A California bill moving through the Legislature would prohibit companies from advertising artificial intelligence chatbots as therapy, and would set new rules for how licensed providers use AI in mental health care.
Senate Bill 903, authored by Senator Steve Padilla of San Diego, passed the Senate unanimously in May and cleared two Assembly policy committees before reaching the Assembly Appropriations Committee, where it was placed on the suspense file this month. It arrives as chatbot use has expanded rapidly and as wrongful death lawsuits, several in California federal courts, allege that AI chatbot makers contributed to users' suicides.
The bill is narrower than headlines about banning AI therapy suggest. It does not outlaw the tools. It targets what companies may call them and how licensed clinicians may deploy them.
The Restrictions Fall on Advertising and Clinical Decision-Making
As introduced, the measure would bar individuals and companies, including those using AI, from offering or advertising therapy or psychotherapy in California unless a licensed professional is responsible for the care. The amended text frames the advertising prohibition around companion chatbots, barring a company from advertising or purporting to offer psychotherapy services delivered through such a chatbot, including by claiming it is a therapist or provides therapy.
The bill would also prohibit AI from independently interacting with clients, making therapeutic decisions, detecting emotions, or generating treatment plans without review and approval by a licensed professional. Licensed professionals could use AI for limited administrative or supplementary support, such as drafting logistical communications that contain no therapeutic advice. Providers would need clear written disclosure and affirmative patient consent before using AI to record or transcribe therapy sessions, or to triage mental health care.
Additional provisions would require confidentiality of therapy records and authorize the Department of Consumer Affairs and the relevant licensing boards to investigate violations and impose civil penalties of up to $10,000 per violation.
The distinction between a marketing restriction and a product ban matters. A general-purpose chatbot a person chooses to talk to about stress would not be outlawed. What would change is a company's ability to present that product as therapy, and a health system's ability to let an algorithm make a clinical determination without a licensed reviewer.
The Evidence Shows Both Measurable Benefit and Documented Failure
Any honest account of AI in mental health has to hold two findings at once, and the debate over SB 903 tends to feature only one at a time.
On the benefit side, there is clinical evidence that structured, purpose-built therapeutic chatbots can help. A Dartmouth randomized trial of Therabot, published in NEJM AI, enrolled 210 adults with clinically significant symptoms of major depressive disorder, generalized anxiety disorder, or high risk for a feeding and eating disorder. Compared with a waitlist control that received no app access, participants using the chatbot for four weeks reported symptom reductions averaging 51 percent for depression, 31 percent for anxiety, and 19 percent for body image and weight concerns. The researchers cautioned that no generative AI agent is ready to operate autonomously in mental health. Access is the strongest argument in the tools' favor: Padilla's office notes that nearly a third of California residents live in an area with an insufficient ratio of providers to patients.
On the risk side, the failures cluster in exactly the situations where stakes are highest. After four months of testing leading chatbots, Common Sense Media and Stanford Medicine's Brainstorm Lab for Mental Health Innovation recommended that teenagers not use AI chatbots for mental health advice or emotional support, finding that the products missed warning signs and steered toward continued engagement rather than professional help. Their research also found that three in four teens use AI for companionship, which includes emotional and mental health conversations.
Le Ondra Clark Harvey, chief executive of the California Behavioral Health Association, told lawmakers that without clear safeguards, chatbots can provide inaccurate information or mishandle crisis situations. "The difference between a licensed clinician and an automated response is not technical," she said, adding that it can be life-altering.
The bill is co-sponsored by the California Psychological Association, the California Association of Marriage and Family Therapists and the California Behavioral Health Association, and is backed by the National Union of Healthcare Workers.
Industry Argues the Bill Slows Access Rather Than Improving It
Opposition has focused less on the advertising ban than on the restrictions covering AI use inside formal clinical settings.
TechNet, which represents technology companies, has told committees it opposes the bill unless amended. Robert Boykin, testifying for the group, said it agrees that AI should not be advertised as replacing licensed professionals, but argued the measure would significantly restrict beneficial uses of AI in health care, including widely used tools such as patient check-ins, journaling support and workflow assistance that help maintain continuity of care.
That objection reaches a real tension. Intake questionnaires, triage algorithms and symptom screeners are already common in large systems, and requiring licensed review of each output carries staffing and cost implications. A dispute involving a labor union and Kaiser Permanente over an electronic visit tool illustrates the ambiguity: the union argues care recommendations generated within seconds are unlikely to have been reviewed by a licensed professional, while Kaiser has said its e-visit tool does not use AI to diagnose patients, make clinical decisions or determine medical necessity.
The Practical Effect on Households and Patients
If enacted, the most visible consumer effect would be in app stores and advertising rather than in what technology exists. Products would need to stop describing themselves as therapy or psychotherapy unless a licensed professional is accountable for the care.
For patients in treatment, the practical change would be consent. A therapist wanting to use AI to record or transcribe a session would need written disclosure and affirmative agreement, which is a right patients can ask about now regardless of whether the bill passes.
Nothing in the measure requires anyone to stop using a chatbot, and no one should stop prescribed treatment or cancel therapy based on legislative news. People who find general AI tools useful for journaling, organizing thoughts, or between-session reflection can continue, while recognizing that these products are not designed for crisis response and are not a substitute for clinical care.
Anyone in crisis should contact a person rather than a program. The 988 Suicide and Crisis Lifeline is available by call or text in the United States. Californians without insurance can access county behavioral health services, and federally qualified health centers offer sliding-scale mental health visits.
Several outcomes remain unknown. The bill has not passed, its final language may change, and Governor Gavin Newsom has not indicated a position. How the Department of Consumer Affairs would enforce the provisions, and whether the rules would reach clinical triage tools already in use, are unresolved. Padilla previously authored SB 243, which established earlier safeguards on AI chatbots, and Illinois enacted its own law last year restricting AI-delivered therapy. MedicalDaily will report on the committee action and any amendments.
Key Questions Answered
Would SB 903 ban AI chatbots in California? No. It would prohibit offering or advertising therapy or psychotherapy unless a licensed professional is responsible for the care, and restrict AI from making therapeutic decisions without professional review. The tools themselves would remain available.
What would change for someone already in therapy? A provider would need written disclosure and the patient's affirmative consent before using AI to record or transcribe sessions, or to triage mental health care.
Does evidence support chatbots for mental health at all? A randomized trial of a purpose-built therapeutic chatbot showed symptom reductions against a waitlist control, and access advantages are real. Testing has also documented failures in crisis situations, which is why researchers advise against teen use for emotional support.
Who supports and who opposes the bill? Professional associations representing psychologists, therapists and counselors co-sponsor it, along with a health care workers union. TechNet, representing technology companies, opposes it unless amended, arguing it could restrict beneficial clinical uses.
Where does the bill stand? It cleared the Senate and two Assembly committees and now sits on the Assembly Appropriations suspense file. It has not passed, and its language could change.
Is a chatbot safe to use during a mental health crisis? No. These products are not designed for crisis response. Anyone in crisis should contact the 988 Suicide and Crisis Lifeline by call or text, or seek emergency care.
Are other states doing something similar? Yes. Illinois enacted a law restricting AI-delivered therapy and setting rules for clinicians' use of AI, and California previously enacted SB 243, which set earlier safeguards on AI chatbots.
