California therapy sessions get a human-only rule

In California, Sen. Steve Padilla’s SB 903 asks a simple but loaded question: when does artificial intelligence stop being a tool and start becoming part of treatment? The bill would prohibit a licensed professional from using AI to assist in providing supplementary support in therapy or psychotherapy when that support is tied to a client’s therapeutic services.

For patients, that matters because therapy depends on trust, judgment and the sense that another person is carrying the clinical responsibility. SB 903 would keep that line sharper, even as software keeps spreading into health care.

A harder line in the therapy room

The proposal is not written as a ban on every piece of software a mental health office might use. Its target is narrower: AI used as supplementary support in the actual delivery of therapy or psychotherapy. In other words, the bill would keep treatment human-led, not machine-assisted, when it comes to the clinical relationship itself.

That approach builds on California’s existing AI rules, which already require some AI-generated patient communications from a health facility, clinic, physician’s office or group practice to include a disclaimer and instructions for reaching a human provider, employee or other appropriate person. SB 903 would move the boundary deeper into the care setting, not just the paperwork around it.

What patients would notice

For people in treatment, the practical effect is a cleaner answer to a question that can be hard to see from the outside: who, exactly, is helping shape the care? The bill would make it harder for AI to sit inside that role, even in a supporting capacity tied to therapy or psychotherapy.

Available vote records show the bill advanced without recorded no votes.