Colorado HB26-1195: a therapist may use AI, but may not let it be the therapist

Colorado HB26-1195 addresses a question most AI statutes leave open: not whether a professional may use AI, but what remains irreducibly the professional's job. It was signed by the Governor on 3 June 2026 as Chapter 358 of the Session Laws, and takes effect on 12 August 2026.

The central restriction is about presence. A regulated psychotherapy professional may not allow an AI system to interact with a client without synchronous, real-time interaction between the professional, the AI system and the client. An AI system may not generate a treatment plan that is not reviewed by the professional, and AI-generated therapeutic recommendations require professional review and approval. The professional remains responsible for reviewing what the system produces — responsibility does not transfer to the tool.

A second set of provisions targets misrepresentation. A professional may not present AI outputs as equivalent to the services of a licensed professional, and may not falsely claim that data handled by an AI system carries the confidentiality protections that attach to therapist-client privilege. That second point matters commercially: confidentiality is frequently the implied promise in mental-health AI marketing, and the law makes the false version of that promise actionable.

Clients must be told about the AI restrictions at initial contact. Where sessions are recorded or transcribed using AI, advance disclosure and written informed consent are required.

Enforcement runs on two tracks. Regulated professionals may be disciplined by the appropriate licensing board, and violations may also constitute unfair or deceptive trade practices under the Colorado Consumer Protection Act — which extends exposure beyond licensed clinicians to persons marketing AI systems as mental health tools.