Independent guides to AI tools for working professionals · No paid rankings · Every source linked

The 2026 Map: 20 States Now Regulate AI Chatbots in Mental Health

Prices and facts checked October 13, 2026. We don't accept payment for rankings.

AI therapy laws by state register, September 2026 version
Screenshot: Psychology.com's AI therapy laws register, September 2026 version (psychology.com), captured October 13, 2026
On this page
  1. Three types of law
  2. The state-by-state list
  3. The line most therapy laws draw
  4. What this means for your practice
  5. FAQ

What happened: state lawmaking on AI and mental health exploded in 2026. A September 12, 2026 state register lists roughly 20 states with enacted laws on AI therapy or AI companion chatbots, with effective dates running into 2027 (Psychology.com register). The Future of Privacy Forum counted 98 chatbot bills across 34 states this year (AIMOCS tracker).

Why it matters today: these laws follow where the patient is, not where you practice (Acuity News). Telehealth therapists and practices with AI intake bots or between-session apps now face a patchwork, and some laws restrict what clinicians themselves can do with AI.

Not legal advice. Several entries below rely on summaries; some texts weren't fully verified by the trackers. Confirm with your state's legislature and licensing board.

Three types of law

  1. Therapy restrictions: AI can't provide therapy or make treatment decisions; licensed clinicians stay in charge. Examples: Illinois, Nevada, Colorado, Maine, Rhode Island, Vermont.
  2. Companion and chatbot safety: crisis protocols, "not human" notices, minor protections. Examples: California, New York, Washington, Oregon, Connecticut.
  3. Representation limits: AI can't be marketed or presented as a licensed professional. Examples: Tennessee, Utah, California AB 489.

The state-by-state list

State Law Effective Focus
Illinois WOPR Act Aug 2025 Therapy restriction; $10,000/violation
Nevada AB 406 Jul 2025 Therapy restriction
Utah HB 452 May 2025 Mental health chatbot disclosures and data limits
New York GBL Art. 47 Nov 2025 Companion crisis protocols and notices
California SB 243; SB 1119; SB 867 Jan 2026; Jul 2027; Jan 2027 Companion safety; child safety assessments; toys
Texas HB 149 Jan 2026 Healthcare AI disclosure
New Hampshire HB 143 Jan 2026 Liability for harms to children
Tennessee Pub. Ch. 647 Jul 2026 No representing AI as a mental health professional
Hawaii Act 248 Jul 2026 Identity, self-harm response, minors
Maine LD 2082 Jul 29, 2026 Human-delivered therapy; consent for AI support
Colorado HB 26-1195 Aug 12, 2026 Therapy restriction; recording consent
Vermont H 816 Jun 2026 Limits on independent AI mental health services
Rhode Island S 2197 / H 7349; S 2195 Jun 2026; Jan 2027 Therapy restriction; companion safety
Connecticut PA 26-15 Jan 2027 Crisis protocols, disclosures
Oregon SB 1546 Jan 2027 Disclosure, self-harm safeguards
Washington HB 2225 Jan 2027 Crisis protocols, minors
Georgia SB 540 Jul 2027 Identity, crisis, professional claims
Iowa SF 2417 Jul 2027 Disclosures, self-harm protocol
Nebraska LB 525 Jul 2027 Chatbot disclosures and crisis protocols

Source: Psychology.com register, status Sep 12, 2026. Idaho's S 1297 was reported enacted but not verified. Pending: Ohio HB 525, Pennsylvania, New Jersey, Michigan, Massachusetts, North Carolina and others.

The line most therapy laws draw

Across the restriction states, the pattern is consistent (AIMOCS; Acuity News):

  • Allowed: scheduling, billing, documentation and, with consent, supplementary support under clinician supervision
  • Not allowed: AI making therapeutic decisions, holding therapeutic conversations on its own, or being marketed as therapy

Illinois was first in August 2025 (Baker Botts). Rhode Island also bars providers from using AI to simulate emotional attachment or dependency (Transparency Coalition).

What this means for your practice

  1. Map where your clients are, especially telehealth clients.
  2. Apply the strictest rule: disclose AI use, get written consent, keep a clinician in the loop (Acuity News).
  3. Audit client-facing AI: intake bots, between-session apps, chat features. These are what most laws target.
  4. Watch your marketing: don't describe any AI feature as therapy or a therapist.
  5. Don't rely only on vendor assurances. The practice using the tool usually carries the exposure.

AI note-taking has its own consent rules: see AI scribe consent laws. Companion chatbots used by clients are covered in AI companion chatbot laws.

FAQ

Is there a federal law on AI therapy?

No. A December 2025 executive order directs challenges to some state AI laws, but its effect is unresolved; treat state laws as binding.

Do these laws ban AI note takers?

No. Documentation support is generally allowed, often with consent.

Will more states follow?

Yes. Dozens of bills are pending for 2027 sessions.

Sources

Pages we used for this article, checked on or before the update date above.

  1. Psychology.com register · psychology.com
  2. AIMOCS tracker · aimocs.com
  3. Acuity News · acuity.news
  4. Baker Botts · ourtake.bakerbotts.com
  5. Transparency Coalition · transparencycoalition.ai

Changelog: Oct 13, 2026: first published.