Governor Newsom vetoed California AB 2575 on patient-care workers overriding clinical decision support AI

Key takeaways

• The message is dated 30 September 2026 and opens "I am returning Assembly Bill 2575 without my signature."
• The bill, as the message describes it, would have let a patient-care worker override clinical decision support output when the worker judged it necessary to meet the standard of care or comply with law.
• The Governor said the anti-retaliation provisions "do not protect workers as intended" and would put the Labor Commissioner in the position of adjudicating the standard of care.
• We read only the veto message, not the bill text.

What did the Governor do with AB 2575, and is it law?

He returned it unsigned. The message is addressed "To the Members of the California State Assembly", is dated 30 September 2026 and begins "I am returning Assembly Bill 2575 without my signature." It closes: "For these reasons, I cannot sign this bill." The message does not enact the bill. We did not review the legislative record, so we cannot say whether anything has happened since.

The message is a two-page scanned letter, which we read as an image. It is a statement of reasons, not a regulation, and it sets no duty for employers, hospitals or AI vendors.

What would the bill have done, according to the veto message?

The message describes two things. First, the bill "would authorize a worker providing direct patient care to override the output of a clinical decision support system if, in the judgment of the worker acting within their scope of practice, an override is necessary to meet the applicable standard of care or comply with applicable law." Second, it "would prohibit employers from retaliating against workers providing direct patient care solely for relying on or overriding a clinical decision support system."

That is the Governor's summary. We did not open the bill, so definitions such as who counts as a worker providing direct patient care, or what counts as a clinical decision support system, are not described here. Separate California measures on physician judgment and clinical decision support are covered in our article on AB 1979 and SB 503. We did not check how, if at all, this message bears on them.

Why did the Governor return the bill?

He gave two reasons, both about enforcement. He wrote: "While I commend the author's commitment" to safe patient care and professional judgment, "it is essential that new laws in this area be practical and enforceable in their implementation to achieve their intended goals."

The first reason concerns the wording of the retaliation rule. By prohibiting adverse actions based "solely" on an override of, or reliance on, a clinical decision support system, the message says the bill "ties the Labor Commissioner's hands in establishing a violation", a higher bar than other retaliation protections, which "simply prohibit action because of a protected activity."

The second, which he introduces with "More importantly,", is that linking anti-retaliation protection to a scope-of-practice determination "puts the Labor Commissioner in the position of adjudicating the standard of care for patients." The message says the Labor Commissioner lacks the medical expertise to do that and so "cannot effectively and fairly enforce the bill's anti-retaliation provisions as the author intends."

What does the veto mean for counsel and clinical AI vendors?

The message sets no rule, so the question stays with contracts, hospital policy and whatever existing law applies. For health-system counsel, the message gives nothing to cite as a statutory protection or barrier when a clinician overrides an AI recommendation. For clinical AI vendors, the message does not turn the bill's override concept into a product requirement. For labor lawyers, the message shows the Governor's stated concern about two drafting choices, a "solely" causation standard and a scope-of-practice trigger, which drafters of any future bill could consider.

That last point is our inference. The message does not say a revised bill is expected or what it would contain.

What should health-system counsel, AI vendors and labor lawyers do now?

• Do not describe AB 2575 to clients as a requirement based on this message. Cite it as a veto message and confirm current status first.
• Confirm that your own policy says who may override clinical decision support output, who is told, and how the override is recorded; AB 2575 as vetoed would not supply that.
• Review how your anti-retaliation and escalation policies treat a clinician who relies on or overrides AI output, using existing employment and healthcare law as the baseline.
• Check the legislative record for later action or a reintroduced bill and read its text before advising, because the message does not preview one.

What we did not verify

What we opened: the Governor's veto message on AB 2575, a two-page scanned PDF, read as an image on 2026-10-25.

What we did not open: the bill text and any amendments, the legislative analyses and vote record, any statement from the author, and any later bill. The bill description above is the Governor's summary only.

What we refuse to claim: that AB 2575 or any part of it is in force or has since been acted on, that the message changes any existing duty, that a clinician is or is not protected under current law, or that the Governor will sign a revised bill.

Related briefings

• AI Regulation News hub
• North America AI regulation tracker
• Attorneys and AI regulation
• California AB 1979 and SB 503 on clinical decision support and physician judgment
• California SB 1120 on AI utilization review and physician oversight
• AI in healthcare compliance
• California SB 947 on automated decision systems in discipline and discharge