California has moved one step closer to barring AI chatbots from posing as therapists and making mental health decisions without human oversight, after a key Assembly committee advanced Senate Bill 903.

Introduced on January 21, SB 903 cleared the Assembly Appropriations Committee in a 13-0 vote on Thursday, August 13, and was ordered to a third reading. No date has been announced for the full Assembly vote.

"AI algorithms are not fit to take over the job of human therapists, who have skills and training that AI is incapable of replicating," bill author California State Senator Steve Padilla said in a statement when introducing the measure. "We must act and place guardrails that prevent AI algorithms from being deployed in a way that is potentially harmful to patients."

Also known as the Wellness and Oversight for Psychological Resources Act, SB 903 would limit AI in psychotherapy to support tasks, bar companion chatbots from posing as therapists, and require clinician oversight and patient consent. AI could assist clinicians but could not make independent therapeutic decisions. The bill would also protect therapy data while exempting peer support, religious counseling, research, and federally approved medical tools.

The legislation has advanced with broad support. The Senate passed SB 903 by a 39-0 vote in May, and it later cleared two Assembly policy committees by votes of 17-0 and 14-1. If the Assembly approves the amended bill, it would return to the Senate for concurrence before going to California Governor Gavin Newsom for signing. Newsom signed a first-in-the-nation package of 18 AI safety laws in September 2024 but vetoed the broader frontier-model safety bill SB 1047 that same month, and in the absence of comprehensive federal AI rules California has remained a leading state arena for AI oversight.

Critics, including Robert Boykin, executive director of the technology industry group TechNet, argue that requiring clinician involvement could delay care as counties face shortages of behavioral healthcare workers.

"At a time when every county in California faces a shortage of behavioral healthcare workers, SB 903 still puts a clinician bottleneck in front of the intake and screening tools that help patients reach care faster," Boykin told the non-profit news outlet CalMatters in a statement.

The bill's progress comes amid growing concern over AI's mental health effects. In June, an American Psychological Association survey found that 77% of psychologists had patients who used AI for mental health purposes, while researchers at King's College London, the Protestant University of Applied Sciences, and the City University of New York found that chatbots can reinforce delusions through excessive agreement and personalized responses. Lawmakers' scrutiny of companion chatbots has also been sharpened by lawsuits against Character.AI, whose roleplay bots are widely used by teenagers, alleging that the product contributed to minors' deaths by suicide; the company has since added safety features for users under 18.

"The tendency of AI chatbots to agree with user opinions has been likened to social media echo chambers and, in its most extreme form, to an 'echo chamber of one,' where the positive corrective influence of real-life social interactions is absent," the researchers wrote.