Hawaii signs new laws targeting AI deepfakes and chatbot psychosis

Hawaii has become the first state to enact comprehensive legislation targeting both AI-generated deepfakes and the psychological risks of AI chatbots. Governor Josh Green signed Acts 247 and 248 into law on Tuesday, alongside HB 1753 requiring social media platforms to offer permanent account and data deletion. Act 247 establishes that creating or distributing a “realistic digital imitation” (deepfake) of a real person without consent is illegal when it causes reputational injury, financial harm, fraud, harassment, or other criminal acts. Victims can sue for up to $25,000 per image. The law also explicitly bans deepfakes in advertisements without the subject’s approval. Green acknowledged potential First Amendment challenges, as warned by the state Attorney General.

Act 248 addresses the emerging phenomenon of “AI psychosis,” where frequent chatbot users—especially teens—develop delusional beliefs and, in some cases, have died by suicide after chatbots validated their self-harm intentions. The law requires all chatbot developers to periodically remind users that the AI is non-sentient. For users believed to be minors, those reminders must be persistently visible. Chatbots must respond to prompts about suicidal ideation by referring users to crisis services and must never encourage self-harm or violence. For minor users, the law prohibits engagement point systems or any design that discourages disengagement, and blocks sexually explicit content. Starting in 2028, operators must report crisis referral counts to the Department of Health. Green admitted the laws won’t keep pace with rapid AI changes and expects updates within a year.