California’s New Health Care AI Law (AB 1979): What Orange County Seniors and Families Should Know Before 2027

Robert Gordon
Home Care Policy Analyst · LinkedIn · October 6, 2026
10 min read

Your mother’s cardiologist glances at a screen, and the screen has already suggested a diagnosis. Your father’s hospital discharge summary was drafted by software before a nurse ever read it. A friendly “health assistant” on a clinic’s website answers your questions at 2 a.m. and remembers everything you typed. Artificial intelligence is already part of how older adults in Orange County get medical care, and until now California law said surprisingly little about who is actually in charge. On September 30, 2026, Governor Newsom signed Assembly Bill 1979 (Chapter 854, Statutes of 2026). Starting January 1, 2027, it draws a clear line: the computer can advise, but a licensed human has to decide. Here is what the new law does, what it does not do, and the questions every family should start asking now.

Sept 30AB 1979 Signed (2026)
Ch. 854Statutes of 2026
Jan 1, 2027Date the Law Takes Effect
4Care Settings Covered: Facilities, Clinics, Doctor Offices, Group Practices
Clinician in scrubs and glasses holding a stethoscope while reviewing a laptop screen

Under AB 1979, an AI tool can inform a clinician’s decision. It cannot replace the clinician’s judgment.

What AB 1979 Actually Says

AB 1979, authored by Assemblymember Mia Bonta, is short, and it does three things. Each one is aimed at a different way AI shows up in health care.

1. A licensed professional keeps the final say

When a patient’s care is informed by a clinical decision support system, the software that flags a risk, suggests a diagnosis or recommends a treatment, the law requires a health facility, clinic, physician’s office or group practice office to take reasonable steps so that the licensed health care professional, acting within their scope of practice, keeps the ability to use independent professional judgment. In plain terms, a doctor or nurse must be able to disagree with the machine and act on that disagreement.

2. AI cannot do licensed work on its own

The same settings may not use a tool that includes artificial intelligence to independently perform a clinical function that state law says must be performed by a licensed person. Nor can AI be used to direct unlicensed staff in doing that licensed work. If California law requires a registered nurse to assess a patient, an algorithm cannot quietly do the assessing while an aide carries out its instructions.

3. Health chatbots now answer to California’s medical privacy law

California’s Confidentiality of Medical Information Act (CMIA) has long limited how doctors, hospitals and health plans may share your medical information. AB 1979 deems a business that offers a health care chatbot to be a “provider of health care” under the CMIA. That matters because a symptom you type into a chatbot is medical information, and the company behind it now carries the same confidentiality duties a clinic does.

Who is covered, and who is not. AB 1979’s clinical rules are written for health facilities (such as hospitals and skilled nursing facilities), clinics, physicians’ offices and group practice offices. The chatbot rule applies to businesses that offer health care chatbots. Non-medical in-home care agencies, including At Home VA Staffing, are not named in the law. Our caregivers are people, and every care decision in our households is made by people.

What Was Already Law, What Is New, and What Was Vetoed

California has been building AI rules for health care one bill at a time, and it is easy to give a new law credit for something an older one already did. Here is the honest scorecard.

RuleSourceStatus
When generative AI writes a message to a patient about their clinical care, the message must carry a disclaimer and tell the patient how to reach a human providerAB 3030 (2024)Already in effect since January 1, 2025
A health plan’s AI cannot deny, delay or change care for medical necessity on its own; a licensed physician or qualified health professional must make that callSB 1120 (2024)Already in effect since January 1, 2025
A licensed clinician keeps independent judgment when a clinical decision support system informs careAB 1979 (2026)New, effective January 1, 2027
AI may not independently perform clinical functions the law reserves for licensed professionalsAB 1979 (2026)New, effective January 1, 2027
Businesses offering health care chatbots are treated as health care providers under the CMIAAB 1979 (2026)New, effective January 1, 2027
A separate bill with a nearly identical title, “Health care services: artificial intelligence”AB 2575 (2026)Vetoed September 30, 2026; not law

That last row is worth remembering. Because the two bills share nearly the same title, they are easy to mix up. AB 2575 was vetoed; AB 1979 is the bill that became law.

Why This Matters More for Older Adults

Seniors see more doctors, take more medications, and spend more nights in the hospital than any other group, so they meet more of these tools than anyone. They are also the patients most likely to be underserved by software trained mostly on younger, healthier people, and the most likely to have a family member managing their care from a distance.

