Hawaii law requires AI companion operators to disclose that users are chatting with artificial intelligence and follow added safeguards when they know a user is under 18.
AI companion operators must tell users when a chat could seem like a conversation with a person. When they know or are reasonably certain a user is under 18, they must keep an AI notice visible or repeat it during chats.
Operators must set responses for self-harm concerns and take steps to prevent dangerous or misleading chats. For minors, they must also limit certain sexual content and pressure to keep chatting, and offer screen-time and account controls.
Operators must report yearly on crisis referrals and safety steps, without including information that identifies users. Reports begin in 2028.
Operators must notify users when an AI companion could be mistaken for a human. An operator is a person who develops and makes an AI companion available to the public. App stores and search engines that only provide access are not operators under this section. The law took effect when the governor approved it on July 14, 2026.
Operators must use safety steps for self-harm prompts and certain interactions with minors. These rules apply to operators of AI companions. The reward, disengagement, sexual-content, and account-control rules apply when an operator knows or is reasonably certain the user is under 18. For self-harm or suicide prompts, operators must adopt a response protocol that makes reasonable efforts to refer users to crisis services and use evidence-based ways to assess risk. Operators must not present an AI companion as providing professional mental or behavioral health care. They must take reasonable steps to prevent it from seeming human during crisis support or encouraging serious bodily injury to another person. When operators know or are reasonably certain a user is under 18, they must not use unpredictable rewards to increase engagement or allow the companion to discourage the user from stopping. They must take reasonable steps to prevent sexually explicit material and sexual comments about the user, and offer screen-time and account controls. The law took effect when the governor approved it on July 14, 2026.
Operators must send annual reports about crisis referrals and safety protocols to the state behavioral health administration. The duty applies to AI companion operators, who report to the Department of Health’s behavioral health administration. Reports may contain only the listed information and must not include user identifiers or personal information. A violation counts as an unfair or deceptive business practice, but this section does not create a private right to sue. It also does not make a model developer liable for a third party’s AI companion system. Annual reports are required beginning January 1, 2028.
No hearings or votes scheduled.