Illinois told AI it cannot make the therapy decisions

The conversation

On 1 August 2025, Illinois enacted House Bill 1806 as Public Act 104-0054, the Wellness and Oversight for Psychological Resources Act. The bill's own status record shows the Governor approved it and that it took effect the same day. The Act's purpose clause states it exists to safeguard people seeking therapy from "unlicensed or unqualified providers, including unregulated artificial intelligence systems." It does not describe one incident; it describes a category of conversation the legislature decided needs a licensed human on the other end.

What the documents show

The statute's text draws a specific line. Section 20 bars any individual, corporation or entity from providing or advertising therapy or psychotherapy services, including through internet-based artificial intelligence, unless a licensed professional conducts them. A licensed professional who does use AI may not let it make independent therapeutic decisions, directly interact with clients in therapeutic communication, generate treatment plans without review, or detect emotions or mental states. Section 15 separately defines a permitted use for administrative support, such as scheduling and billing, and supplementary support, such as note-taking, performed under the licensed professional's full responsibility. Where a session is recorded or transcribed and AI assists with that supplementary work, the patient must be told in writing that AI is used and why, and must consent. Violations carry a civil penalty of up to $10,000 per violation, enforced by the Illinois Department of Financial and Professional Regulation after a hearing, as the same enacted text states.

The system boundary

The Act's own architecture is the boundary: a licensed human professional must remain the party who diagnoses, treats, or holds a therapeutic conversation, while software may only support that professional's non-clinical work. The statute exempts religious counseling, peer support, and public self-help materials that do not claim to be therapy. This is narrower than a ban on AI near mental health; it is a rule about who may conduct therapeutic communication and who may only assist the person who does.

Where it fails

The Act does not specify how a patient, a platform, or a regulator distinguishes permitted supplementary support from an AI system drifting into therapeutic communication once it summarizes session notes or flags a trend, nor does its text define what counts as detecting an emotion in a product marketed for general wellness rather than diagnosis. Builders operating in Illinois should watch for that boundary in their own product design rather than assume a generic AI disclaimer satisfies it.

• Does our system ever produce output a reasonable user could read as a therapeutic recommendation, even when marketed as supplementary support?
• If a session is recorded or transcribed, have we built the written notice and consent step the statute requires before AI touches that record?
• Who at our organization is the licensed professional accountable for every AI-assisted interaction, and can we show they reviewed it?

This is an editorial checklist, not legal advice; the statute's own enforcement runs through the Illinois Department of Financial and Professional Regulation, and its text is the only authority a builder should rely on for compliance.