If enacted, the bill would limit providers’ use of AI in psychotherapy, require consent for certain uses, and bar advertising chatbots as therapists.
Providers could use AI for therapy-related office tasks and other support, but licensed professionals would remain responsible for clinical decisions and communications.
Before using AI to record or transcribe therapy sessions or screen clients, providers would have to explain its purpose and get consent. Refusing would not affect the client’s right to care.
A person or organization could not advertise a companion chatbot as a therapist or claim that it provides therapy.
AI could not make therapy decisions, assess clients, or detect emotions without a licensed professional’s review and approval. Limited exceptions apply, including for some direct communication tools cleared for that use.
AI use in therapy records would have to follow state medical privacy rules. Companies could not share, sell, store, or train models on therapy data in ways that violate applicable law.
Providers could use AI only for office tasks or other support that does not involve therapeutic communication. This covers services to diagnose or treat mental health or substance use disorders; religious counseling and peer support are excluded. Supplementary support can include preparing records and tracking progress, but licensed professionals remain responsible for clinical decisions and communications. Religious counseling, peer support, public self-help materials, training or simulation, and certain research are outside this chapter.
Providers could not use AI to record or transcribe therapy communications or sessions, or to screen clients, without notice and consent. The patient or client, or their legally authorized representative, may give consent. The provider must explain that AI will be used and its specific purpose. Consent must be informed, voluntary, documented, and revocable. A client who refuses consent does not lose any right to care.
People and organizations could not advertise companion chatbots as providing psychotherapy. The prohibition includes claiming that a companion chatbot is a therapist or provides therapy.
AI could not perform specified clinical tasks without review and approval by a licensed professional. These rules apply when AI is used in connection with psychotherapy services or triage or screening. Covered tasks include making therapy decisions, generating recommendations, assessment results, diagnoses, or treatment plans; detecting emotions or mental states; and triage or screening. AI could directly communicate with clients without that review only if it is approved or cleared by the U.S. Food and Drug Administration for that use and complies with federal health privacy law. The bill allows standardized questionnaires and appointment tasks under stated limits.
AI use in psychotherapy records would have to follow state medical privacy rules, and companies could not handle therapy data in ways that violate applicable law. This concerns patient or client records and data obtained from psychotherapy. The restriction covers sharing, selling, storing, or training models on psychotherapy data only when that handling is inconsistent with applicable law.
No hearings or votes scheduled.