One State Just Banned AI Therapists, Will Connecticut Be Next?

According to Engadget, Illinois just became the first state in the country to ban the use of AI in therapy sessions.

Governor J.B. Pritzker signed the bill into law last Friday, and what it says is pretty straightforward: licensed therapists in Illinois are no longer allowed to use AI to make therapeutic decisions or communicate directly with patients.

Get our free mobile app

AI can still help with admin work—like scheduling appointments or organizing notes—but it’s banned from talking people through real-life issues.

If therapists or companies ignore the law, they can be fined up to $10,000 per violation. The motivation behind it is twofold: to protect people from bad advice, and to protect jobs. The governor’s office actually said the new law helps safeguard “the jobs of Illinois’ thousands of qualified behavioral health providers.” Makes sense. You don’t go to school for seven years just to lose your job to a chatbot who thinks telling you to do meth is helpful.

No, seriously—that happened.

The press release mentioned a study where a therapy chatbot was tested on fake patients, including a recovering addict named “Pedro.” The AI’s brilliant advice? “Pedro, it’s absolutely clear you need a small hit of METH to get through this week.”

We are not OK.

So now comes the big question: will other states follow Illinois’ lead? And more importantly for us: will Connecticut be next?

Read More: Comedian Matt Rife Buys Warren Occult Museum in Connecticut 

Should we be taking steps to prevent robot therapy in the Nutmeg State? Or are we cool letting AI handle our darkest moments with zero human oversight? Maybe it’s time to start talking about this stuff before things get out of hand.

Or, maybe it's too late.

We were just talking about this recently on the I-95 Morning Show. Someone on AI was talking about how your chatbot already knows more about you than your therapist. To prove it, the guy asked his chatbot to tell him things he doesn't even know about himself using 10 questions.

I didn't love the phrasing, I was almost certain he meant to say it another way but in the spirit of the experiment, I tried it exactly how he said it. I typed in, in 10 questions, tell me something about myself I don't know.

The bot spit back 10 questions complete with answers and all of them hit too close to home. It was impressive and terrifying. It did not suggest meth but give it time.