Board backs new consumer-protection rules for AI in therapy and clinical decision tools

The California Board of Psychology on May 15 voted to support several bills addressing the use of artificial intelligence in clinical care and advertising, saying new rules are needed to protect patients and preserve clinician judgment. The board backed SB903, which would require clinicians to obtain clear patient consent when AI is used and prohibit AI from making independent therapeutic decisions, and signaled support for civil-liability and disclosure measures in AB2575 and AB1979.

Board members and staff framed the measures as consumer-protection tools aimed at keeping clinical judgment with licensed practitioners. Enforcement Committee Chair Mr. Fu summarized the committee's view that SB903 'is consistent with our desires for consumer protection' and stressed that AI should not replace licensed clinical decision-making. Executive Officer John Burke told the board staff had reviewed April amendments that clarified enforcement responsibilities and that the Department of Consumer Affairs would have investigatory authority under SB903 as drafted.

The board also considered AB2575, a broader health-sector measure that would require written notice about AI tools used for clinical decision support, and AB1979, which would require notices if generative AI produces patient-facing clinical communications and ensure human contact information is available. Enforcement manager Mr. Phillips told the board that if these measures pass the board should expect increased complaint volume and higher investigative costs for AI-related cases, noting that expert reviews and Division of Investigation work can add thousands to case costs.

Why it matters: Boards and regulators are confronting rapid adoption of AI tools in health care. The measures the board supported aim to preserve patient protections, clarify liability, and ensure clinicians retain professional authority when technology is used. For now the board is monitoring related bills (SB1146, SB1159) that would require AI advertising disclosures and clarify that AI systems are not "persons" for public-participation laws. The board took formal votes to record support where committees recommended positions and will follow bill amendments as they move through the Legislature.

The board recorded formal support votes on legislation during the May 15 meeting and will continue tracking amendments and implementation issues should the bills become law.