Three situations come up again and again in Orange County families:

  • The hospital stay. Discharge planning, fall-risk scores and sepsis alerts increasingly run on software. If a tool says your father is ready to go home and his nurse is not so sure, AB 1979 protects the nurse’s ability to act on that concern. (If he was held under observation rather than admitted, read our guide to Medicare observation status and the 3-day rule too.)
  • The doctor’s visit. Ambient AI scribes that listen and draft visit notes are becoming common. The note in your mother’s chart should reflect what her physician concluded, not what the software inferred.
  • The late-night question. An adult child in Irvine types a parent’s symptoms, medications and diagnosis into a chatbot. Starting in 2027, that company is bound by the same medical confidentiality law as a clinic.
Physician in a white coat with a stethoscope listening to a patient across a desk with a computer monitor

The screen can inform the visit. Under AB 1979, the decision belongs to the licensed professional in the room.

Where You Will Notice It, and What to Ask

The law puts duties on providers, not on patients. But families who know the rules ask better questions, and better questions get better care.

SettingWhere AI may show upA good question to ask
Hospital or skilled nursing stayRisk scores, discharge timing, monitoring alerts“Who reviewed this recommendation, and do you agree with it?”
Physician or specialist officeVisit-note scribes, diagnosis suggestions, imaging reads“Is anything in today’s plan based on an AI tool? What did you decide?”
Patient portal messagesAI-drafted replies to your questions“Was this reply reviewed by a clinician? How do I reach a person?”
Health chatbots and symptom checkersSymptom triage, medication questions“Who runs this service, and what happens to what I type?”
Medicare Advantage or Medi-Cal planPrior-authorization screening“Was this denial reviewed by a physician?” (SB 1120 already requires it)
A note on timing. AB 1979 takes effect January 1, 2027. Between now and then, the AB 3030 disclaimer rule and the SB 1120 health-plan rule already apply, and you can ask every question above today. Asking whether a human reviewed a recommendation is always reasonable.

Health Chatbots and Your Parent’s Privacy

Many families use chatbots and symptom checkers to make sense of a new diagnosis or a confusing medication list. They can be genuinely helpful. They can also collect a remarkable amount of sensitive detail: diagnoses, prescriptions, cognitive changes, even a parent’s address and date of birth.

Before AB 1979, whether a chatbot company owed you medical confidentiality depended on how it was set up. Starting January 1, 2027, a business offering a health care chatbot is treated as a provider of health care under the CMIA, so it generally needs authorization before sharing that information. Even so, a few habits remain smart:

  • Use the chatbot inside your parent’s own health system portal when one is offered, rather than an unknown app.
  • Leave out full names, Social Security numbers and Medicare numbers. A symptom question rarely needs them.
  • Treat the answer as a starting point for a call to the doctor, never as the plan itself.
  • Be wary of any “assistant” that calls or texts your parent first and asks for personal details. That is a scam pattern, not a new law at work. Our article on AI voice-cloning scams targeting seniors explains how those calls work.
Older woman with gray curly hair in a white jacket and scarf looking at her smartphone

From January 2027, companies offering health care chatbots carry the same confidentiality duties as a medical provider under California law.

If You Think a Machine Made the Call

Most of the time, the clinician did review the recommendation, and asking simply gets you a clearer explanation. If you still believe an AI tool, not a professional, decided something about your parent’s care, these are the right doors:

  • Start with the care team. Ask the attending physician, charge nurse or patient advocate to explain who made the decision and why. Hospitals have patient relations offices for exactly this.
  • Licensed professionals: the Medical Board of California handles complaints about physicians at (800) 633-2322, and the Board of Registered Nursing handles complaints about nurses.
  • Hospitals, nursing homes and clinics: the California Department of Public Health licenses health facilities and investigates complaints about them.
  • Plan denials: for a Medicare plan, follow the appeal steps in the denial notice. Orange County’s free Medicare counseling program, HICAP, can help at (714) 560-0424. Medicare open enrollment runs through December 7, so it is also a good time to review what is changing in Medicare plans for 2027.

The Bigger Picture for Families Planning Ahead

AB 1979 is part of a busy year for California health law. If you are updating a parent’s care plan this fall, it pairs naturally with the state’s new POLST rules, which govern the medical orders that travel with a seriously ill patient. Our guide to SB 1088 and California’s POLST changes covers that side. The common thread is simple: the people who know your parent, their doctors, their family and their caregivers, are the ones who should be making decisions about their care.

