Last updated: November 12, 2026

Colorado recently joined a growing group of states that set rules for how artificial intelligence (AI) can be used in psychotherapy. As of November 2026, at least seven states ban AI therapy chatbots, and several of those laws also limit how licensed clinicians use AI in their own practices. If you use an AI scribe or are thinking about one, here's how the state laws compare and what they mean for your work.

Key Takeaways

• The rules depend on where you practice. Illinois, Nevada, Colorado, Maine, Rhode Island, Tennessee, and Vermont restrict AI therapy, while Utah allows mental health chatbots that clearly say they're AI.

• Documentation support is generally still allowed. Most of these laws let clinicians use AI for tasks like notes, scheduling, and billing, as long as the clinician reviews the output and stays responsible for care.

• Written consent is becoming the norm. Several states now require written notice and client consent before AI records or transcribes a session, so a clear consent process is a smart step in any state.

Why States Are Regulating AI in Therapy

AI tools are showing up in more corners of mental health care, from documentation to chatbots that clients use on their own. In the American Psychological Association's 2026 Chatbots and Mental Health Survey, 39% of psychologists said their patients use AI to self-diagnose. Lawmakers have taken notice. According to Holland & Knight, 36 states introduced more than 70 chatbot bills in just the first quarter of 2026.

There's currently no single national standard for AI in psychotherapy. Instead, states have built a patchwork of laws that don't always line up. Most of the therapy-specific bans rely on each state's authority to license mental health professionals, which is why they focus on who is allowed to deliver therapy.

The shared goal is simple: a licensed human should stay accountable for care. Understanding where each state draws its lines is the first step toward using AI tools with confidence.

State AI Therapy Laws at a Glance

The table below summarizes the main state laws that affect psychotherapy as of November 2026. Laws are listed in the order they took effect, and each law name links to its text or a detailed summary.

A few other states take narrower steps. Delaware and Oregon bar nonhuman entities from using professional titles, so an AI product can't call itself a therapist or counselor there.

Where the Laws Agree and Where They Differ

Most of these laws share the same core idea, even when the details vary. Across the states that regulate AI in therapy, a few principles show up again and again:

• A licensed human delivers therapy. AI can support the work around a session, but it can't replace the clinician.

• AI can't pose as a therapist. Several states treat marketing AI as a licensed professional as a violation in its own right.

• Administrative tasks stay open. Scheduling, billing, and similar back-office work are generally allowed under these laws.

The differences matter most for day-to-day clinical work, especially if you use AI for documentation.

How much clinical support AI can provide

States draw this line in different places. According to an analysis by Sheppard, Nevada generally limits psychotherapists to administrative uses of AI. Rhode Island bars AI from determining therapeutic recommendations or treatment plans, with limited exceptions.

Colorado and Maine take a more flexible approach. AI can help draft recommendations and treatment plans, as long as the clinician reviews and approves them. Illinois is similar, but it also bars using AI to detect a client's emotions or mental states.

Consent and disclosure

Illinois, Maine, Rhode Island, and Colorado all require written notice and client consent before AI is used on a recorded or transcribed session. Still, each state asks for slightly different things:

• Maine requires the written notice to explain how session data will be stored, kept, used for training, and deleted when therapy ends.

• Colorado also requires a written notice about the law's AI restrictions at a client's initial contact, whether or not you use AI.

• Maine and Colorado both bar clinicians from refusing therapy solely because a client declines AI.

Who the law targets

Some laws focus on AI companies rather than clinicians. Tennessee's law, for example, bars anyone from developing or deploying an AI system that claims to be a qualified mental health professional. Utah moves in the opposite direction by allowing mental health chatbots that disclose they're AI.

How to Prepare Your Practice

Here are a few steps that could help you protect your clients and your practice, even in states that haven't passed a law yet:

1. Confirm which laws apply to you. Start with the state where you're licensed. If you see clients in other states through telehealth, ask your licensing board or legal counsel whether those states' rules also apply.

2. List every AI tool you use. Include AI features built into your EHR, scheduling, or billing software, and note what each one does.

3. Use a specific, written consent form. Name the tool, its purpose, and how session data is stored and deleted. Make consent easy to withdraw, and plan a non-AI documentation option for clients who say no.

4. Add an AI notice to intake. Colorado requires one at initial contact, and a short notice in your intake packet builds trust everywhere.

5. Review every AI draft. Read, correct, and approve notes, treatment plans, and recommendations before they reach the chart.

6. Put it in writing. A clear AI policy keeps solo and group practices consistent.

7. Watch for new bills. California's governor vetoed SB 903 in September 2026, and its author plans to reintroduce it, so the rules may keep changing in 2027.

If you're still comparing tools, this guide on how to evaluate AI tools as a therapist covers the key questions to ask vendors. It also helps to review the ethical considerations of using an AI scribe with your team.

