Requirements of tier III companion AI chatbots

A covered entity shall— (1) prohibit the use of a tier III companion AI chatbot by a minor user unless such use is supervised by a licensed medical or mental health professional; (2) require such a companion AI chatbot to— (A) provide a disclosure not less frequently than every 60 minutes during a continuing interaction between such chatbot and a minor user that the chatbot is not a human or licensed professional in any field; (B) provide a nudge-out notification not less frequently than every 60 minutes during any such interaction; (C) maintain, publish on a publicly available website in clear, accessible language, and require such companion AI chatbot to follow, evidence-based protocols designed to— (i) detect expressions of suicidal ideation or self-harm; (ii) refer minor users to— (I) the 988 Suicide & Crisis Lifeline; (II) licensed medical and mental health professionals; and (III) other appropriate crisis service providers; and (iii) prohibit such companion AI chatbot from providing crisis counseling or de-escalation services to any minor user; and (D) notify the parent or guardian of a minor user without unreasonable delay if such minor user expresses suicidal ideation; (3) prohibit a tier III companion AI chatbot from— (A) providing a minor user with any diagnosis or treatment decision; and (B) functioning as a standalone provider of professional mental-health care; and (4) establish and maintain a risk management program for tier III companion AI chatbots containing written policies providing for employees of such entity to— (A) continually monitor such chatbots; (B) become familiar with crisis de-escalation measures; and (C) engage in recordkeeping regarding such chatbots.