Robot Therapists? Colorado's New Bill Draws a Hard Line on AI in Mental Health.

The Bottom Line

If you're seeing a therapist, you might have noticed AI note-takers joining the session. This bill puts strict guardrails on how mental health professionals can use artificial intelligence, banning AI from conducting therapy or reading your emotions while requiring your explicit, written permission before an AI listens in.

What This Bill Actually Does

As artificial intelligence gets smarter, the line between a helpful clinical tool and an automated therapist is blurring. HB26-1195 draws a hard line in the sand for regulated mental health professionals—including psychologists, social workers, counselors, and even unlicensed psychotherapists. Section 12-245-224.5 of the legislation explicitly bans professionals from letting an artificial intelligence system conduct therapeutic communication directly with clients.

Under the new law, a therapist cannot let an AI do any of the following:

• Directly interact with clients to provide therapy or emotional support.
• Generate treatment plans or recommendations without human review and approval.
• Detect emotions or mental states during a session.

However, the bill doesn't ban AI entirely. It allows therapists to use AI for administrative support (like scheduling appointments and processing insurance) and supplementary support (like drafting session notes or analyzing anonymized data). But there is a major catch: if a therapist wants to use an AI tool to record or transcribe your session, they must get clear, written, informed consent from you in advance. A vague "terms of service" pop-up or a box you accidentally check while scrolling does not count under this law.

What about apps like Calm or BetterHelp? The bill creates an exemption for self-help and wellness apps, provided they do not diagnose or treat mental health disorders and clearly state they are not a substitute for clinical care. It also makes it a deceptive trade practice (Section 6-1-1705.2) to advertise an AI as being equivalent to a licensed Colorado therapist or to falsely claim that an AI's data is completely confidential.

What It Means for You

As a patient, this legislation ensures that your therapy sessions remain fundamentally human. You won't be pawned off on a chatbot for "pre-counseling," and your therapist cannot legally let an algorithm dictate your treatment plan without verifying it themselves. When you seek professional mental healthcare in Colorado, you are guaranteed an expert, not a string of code.

Most importantly, your digital privacy gets a major upgrade. If your therapist wants an AI assistant—like an automated transcription service or a clinical scribe—to listen in and take notes, they have to ask your permission first. The written consent mandate is highly specific. It must tell you exactly what the AI is doing and why. If the idea of a machine listening to your deepest traumas makes you uncomfortable, you have the absolute right to refuse. The bill legally protects that choice, stating your therapist cannot deny you services just because you opted out of the AI transcription.

For those exploring mental health apps on their phones, you will start seeing clearer boundaries and disclaimers. Any app offering guided meditation, coaching, or self-help must prominently disclose that it isn't clinical care. If a company tries to market an AI bot as "just as good as a licensed therapist," they will face the wrath of the Colorado Consumer Protection Act. It's a layer of consumer protection designed to ensure you aren't misled when you're looking for help.

What It Means for Your Business

If you own a mental health practice or work as a licensed professional, this legislation requires an immediate audit of your tech stack. You are legally clear to use AI for administrative support (scheduling, billing) and supplementary support (summarizing notes), provided you retain full responsibility for the output. However, you must implement new intake protocols for patient consent if your AI tools "listen" to sessions.

To comply with the new consent rules, your intake forms must:

• Be a clear, explicit, and unambiguous written agreement (electronic is fine).
• State specifically that an AI system will be used and explain exactly what its purpose is.
• Be entirely separate from broad, general terms-of-service agreements.
• Be revocable at any time without penalty to the patient's care.

For tech developers and startups building in the "med-tech" space, Colorado is establishing a strict regulatory environment. The bill expressly outlaws systems designed to "detect emotions or mental states" in clinical settings. Furthermore, your marketing teams must be extremely careful. Falsely advertising your AI tool as a replacement for a licensed therapist, or implying that your AI's outputs are endorsed by regulated professionals, is now classified as an unfair trade practice. You must ensure your marketing materials clearly distinguish between clinical tools and general wellness products.

There are useful carve-outs for businesses focused on training and academia. If you are developing AI for research under a federally registered Institutional Review Board, or using AI in an accredited educational program for clinical simulations, you are largely exempt from the public-facing restrictions—just ensure these tools aren't deployed to the general public. Violating these new rules carries heavy consequences: up to a $5,000 fine per violation from the Department of Regulatory Agencies, and up to $20,000 per violation from the Attorney General for deceptive trade practices.

Follow the Money

According to the nonpartisan fiscal note, this bill does not require a new state appropriation or major taxpayer funding. The regulatory bodies overseeing this space—the Department of Regulatory Agencies (DORA) and the Behavioral Health Administration (BHA)—will absorb the minor workload of updating rules and investigating complaints within their existing operating budgets.

However, the legislation has the potential to generate minimal new state revenue driven by penalties. Licensed professionals who violate the AI rules could face administrative fines up to $5,000 per violation, which would flow directly into DORA's licensing cash funds. On the corporate side, companies committing deceptive trade practices by falsely advertising AI therapy could face civil penalties up to $20,000 per violation, generating damage awards for the state.

Where This Bill Stands

HB26-1195 is currently Signed Into Law. The latest official action came on 06/03/2026: Governor Signed.

That means the legislative process is complete and the bill is now law. The remaining questions are about implementation timing and how agencies, businesses, or local governments respond.