Governor Newsom vetoes California SB 903, the bill to ban AI companion chatbots in psychotherapy

Key takeaways

• The Governor returned the bill without his signature on 30 September 2026.
• He said the bill "is overly broad and would drastically limit a clinician's use of tools that benefit the delivery of care today".
• He said the bill's definition of "psychotherapy services" would capture general-purpose AI systems not even structured or deployed to deliver that care.
• He supported the author's intent on guardrails and encouraged revisiting the issue next year.

What did Governor Newsom do with SB 903, and is it law?

He returned it unsigned. The two-page letter, dated 30 September 2026 and addressed to the Members of the California State Senate, opens: "I am returning Senate Bill 903 without my signature." It closes: "For these reasons, I cannot sign this bill." The message is a statement of the Governor's reasons, not a rule, and it sets no duty for clinicians or technology providers.

What would the bill have done?

The message describes two things. First, the bill "would prohibit the use of companion chatbots in the provision of psychotherapy services". Second, it "would create restrictions on the use of artificial intelligence (AI) by licensed professionals who provide psychotherapy services". The message gives no bill text, section numbers, penalties or effective date, so those details are outside what this article can state. Readers who need the exact drafting would need to read the bill itself. For comparable proposals and laws elsewhere, see our pieces on the Colorado psychotherapy AI restrictions and the Rhode Island mental health chatbot laws, which are separate instruments and do not apply in California.

Why did the Governor say he could not sign it?

He gave two objections and one statement of support. On support, he wrote that he supports "the author's intent to establish guardrails for the use of AI in healthcare settings" and referred to "recently signed legislation to mitigate bias". On the first objection, he said: "this bill is overly broad and would drastically limit a clinician's use of tools that benefit the delivery of care today", adding that this includes "requiring routine screening determinations to receive direct approval". On the second, he said the bill's broad definition of "psychotherapy services" is "poorly crafted" and "would capture general-purpose AI systems not even structured or deployed to deliver that care". He closed by encouraging "the author and stakeholders to revisit this issue next year" to find a solution that ensures professional judgment is not replaced while still allowing licensed professionals to use tools that enhance their delivery of quality care. That is the Governor's stated test in the message.

What are the key dates and parties?

What we did not verify

What we opened: the Governor's veto message PDF (two pages) from gov.ca.gov, on 25 October 2026 UTC. The PDF is a scan with no text layer, so we read it as page images.

What we did not open: the text of SB 903, its legislative history, votes, analyses, the author's or supporters' statements, or the bias-related legislation the message mentions. We did not check the bill's current status in the Legislature, including any later vote or successor bill.

What we refuse to claim: that SB 903 is or will become law, or that no later legislative action has occurred, what its sections provide beyond the message's two-sentence description, who supported or opposed it, or that it will return in 2027. The message only encourages revisiting the issue next year.

Related briefings

• AI Regulation News hub
• North America AI regulation tracker
• Attorneys and AI regulation
• Colorado HB 1195 psychotherapy AI restrictions
• Rhode Island AI mental health chatbot laws
• California SB 243 companion chatbot safety
• Nevada AB 406 AI mental and behavioral health ban