California bill would ban AI chatbots from advertising as therapy

California legislators are moving to regulate artificial intelligence in mental healthcare as millions of people—including one in eight teens and young adults—turn to chatbots for emotional support and advice.

Senate Bill 903, authored by State Sen. Steve Padilla, would prohibit companies from advertising chatbots as therapy and require licensed professionals to review any therapeutic decisions made by AI systems. The legislation would also mandate that healthcare providers disclose AI use and obtain patient consent before deploying tools to record sessions or triage mental health cases.

According to reporting first published by the Los Angeles Times, the bill arrives as federal courts in California hear wrongful death lawsuits accusing chatbot makers of contributing to users' suicides. OpenAI disclosed last fall that approximately 1.2 million users per week share suicidal thoughts on ChatGPT.

Why it matters

The legislation reflects a collision between urgent mental health needs and largely unregulated AI deployment. While chatbots offer 24/7 availability and no insurance barriers, they lack the clinical judgment to detect serious conditions or manage crisis situations. The bill attempts to draw boundaries around what constitutes legitimate clinical support versus potentially harmful substitutes—a distinction that could determine whether AI expands access to care or displaces qualified practitioners.

Two-front regulatory approach

The bill targets both consumer-facing chatbot apps and AI tools used within clinical settings. Platforms like Character.ai host chatbot "psychologists" that offer empathy and advice with only small-print disclaimers that responses should be treated as fiction. These general chatbots operate with no professional oversight despite their popularity for companionship and informal therapy.

In formal healthcare environments, the National Union of Healthcare Workers has filed a complaint against Kaiser Permanente alleging its e-visit screening tool generates care recommendations too quickly for human review. The union argues such automated triage violates existing state law and puts patients at risk.

Benjamin Eichert, NUHW's director of public policy, told the Times that intake and screening represent the fastest-growing area of AI implementation in behavioral health. The bill's triage provisions directly address tools already deployed without regulatory guardrails.

Industry pushback on clinical restrictions

Professional associations representing psychologists, therapists, and counselors back the legislation, as does NUHW. Le Ondra Clark Harvey, CEO of the California Behavioral Health Association, emphasized that the difference between licensed clinicians and automated responses "can be life altering."

TechNet, representing tech-sector companies, opposes the bill's restrictions on clinical AI use. Robert Boykin, TechNet's California executive director, argued the legislation could limit tools that help address the state's behavioral healthcare worker shortage by creating "a clinician bottleneck in front of the intake and screening tools that help patients reach care faster."

Regulatory momentum builds

Illinois became the first state to ban AI in therapy services last year, with several states following with similar restrictions. The California bill awaits a vote in the Assembly's fiscal committee.

Dr. Jodi Halpern, a UC Berkeley professor of bioethics and medical humanities, noted that people turn to chatbots because of unmet needs but companies then rely on business models built around maximizing engagement through manipulative tactics like praise and agreement.

State regulators are separately investigating NUHW's complaint against Kaiser. In response, Kaiser stated its e-visit tool "does not use AI to diagnose patients, make clinical decisions, or determine medical necessity," though the union maintains the tool's inner workings remain opaque.

These details were first reported by Ana B. Ibarra for CalMatters and published in the Los Angeles Times.