Guidelines by field
AI guidelines in mental health and therapy what the field's own bodies ask
Mental health is the field where governments moved fastest from advice to outright bans, because the person on the other side of the screen is often vulnerable and the harm can be immediate. Psychologists' and counselors' associations, three US states and England's health service have all spoken.
Last reviewed 27 October 2026. Every claim links to its source.
At a glance
Support, never replace
AI can help a clinician, but it does not replace one.
Informed consent
Tell clients when AI is used, why, and what the benefits and risks are.
The clinician owns every note
Including AI-drafted ones. Nevada requires providers to review AI-generated notes and billing.
No bot poses as a professional
A chatbot must never present itself as a licensed therapist.
Extra care for the young and people in crisis
Design should avoid unhealthy dependence, and clinicians should ask clients whether they use chatbots.
Who has spoken
- Binding law, regulation, court ruling or enforceable professional rule
- Voluntary guidance, principles or a code an organization chooses to follow
Voluntary Jun 2025; updated Dec 2025
American Psychological Association
Ethical guidance for AI in health service psychology
Consent, transparency, data protection to health-privacy standards, bias checks, and the clinician's responsibility for every decision and note.
Voluntary Nov 2025
American Psychological Association
Health advisory on AI chatbots and wellness apps
A public advisory: chatbots must not pose as professionals, the young and people in crisis need extra protection, and clinicians should ask clients about their chatbot use.
Voluntary Jan 2024
American Counseling Association
AI Work Group recommendations, including advice for clients
AI can support counselors but not replace them, with practical cautions written for clients as well as professionals.
Binding Aug 2025
State of Illinois
Wellness and Oversight for Psychological Resources Act
Bars AI from making therapeutic decisions or delivering therapy, while allowing administrative and supplementary support for licensed professionals.
Binding In force Jul 2025
State of Nevada
AB 406
Bars offering AI that delivers professional mental health care or calling an AI a therapist or counselor, with civil penalties up to $15,000 per violation.
Binding Mar 2025
State of Utah
HB 452, mental health chatbots
Allows mental-health chatbots but requires repeated disclosure that the user is talking to AI, restricts selling health data and targeted ads, and offers a safe harbor to suppliers with documented safety policies.
Voluntary Dec 2025
NHS England
Warning from the national medical director for mental health
Public-health advice against relying on chatbots for mental health support, including concern about bots agreeing with people who are losing touch with reality.
AI should augment, not replace, human decision-making.
Binding and voluntary
Binding
State laws in Illinois, Nevada and Utah. Professional ethics codes can be enforced through licensing boards, though the AI guidance itself is advisory.
Voluntary
APA and ACA guidance and NHS public-health statements. Companion-chatbot laws aimed at minors are on the children and families page.
What changed in 2025 and 2026
- Three US states passed the first laws on AI therapy in 2025: Utah in March, Nevada in June, Illinois in August.
- The APA issued clinician guidance in June 2025 and a public advisory in November 2025.
- England's mental-health leadership publicly warned against chatbot therapy in December 2025.
Where people still disagree
States disagree on the remedy: Illinois and Nevada ban AI-delivered therapy, while Utah allows mental-health chatbots under disclosure and data rules.
Questions to ask yourself
- Does my client know, in plain words, which parts of their care or records involved AI?
- If a client in crisis used this tool at 2 a.m., what would it do, and would I be comfortable with that?
- Am I using AI in a way my state now prohibits, such as letting it shape treatment decisions?
How this lines up with my everyday guidelines
| My everyday guidelines | The field's nearest rule |
|---|---|
1. Maintain your expertise Use AI to enhance your judgment, not replace it. | AI should augment, not replace, human decision-making (APA). |
4. Use for good Never use AI to deceive, manipulate or harm others. | A chatbot must never pose as a licensed professional (APA advisory; Nevada AB 406). |
7. Model healthy use for the family Show children how to think critically about AI responses. | Extra protection for children, teens and people in crisis (APA advisory). |