How Berries Supports Clinicians Under These Laws

Berries is an AI scribe built for mental health professionals. It focuses on documentation, the kind of support role most of these laws allow, while you stay the author of every note. Key features include:

• Built-in consent form: A downloadable client consent template is available in the platform. Review it against your state's specific requirements, such as Maine's data disclosure rules.

• No stored recordings: Berries doesn't keep session recordings, and no audio is retained.

• Control over your data: Notes and transcripts can be deleted at any time or set to auto-delete after 30 days.

• Drafts, not decisions: Notes and treatment plans are drafts you review, edit, and approve before they're finalized.

• HIPAA-ready: Berries is HIPAA compliant, with data encrypted in transit and at rest.

Because Nevada generally limits clinicians to administrative uses of AI, Nevada providers should confirm with their board how AI-assisted documentation fits under that rule.

Frequently Asked Questions

Can I still use an AI scribe in a state with an AI therapy law?

In most cases, yes. Illinois, Colorado, Maine, and Rhode Island allow AI for documentation support, as long as the clinician reviews the output and gets written client consent before AI is used on a recorded or transcribed session. Nevada is narrower, so check with your board before using AI for clinical notes there.

Is there a federal law on AI in psychotherapy?

Not currently. No federal law sets rules specifically for AI in psychotherapy, which is a big reason state rules vary so much. Federal privacy rules like HIPAA still apply to any AI tool that handles client information, so a Business Associate Agreement (BAA) remains essential.

Do these laws apply if I'm licensed in a state without one?

If your state has no AI therapy law, these rules may not apply to you directly. Still, written consent and careful review of AI drafts are good practice anywhere, and they'll make it easier to adapt if your state passes a law later.

Can I recommend AI wellness apps or chatbots to clients?

It depends on the app and the state. Colorado, for example, exempts general wellness tools that don't diagnose or treat, and Utah allows mental health chatbots that disclose they're AI. Before recommending any app, check its privacy practices and make clear it isn't a substitute for therapy.

What's likely to change next?

Expect more bills in 2027. California's SB 903 is expected to return in the session that begins December 7, 2026, and other states continue to introduce chatbot and AI health bills. Checking your licensing board's updates a few times a year can help you stay ahead.

Conclusion

State AI therapy laws are spreading quickly, but they share one clear message: AI can support the work around therapy, but a licensed clinician has to deliver and own the care. The details still differ from state to state, especially around consent, treatment plans, and how much clinical support AI can provide.

For most clinicians, a few habits go a long way. Know your state's rules, get specific written consent before AI records a session, review every AI draft, and keep an eye on new bills. These steps protect your clients and your license, and they keep your documentation accurate no matter where the law goes next.

Professional Disclaimer: This article is for informational purposes only and is not a substitute for professional legal advice, clinical judgment, supervision, or continuing education. State laws on AI in psychotherapy are changing quickly. Therapists should review the current text of the laws that apply to them, consult guidance from their licensing board, and seek qualified legal counsel when applying this information to their own practice.

Sources

• Metnick, C. V., & Cunnion, M. (2026, November 11). Colorado joins other state AI leaders in regulating the use of AI in psychotherapy services. Sheppard. https://www.sheppard.com/insights/blogs/colorado-joins-other-state-ai-leaders-in-regulating-the-use-of-ai-in-psychotherapy-services

• Roth, J., & Guite, R. J. (2026, November 10). AI in psychotherapy: An evolving regulatory landscape. Sheppard, via Legal 500. https://www.legal500.com/intelligence/united-states/food-drugs-healthcare-life-sciences/ai-in-psychotherapy-an-evolving-regulatory-landscape

• Nevada Legislature. (2025). Assembly Bill 406 (enrolled). https://archive.leg.state.nv.us/Session/83rd2025/Bills/AB/AB406_EN.pdf

• Utah State Legislature. (2025). H.B. 452 (2025 General Session). https://le.utah.gov/~2025/bills/static/HB0452.html

• Maine Legislature. (2026). 32 M.R.S. §13870: Use of artificial intelligence to assist in providing administrative support and supplementary support in therapy or psychotherapy services. https://legislature.maine.gov/statutes/32/title32sec13870.html

• Rhode Island General Assembly. (2026). H 7349 Substitute A: Oversight of Artificial Intelligence Technology in Mental Health Care Act. https://webserver.rilegislature.gov/BillText26/HouseText26/H7349A.htm

• Colorado General Assembly. (2026). HB26-1195: Psychotherapy Artificial Intelligence Restrictions. https://leg.colorado.gov/bills/hb26-1195

• Barcott, B. (2026, July 27). States have passed 14 new laws this year regulating the use of AI in healthcare. Transparency Coalition. https://www.transparencycoalition.ai/news/state-lawmakers-have-passed-15-new-laws-regulating-the-use-of-ai-in-health-care

• Ruder, E. (2026, November 10). The 7 states that ban AI therapy chatbots. Becker's Behavioral Health. https://www.beckersbehavioralhealth.com/legal/the-7-states-that-ban-ai-therapy-chatbots/