· ai regulation · 5 min read
Illinois Said AI Chatbots Can't Practice Therapy. They're Right.
The Wellness and Oversight for Psychological Resources Act prohibits unregulated AI systems from providing therapy or psychotherapy services in Illinois. If you're building mental health AI, this is your compliance briefing.
Photo by Vitaly Gariev on Unsplash
The mental health AI industry has been in a grey zone since the first chatbot claimed it could help with depression. Apps like Wysa, Woebot, and dozens of smaller competitors have offered what they carefully avoid calling “therapy” — framed instead as “emotional support,” “wellness coaching,” “CBT exercises,” or “mental health tools.” The framing isn’t accidental. It’s designed to avoid triggering professional licensing requirements.
Illinois decided to close the gap. The Wellness and Oversight for Psychological Resources Act — WOPRA, effective August 1, 2025 — prohibits unlicensed and unqualified entities, including unregulated AI systems, from providing therapy or psychotherapy services in Illinois. If your app delivers therapy, someone with a license needs to be accountable for it.
What the Law Says
WOPRA’s purpose section is direct:
“The purpose of this Act is to safeguard individuals seeking therapy or psychotherapy services by ensuring these services are delivered by qualified, licensed, or certified professionals. This Act is intended to protect consumers from unlicensed or unqualified providers, including unregulated artificial intelligence systems, while respecting individual choice and access to community-based and faith-based mental health support.”
The key phrase is “unregulated artificial intelligence systems.” Illinois didn’t ban AI from mental healthcare. It banned unregulated AI from practicing psychotherapy — which is a licensed profession in Illinois, as in every other state.
The Scope of the Prohibition
The Act covers “therapy or psychotherapy services” — which encompasses the treatment, diagnosis, and evaluation of mental health conditions. Under Illinois law, providing these services without a license is already illegal. WOPRA makes explicit that an AI system can’t do it either.
The law carves out what it calls “administrative support” — tasks that assist a licensed professional without constituting independent clinical practice. Scheduling appointments, transcribing session notes, processing paperwork, sending reminders. An AI can do all of that. What it can’t do is conduct the clinical assessment, formulate a treatment plan, or provide the therapeutic intervention.
The line between “administrative support” and “clinical practice” is exactly where the fights will happen. CBT exercises dispensed by an app — clinical or administrative? Safety planning prompts from a chatbot during a crisis — clinical or support? The law doesn’t answer these questions explicitly, and the digital health industry will spend years litigating them.
The Companies in the Grey Zone
Wysa describes itself as “an AI-powered mental health app” that uses evidence-based therapeutic techniques. Woebot offers “evidence-based tools” for managing depression and anxiety. Both companies have published clinical research on their tools’ effectiveness.
Here’s the thing: effectiveness isn’t the issue. The issue is licensure. A social worker who provides CBT exercises has a license, carries malpractice insurance, is subject to professional discipline if they harm a client, and has a legal duty to report if a client is in danger. An AI chatbot has none of that accountability infrastructure.
Illinois’s position is that if your product is delivering what functions as therapy, someone accountable under Illinois’s mental health licensing law needs to be responsible for that delivery. The chatbot doesn’t have to go away — it has to operate under licensed supervision.
What Compliance Looks Like
If you’re operating a mental health AI product with Illinois users:
Model 1: Full administrative support. Your AI does scheduling, symptom tracking, psychoeducation content delivery, and reminders. A licensed Illinois provider is the actual treating clinician. The AI is a workflow tool. This is clearly compliant.
Model 2: Licensed oversight of AI-delivered interventions. An Illinois-licensed professional reviews, approves, and supervises the AI’s clinical protocols. The clinician is accountable for the treatment plan; the AI is executing it. This is defensible but requires genuine clinician oversight — not a paper clinician whose name appears on the app while they’re never actually involved.
Model 3: Unsupervised AI therapy. No licensed clinician in the loop, AI delivers therapeutic interventions directly to users. This is what WOPRA prohibits.
The documentation question is important. If you’re operating under Model 2, you need to be able to show the licensing board what “supervision” actually means in your product — what the clinician reviews, how often, and what their authority is when the AI does something they wouldn’t approve.
The faith-based and community support carve-out in the statute is interesting. Illinois explicitly preserved space for peer support, pastoral counseling, and community-based mental health services. That’s a values choice — the legislature drew a line between licensed clinical care and traditional community support structures. Reasonable people can argue about where exactly that line should be.
The Market Reality
The mental health provider shortage is real. There aren’t enough licensed therapists to serve the demand, and AI tools have been filling gaps that the licensed system can’t cover. That’s the legitimate argument for these products.
WOPRA doesn’t eliminate that use case — it structures it. AI can still reach people who can’t access licensed care, but if it’s providing what functions as clinical services, someone with accountability under state law has to be responsible for it.
That’s the correct policy outcome. An AI system that tells a suicidal person to “practice mindfulness” and doesn’t have a licensed human accountable for that recommendation is a product that can hurt people. Regulation that requires licensed oversight isn’t the enemy of mental health access — it’s what makes AI-assisted mental health care actually safe to scale.
The grey zone was always going to close. Illinois just closed it first.
You can find the original text of Illinois HB 1806 (WOPRA) on the Illinois General Assembly’s website.