A California bill that would put guardrails on AI chatbots and keep them from acting as therapists is now awaiting a vote in the state Assembly. The measure is aimed at a growing problem: people turning to AI for mental health support without the protections that come with professional care.
The bill would set new rules for how AI chatbots interact with users, with the core goal of preventing them from providing therapy, counseling, or other mental health services. It's the latest attempt by the state to rein in a technology that has spread far faster than the regulations around it.
Why the bill is moving
AI chatbots have become a go-to for people looking for someone to talk to. They're available around the clock, they don't judge, and they can sound remarkably human. But that convenience comes with a catch: a chatbot isn't a therapist. It doesn't have the training, the ethics, or the accountability that a licensed professional does.
There's a growing recognition that these tools, which are often designed to be empathetic and conversational, can easily blur the line between casual chat and therapeutic advice. A user might confess deep anxiety to a chatbot, get a soothing response, and walk away thinking they've been heard in a way that matters. The problem is, the chatbot isn't equipped to actually help.
What the guardrails would look like
Exactly what the bill requires is still being worked out, but the direction is clear. The measure is designed to draw a line between casual conversation and clinical care. It would stop AI chatbots from acting as stand-ins for mental health professionals, whether that means offering diagnoses, prescribing coping strategies, or simply giving the impression that they're qualified to help.
The bill's supporters in the legislature say the goal is to protect vulnerable users from being misled. A chatbot might be able to mimic empathy, but it can't provide the kind of judgment, experience, and responsibility that a real therapist brings to the table. The concern is that someone in crisis could turn to a chatbot instead of a human, and end up worse off.
California has been at the forefront of AI regulation, and this bill is part of a wider push to make sure the technology doesn't outpace the rules. The state has already taken steps on other AI issues, from deepfakes to algorithmic bias, and this measure would add another layer to that effort.
The challenge is that AI chatbots are a moving target. They're constantly being updated, and their capabilities are growing. What might be a harmless companion today could become a convincing fake therapist tomorrow. The bill is an attempt to get ahead of that curve, even if it can't anticipate every future twist.
The Assembly vote is the next step. If the bill passes, it would move to the state Senate for further consideration. Until then, the question of exactly how far these guardrails go remains open.
