reduced deaths by 16%.
Table of contents
What AB 1979 requires and how hospitals must respond
AB 1979 sets a narrow procedural rule: healthcare facilities must take reasonable steps to ensure licensed professionals retain control over decisions that require a professional license. The bill’s text requires that licensed professionals can exercise independent judgment when AI tools help guide patient care. That language permits clinicians to consult AI‑driven decision support while making clear AI cannot be the sole decision‑maker for prescribing, ordering diagnostic tests or other acts that state law reserves to licensed practitioners.
Operationally, hospitals and clinics will need to show policies, training and audit trails that keep clinicians engaged in decisions. The law allows AI for documentation and communication tasks and for tools that assist clinicians, but facilities must document how human review is integrated into workflows. The requirement is procedural: it focuses on steps taken to preserve clinician authority rather than banning particular software packages outright.
Where the bill came from: unions, cases and vendor tools
The California Nurses Association sponsored the bill after nurses raised concerns that administrators were inserting AI into workflows previously performed by registered nurses and physicians. Michelle Gutierrez Vo, a registered nurse at Kaiser Permanente in Fremont and a union president, said nurses are worried about employers replacing clinical functions with generative AI or algorithmic devices. The National Union of Healthcare Workers filed a complaint in July alleging Kaiser implemented an algorithm to assess patients seeking mental healthcare and produced recommendations without clinician oversight.
Kaiser Permanente responded through spokesperson Elissa Harrington, saying 'At Kaiser Permanente, technology, AI, and other advanced tools do not make care decisions.' Kaiser also pointed to an internal study published in 2020 that it says reduced deaths by 16% using a predictive alert algorithm. Press accounts in the Wall Street Journal documented cases in which an algorithm's recommendation was followed by an advice nurse and the patient later died; an arbitrator later held the nurse responsible in that instance. Those specific episodes and the unions' filings are the context that pushed lawmakers to act.
Clinical and legal implications for clinicians and vendors
Clinicians are already using generative and clinical decision‑support tools such as Open Evidence and Abridge to summarize literature, draft notes and suggest diagnoses. Dr. Donaldo Hernandez, who manages hospitalized patients at two Santa Cruz centers, described Open Evidence as a 'fast librarian' that produces citations and care suggestions clinicians must vet. AB 1979 preserves that advisory role for AI but raises the bar on how hospitals document human oversight.
On the legal front, Henry Norwood, a healthcare attorney at Kaufman Dolowich, expects to see malpractice claims that allege over‑reliance on AI guidance when care goes wrong. The law does not change standards of care directly, but it creates a statutory expectation that clinicians exercise independent judgment when AI is used — a fact litigators can point to in malpractice cases. The California Medical Association, representing 50,000 doctors, remained neutral on the bill after its language was narrowed from earlier versions that would have been broader.
| Actor | Position or use | Example cited in reporting |
|---|---|---|
| California Nurses Association | Sponsor of AB 1979 | Raised concerns about AI replacing nurse functions |
| Kaiser Permanente | Uses predictive algorithms; defends clinical oversight | Cited internal 2020 research showing a 16% reduction in deaths |
| National Union of Healthcare Workers | Filed complaint against Kaiser | Alleged mental‑health screening algorithm recommended care without licensed oversight |
| California Medical Association | Neutral on AB 1979 after narrowing | Represents 50,000 doctors |
How AB 1979 could change AI use in California health care
The case for
- Hospitals that document clinician review and build clear escalation steps will keep decision‑support benefits while meeting the law’s requirements.
- Clinicians may gain leverage to shape AI deployment, ensuring vendors build human‑in‑the‑loop controls and audit logs.
The case against
- Hospitals could respond by limiting AI features to avoid compliance risk, slowing adoption of tools that detect deterioration or flag diagnoses.
- Vendors may move features out of California or harden contracts against local liability exposure, reducing clinician access to advanced tools.
What to be careful about
- A surge in malpractice claims alleging clinician over‑reliance on AI, as predicted by attorney Henry Norwood, could raise insurer premiums or change hospital liability practices.
- Ambiguity about what counts as 'reasonable steps' to preserve independent judgment could produce uneven enforcement and disruptive audits.
- Operational burdens from documenting human review and retraining staff could divert clinician time from patients and slow tool rollout.
The bottom line
AB 1979 draws a clear boundary: AI can inform but not replace licensed clinicians in functions the state reserves for professionals. The law leaves room for clinical decision support, documentation tools and other applications, but it shifts the compliance burden onto hospitals to show clinicians retain independent judgment. Expect operational changes as systems add human‑in‑the‑loop checks, and expect litigation and guidance to follow as courts and regulators interpret what constitutes adequate oversight.
What to watch
- Watch for enforcement guidance or model policies from state regulators explaining what 'reasonable steps' hospitals must take; no date has been set.
- Watch for any malpractice filings that cite AB 1979 or claim clinicians over‑relied on AI guidance; no date has been set.
- Watch how major health systems such as Kaiser Permanente adjust their documented oversight and training; no date has been set.
Frequently asked questions
What does AB 1979 prohibit?
AB 1979 bars hospitals and clinics from letting AI independently perform medical tasks that require a professional license, such as prescribing medications or ordering tests, and requires steps so licensed professionals can exercise independent judgment.
When does the law take effect?
The law takes effect 1 January; the source sets that effective date but does not specify the year in its quoted language.
Will existing AI tools like those Kaiser uses be banned?
The law does not categorically ban AI tools; Kaiser has said its predictive algorithm saved lives and cited a 16% reduction in deaths in a 2020 study, but the law requires human oversight and documented steps to ensure clinician judgment.
How did medical organizations react?
The California Hospital Association and the California Medical Association, which represents 50,000 doctors, took neutral positions after the bill’s language was narrowed.
Related reading
This article is information, not medical advice. Anyone acting on it should speak to a qualified professional.