Your AI-in-Health-Care Family Checklist

Click each step as you finish it.

  • List the hospitals, medical groups and specialists your parent uses
  • Check whether your parent’s patient portal has an AI assistant, and how to reach a human instead
  • At the next appointment, ask whether any part of the plan came from an AI tool and what the doctor decided
  • Look for the required disclaimer on AI-written portal messages, and note who to contact
  • Review which health apps or chatbots your parent uses and who operates each one
  • Remove full names, Social Security and Medicare numbers from chatbot questions
  • If a plan denies care, ask whether a physician reviewed the denial
  • Save the Medical Board of California and HICAP phone numbers in your phone
  • Add “who reviewed this?” to your list of questions for every hospital discharge
  • Talk with your parent about how they want health technology used in their care

Test Your Knowledge

1. When does AB 1979 take effect?
A) September 30, 2026, the day it was signed
B) January 1, 2027
C) July 1, 2028
2. Under AB 1979, who has the final say when an AI tool recommends a treatment?
A) The software, if it is more accurate
B) The hospital’s billing department
C) The licensed health care professional, using independent judgment
3. Which rule was already California law before AB 1979?
A) AI-written clinical messages to patients must include a disclaimer and a way to reach a human
B) Health chatbot companies are providers under the CMIA
C) AI may never be used in hospitals
4. What does AB 1979 change for health care chatbots?
A) It bans them in California
B) A business offering one is treated as a health care provider under California’s medical privacy law
C) It requires every chatbot to be run by a doctor
5. What happened to AB 2575, the bill with a nearly identical title?
A) It was signed alongside AB 1979
B) It was vetoed, so it is not law
C) It takes effect in 2028

Frequently Asked Questions

Does AB 1979 ban AI in California hospitals and doctors’ offices?+

No. AI tools can still support care. The law requires that a licensed professional keeps the ability to use independent judgment when a clinical decision support system informs care, and it bars AI from independently performing clinical work that state law reserves for licensed professionals.

When does the law start?+

AB 1979 was signed on September 30, 2026, as Chapter 854 of the Statutes of 2026, and takes effect January 1, 2027.

Does it apply to home care agencies?+

The clinical rules are written for health facilities, clinics, physicians’ offices and group practice offices. Non-medical in-home care agencies are not named. At Home VA Staffing’s caregivers are trained people, and care decisions in our clients’ homes are made by people.

Is my information private when I use a health chatbot?+

Starting January 1, 2027, a business offering a health care chatbot is treated as a provider of health care under California’s Confidentiality of Medical Information Act, which generally requires authorization before medical information is shared. It is still wise to leave out identifiers such as Social Security and Medicare numbers.

How will I know if AI wrote a message from my parent’s doctor?+

Under AB 3030, in effect since January 1, 2025, clinical messages generated by generative AI must include a disclaimer and instructions for contacting a human provider, unless a licensed provider reviewed the message first. That rule predates AB 1979.

Who can help if a Medicare plan denies my parent’s care?+

Follow the appeal instructions in the denial notice. Orange County’s HICAP program offers free, unbiased Medicare counseling at (714) 560-0424. California’s SB 1120 already requires a licensed physician or qualified health professional, not an algorithm alone, to make medical necessity decisions for health plans it covers.

Care Decisions Made by People Who Know Your Parent

At Home VA Staffing provides in-home care across Orange County, from a few hours of help each week to daily support. Our caregivers notice changes, keep families informed and help seniors get to the appointments where these conversations happen.

Talk to Our Team: (213) 326-7452

Proudly Serving Orange County Communities

Aliso Viejo Anaheim Brea Buena Park Costa Mesa Cypress Dana Point Fountain Valley Fullerton Garden Grove Huntington Beach Irvine La Habra La Palma Laguna Beach Laguna Hills Laguna Niguel Laguna Woods Lake Forest Los Alamitos Mission Viejo Newport Beach Orange Placentia Rancho Santa Margarita San Clemente San Juan Capistrano Santa Ana Seal Beach Stanton Tustin Villa Park Westminster Yorba Linda

This article is for general informational purposes and is not legal or medical advice. It describes Assembly Bill 1979 (Chapter 854, Statutes of 2026), signed September 30, 2026 and effective January 1, 2027, along with AB 3030 (2024) and SB 1120 (2024). How the law applies to a particular provider or situation may depend on regulations and guidance not yet issued. At Home VA Staffing is a non-medical home care agency and is not affiliated with the State of California, the Medical Board of California, or any health plan.