What a Behavioral Health Group Can Automate Under Illinois's AI Therapy Law

If you run a behavioral health group in Illinois, artificial intelligence can take over much of your front office—scheduling, billing, insurance follow-up, records—but a 2025 state law now draws a hard line at the clinical work itself. The practical question for a practice administrator is no longer whether to use AI, but which tasks the law lets you automate, which ones require written patient consent first, and which a vendor can never deliver no matter how good the demo looks.

What Illinois changed in 2025

Illinois enacted the Wellness and Oversight for Psychological Resources Act in August 2025, and it took effect on enactment. The Illinois Department of Financial and Professional Regulation (IDFPR) states the core rule plainly: AI may not provide therapy or make therapeutic decisions, but it may be used for administrative and supplementary support of licensed behavioral health professionals. The Act reaches any individual, corporation, or other entity that provides, advertises, or offers therapy in Illinois, including internet-based AI systems and mental-health chatbots.

The enforcement teeth matter for a buying decision. IDFPR enforces the Act and can impose civil penalties of up to $10,000 per violation after a hearing. A tool you buy to save front-desk hours can also create regulatory exposure if it is pointed at the wrong task, so the category a workflow falls into is not a compliance footnote—it is the buying decision.

See the administrative workflows Clairvance helps healthcare practices connect.

What you can automate today

Administrative support is permitted without separate patient consent because it does not involve therapeutic communication. Holland & Knight's reading of the statute groups the clearly allowed work into clerical tasks:

• Managing appointment scheduling, reminders, waitlists, and cancellations.
• Processing billing and submitting or following up on insurance claims.
• Drafting general communications about therapy logistics—directions, intake paperwork requests, appointment confirmations—that carry no therapeutic advice.

These are the workflows where a Chicago practice loses the most hours to retyping the same patient details across a phone line, a payer portal, and the practice management system. They are also the safest place to start, because the Act treats them as ordinary office automation rather than care.

What needs written consent first

Supplementary support—work that helps deliver care without being therapeutic communication—is allowed only after the practice obtains written patient consent. This is the category that trips up the most popular AI purchases: ambient scribes and note generators. The statute permits AI scribes and ambient documentation, along with maintaining therapy notes and analyzing anonymized data, but each remains subject to the licensed professional's review and approval.

The consent itself has a specific shape. Under Orrick's analysis, the patient must be told in writing that AI will be used and why, and must consent in writing; the consent can be electronic, but it has to be a separate, stand-alone document. You cannot fold it into a general terms-of-use agreement, and consent collected through a pre-checked box or a dialog the patient simply closes does not count.

What no vendor can deliver in Illinois

Some capabilities are off the table regardless of how a product is marketed. The Act prohibits AI from making independent therapeutic decisions, interacting directly with a client in any form of therapeutic communication, generating treatment plans or recommendations without the licensed professional's review and approval, and detecting a client's emotions or mental state. A clinician may still review and endorse AI-drafted communications, but the judgment has to stay with the person.

So if a demo shows a chatbot triaging symptoms, suggesting a diagnosis, or reading a patient's mood from their voice, that feature is not one you can switch on in an Illinois practice—even when the vendor sells it freely in other states.

A workflow-by-workflow map

This table maps the administrative workflows a behavioral health group actually runs to the three categories the Act creates. It is an operating method for sorting your own task list, not legal advice; confirm any close call with your counsel and IDFPR's current guidance.

Run every candidate tool through the same two questions: does it communicate therapeutically with a patient, and does it decide anything clinical on its own? If it only moves information or drafts work a clinician will review, it usually belongs in the first rows, not the last three.

What to require from a vendor before you sign

Because the penalty attaches to how a tool is used, not to the wording of a sales contract, the safeguards have to live in how the product is configured. Before you sign, require that the tool can:

• Turn off any clinical, triage, or emotion-detection feature at the account level, with evidence it stays off.
• Route every AI-generated note or message to a named licensed professional for review and signature before it reaches a chart or a patient.
• Capture and store the stand-alone written AI-use consent, timestamped per patient, and block scribe features until that consent is on file.
• Produce an audit log of who reviewed what and when, so you can answer a regulator if asked.
• Hold patient data under a business associate agreement consistent with your existing HIPAA obligations.

Our open AI vendor evaluation scorecard gives you a neutral place to score competing tools against criteria like these; it is a planning aid, not a certification that any product complies with Illinois law.

A consent-gated scribe rollout, step by step

Here is a labeled rollout for a four-clinician Chicago counseling group adopting an AI scribe. The numbers are illustrative and show the sequence, not a measured result from a real client.

1. Week 1 — scope. Confirm the scribe only drafts notes for clinician review and has no triage or emotion-detection mode; disable anything that does.
2. Week 2 — consent. Build the separate written AI-use consent into intake—electronic but stand-alone—and configure the tool so it will not record a session until that consent is logged.
3. Week 3 — pilot. Run one clinician on the scribe. Every draft note is edited and signed by that clinician; nothing auto-posts to the chart.
4. Week 4 — review. Check the audit log: were all recorded sessions consented, and did a licensed professional sign every note? Only then expand to the other three clinicians.

The point of the sequence is that consent and clinician review are gates, not paperwork added at the end. If either gate cannot be enforced by the tool itself, the rollout stops there.

Quick answers

Can an Illinois practice use an AI scribe?

Yes, as supplementary support, but only with the patient's separate written consent, and a licensed professional must review and approve every note the scribe drafts.

Does scheduling or billing automation need patient consent under the Act?

No. Scheduling, reminders, billing, and insurance follow-up are administrative support that does not involve therapeutic communication, so the Act permits them without a separate AI-use consent.

What happens if a tool crosses the line?

IDFPR enforces the Act and can impose civil penalties of up to $10,000 per violation after a hearing, so the exposure follows how a tool is used, not only what a contract says.

Does this apply outside behavioral health?

The Act is specific to mental health and therapy services in Illinois. Medical, dental, and other practices answer to their own rules; this article does not cover them.

Sources

1. The Illinois Department of Financial and Professional Regulation (IDFPR) · idfpr.illinois.gov
2. IDFPR enforces the Act · orrick.com
3. Holland & Knight's reading of the statute · hklaw.com