Ahead of Jan. 1, 2027, the authors examine how practices can preserve independent clinical judgment when physicians review AI recommendations in batches.
LOS ANGELES, CA, UNITED STATES, November 11, 2026 /EINPresswire.com/ — Starting Jan. 1, 2027, California practices must take reasonable steps to ensure that physicians retain the ability to exercise independent judgment whenever AI informs patient care. The law does not say how much review is enough. Practices that approve AI recommendations in high volume should confirm before then that each physician sign-off reflects an actual clinical decision.
Gov. Gavin Newsom signed AB 1979, by Assemblymember Mia Bonta (D-Oakland), on Sept. 30. The California Nurses Association sponsored the bill, which adds Section 22758.5 to the Business and Professions Code. It applies to health facilities, clinics, physician offices and group practice offices. Newsom also signed SB 503, which targets bias in clinical decision support tools, and vetoed two other health care AI bills, AB 2575 and SB 903.
“Technology should not replace human decision-making in health care,” Bonta said in a statement.
Physicians are adopting these tools quickly. In the American Medical Association’s 2026 Physician Survey on Augmented Intelligence, 72% of physicians who answered the relevant question reported incorporating at least one AI use case into practice, compared with 38% in 2023. Another 9% were uncertain which AI tools their practice offered. Yet 88% expressed at least some concern about AI-related skill loss.
If your practice uses AI to triage requests, screen for contraindications or draft treatment recommendations, compliance with AB 1979 turns on whether the physician had a meaningful opportunity to reach a different decision, and a signature alone may not show that.
What the law requires
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 the timing of care, diagnosis or treatment.
Practices also may not use an AI tool to “independently perform any clinical function” that the law requires a licensed person to perform. Nor may they use AI to “direct, guide, supervise, or instruct unlicensed personnel” in those functions.
The law exempts documentation and communication functions that do not require professional judgment, such as automated notices about record updates, appointment reminders and tools that help patients find information. It also excludes trainees using a tool in a supervised course of study. Violations by physicians fall within the jurisdiction of the Medical Board of California or the Osteopathic Medical Board of California. Where a violation amounts to unlicensed practice, a licensing board may seek an injunction or restraining order.
AB 1979 permits AI-assisted decision support. Technology may inform clinical decisions, provided professional judgment stays with licensed health care providers.
The enacted version is narrower than earlier drafts. An analysis by the California Board of Behavioral Sciences shows lawmakers deleted a requirement that no clinical decision be based solely on an AI system’s output. They also revised language requiring a licensee to exercise independent judgment when reviewing and approving AI clinical decisions. As enacted, the duty falls on the practice, which must take reasonable steps so the clinician retains the ability to exercise that judgment.
As a result, the law does not expressly address batch approvals or set a minimum review time per patient. Nor does it make existing AI-assisted workflows unlawful per se on Jan. 1, 2027.
The batch approval problem
Consider a telehealth platform that treats straightforward complaints, such as suspected uncomplicated urinary tract infections or sinus infections. The platform collects symptoms, history, medications and contraindications. An AI tool flags cases that appear to meet clinical criteria and presents recommendations for physician review.
Suppose the platform receives hundreds of requests per hour. Can a physician approve a group of apparently low-risk cases together, or must each recommendation receive individual review? At what point does approval become a rubber stamp?
Our concern is that health care organizations will treat a physician’s formal sign-off on an AI-generated recommendation as sufficient without asking whether the physician has a meaningful opportunity to evaluate the underlying information and reach an independent decision. Having a human in the process does not by itself establish meaningful oversight.
Because AB 1979 places duties on health facilities and practice settings, a medical group delivering care through a telehealth platform should not assume that the technology vendor alone is responsible for compliance.
We are equally wary of reading the law too strictly. If physician review requirements block responsible use of automated screening and decision support, practices may face higher costs, slower processing and fewer opportunities to expand lower-cost care. The task for regulators and practices is to set workable standards for how physicians and AI tools operate together.
Oversight also depends on training. In our 2026 book, Digital Diagnosis: The AI Revolution in Healthcare and the Future of Regulation, we discuss the risk of treating AI output as infallible. Physicians need to understand when to question a system’s recommendations rather than simply defer to them.
Chatbots and patient disclosures
AB 1979 also brings some consumer health chatbots under the Confidentiality of Medical Information Act (CMIA). A business that offers a health care chatbot to consumers for managing their information, or for diagnosing, treating or managing a medical condition, is treated as a health care provider for CMIA purposes.
These requirements build on AB 3030, in effect since Jan. 1, 2025. When a practice uses generative AI to draft patient communications containing clinical information, the message must include a disclaimer stating that it was generated by AI and instructions for reaching a human. The requirement does not apply if a licensed or certified health care provider reads and reviews the message before it is sent.
Steps to take before Jan. 1, 2027
Practices should review how AI is used in their clinical workflows:
• Sort your tools. Separate administrative and documentation tools from those that influence diagnosis, treatment or other licensed functions. The exemption covers only uses that do not require professional judgment.
• Map the review. For each clinical tool, check what information the physician sees, how much time the workflow allows and whether rejecting a recommendation is easy. Identify productivity targets that encourage approvals without meaningful evaluation.
• Set rules for grouped review. If physicians review cases in batches, define which cases qualify and which findings trigger individual review. Put those rules in writing.
• Document overrides. When a physician rejects an AI recommendation, record the decision and the reason. Documented overrides can help show how independent professional judgment is exercised and may also reveal problems with a tool.
• Train for limits. Teach physicians the limitations of each tool and when to disregard its output.
• Check patient messages. Confirm that AI-drafted clinical messages carry the AB 3030 disclaimer or get a documented clinician review, and that any consumer chatbot complies with the CMIA.
Do not treat an electronic signature as conclusive evidence of review. A sign-off alone does not show how the physician evaluated the underlying information or exercised independent professional judgment.
Until regulators provide further guidance, records showing how physicians review, challenge and document AI output can help demonstrate that professional judgment remained central to patient care.
AB 1979 makes clear that AI cannot independently assume responsibilities reserved for licensed health care professionals.
How much physician involvement is enough remains an open question.
Harry Nelson and Nicholas Merkin
https://www.healthcarecompliance.ai/
+1 3104697260
info@healthcarecompliance.ai
Visit us on social media:
LinkedIn
Instagram
YouTube
Legal Disclaimer:
EIN Presswire provides this news content “as is” without warranty of any kind. We do not accept any responsibility or liability
for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this
article. If you have any complaints or copyright issues related to this article, kindly contact the author above.
Media gallery
