GLOSSARY

AI therapy disclosure

Also: AI disclosure in mental health, chatbot disclosure

DEFINITION

AI therapy disclosure is the practice, and in some places the legal duty, of telling a person that the system they are talking to is an AI, not a human clinician.

In plain words

AI therapy disclosure means being told, clearly and early, that you are talking to an AI. A good disclosure also says what the AI is for, what it is not, and where to go when something is serious. The point is that you can decide how much weight to give what it says.

Where the idea comes from

Rules differ by place, and none is quoted here as legal advice. As of 2026-10-30:

  • European Union. Article 50 of the AI Act requires providers of AI systems that interact directly with people to design them so people know they are dealing with an AI, unless that is obvious to a reasonably well-informed person from the context. The information must be given no later than the first interaction. The page gives 2 August 2026 as the date the article applies.
  • Utah (HB 452). Signed 25 March 2025, effective 7 May 2025. Per CASRAI's comparison, mental health chatbots remain legal but users must be told they are talking to AI before access, again after seven days of non-use, and whenever they ask. It also limits selling identifiable health data and requires advertising to be labelled.
  • Illinois (HB 1806). Signed 4 August 2025. It prohibits AI from providing therapy and requires written, informed consent before AI is used on a recorded or transcribed session.
  • Nevada (AB 406). Effective 1 July 2025. It makes it unlawful to offer an AI system that provides professional mental or behavioural health care, or to claim that an AI can.

How it shows up

For a person choosing an app, disclosure shows in the first screen, in what the agent calls itself, and in whether it ever describes itself as a therapist. On TwentyThird, Oneli is an AI and says so on every surface. TwentyThird is not therapy, not an AI therapist, not a medical device and not a crisis service, and it is not monitored in real time. If you are in danger, contact local emergency services now. In the EU, call 112. See is AI therapy legal and safe.

What it is not

Disclosure is not consent to anything, and it does not make a product safe or lawful. In Illinois and Nevada, as the CASRAI comparison notes, honest labelling does not cure a prohibited service. Disclosure rules also change quickly, and this entry covers only the sources named above. Check the current text of the law, or ask a qualified lawyer, before relying on it.

Sources

  1. Article 50: Transparency obligations, EU Artificial Intelligence Act. Future of Life Institute (artificialintelligenceact.eu), 2026.
  2. AI therapy laws: Illinois, Nevada and Utah compared. CASRAI, 2025.
  3. Chatbots in check: Utah's latest AI legislation. Future of Privacy Forum, 2025.

If you are in danger, contact local emergency services now — 112 anywhere in the EU. TwentyThird is not an emergency service. How safety works