SB 903 would ban companion chatbots from providing psychotherapy.
AI may support clinicians but cannot make independent therapeutic decisions.
The bill passed the Assembly Appropriations Committee unanimously, 13‑0, on August 13.
California is moving closer to prohibiting AI chatbots from posing as therapists and making mental health decisions without human oversight.
Introduced on January 21, Senate Bill 903 cleared the Assembly Appropriations Committee with a unanimous 13‑0 vote on Thursday and was ordered for a third reading. A date for the Assembly vote has not yet been set.
Myriad: When will OpenAI release GPT-6? Click to make your prediction.
“AI algorithms cannot replace human therapists, who possess skills and training that AI cannot replicate,” said California State Senator Steve Padilla, author of the bill, in a statement when introducing the measure. “We must enact safeguards to prevent AI algorithms from being deployed in ways that could harm patients.”
Also known as the Wellness and Oversight for Psychological Resources Act, SB 903 would restrict AI’s role in psychotherapy to supportive tasks, prohibit chatbots from posing as therapists, and mandate clinician oversight and patient consent. The bill would also safeguard therapy data while exempting peer support, religious counseling, research, and federally approved medical tools.
The Senate approved SB 903 by a 39‑0 vote in May. The bill then cleared two Assembly policy committees with votes of 17‑0 and 14‑1. If the Assembly adopts the amended version, it will return to the Senate for concurrence before heading to California Governor Gavin Newsom for signature.
Critics, including Robert Boykin, executive director of the technology industry group TechNet, warned that mandating clinician involvement could delay care, given the shortages of behavioral health workers across many counties.
“With every California county experiencing a shortage of behavioral health workers, SB 903 creates a clinician bottleneck ahead of the intake and screening tools that help patients access care more quickly,” Boykin said in a statement to the nonprofit news outlet CalMatters.
The development follows rising concerns about AI’s impact on mental health.
In June, an American Psychological Association survey revealed that 77% of psychologists had patients who used AI for mental health purposes. Meanwhile, researchers from 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.
The researchers noted that the tendency of AI chatbots to agree with users resembles social media echo chambers and, at its extreme, creates an ‘echo chamber of one,’ where the corrective benefit of real‑life social interaction is missing.
