Companion chatbots are AI systems built for ongoing, personal conversation: AI friends, partners and characters. These laws set duties for the companies that run them, most often an AI notice, a protocol for talk of suicide or self-harm, and extra protections for minors.
14 enacted laws in 10 states, plus 20 bills and actions, as of November 4, 2026.
Operators must keep a documented protocol to reduce material risk of statements promoting suicidal ideation, suicide or self-harm, including referral to crisis services. If the operator determines there is a credible and imminent threat, it must do at least one of: notify a linked parent as soon as practicable (if that does not risk serious harm) and tell the child, or give streamlined access to 988 or an equivalent helpline.
An operator may not let its companion chatbot engage with users unless it maintains a protocol to prevent the chatbot producing suicidal ideation, suicide or self-harm content, including a notice referring users who express those thoughts to crisis services such as a suicide hotline or crisis text line. The operator must publish details of the protocol on its website.
Operator must implement a protocol for prompts about suicidal ideation or self-harm that refers users to a crisis service provider (not law enforcement) and has escalation procedures for repeated or severe indicators.
Operator must use evidence-based methods to detect clear risk of suicide, self-harm or imminent violence and to prevent outputs encouraging them, refer users to resources including 988, refer again to mental health services on repeat expressions, and post the protocol on its website.
Operator may not make the chatbot available without a protocol for detecting and addressing severe harm: methods to identify self-harm and eating-disorder-related self-harm, referral to resources such as the 988 Lifeline, measures against content that encourages severe harm, and escalation for repeated or severe indicators.
Operator must adopt a protocol for prompts about suicidal ideation or self-harm that tries to refer users to crisis services, use evidence-based methods to measure risk, prevent human-claims during crisis conversations, and prevent outputs that encourage serious injury to others.
Operator must adopt a protocol for responding to prompts about suicidal ideation, including reasonable efforts to refer users to crisis services such as a suicide hotline or crisis text line.
Unlawful to operate an AI companion without a protocol that reasonably detects and responds to user expressions of suicidal ideation or self-harm and refers the user to crisis services such as the 988 hotline or a crisis text line.
Operator may not give Oregon users access unless it has an evidence-based protocol to detect suicidal or self-harm ideation and prevent content encouraging it, including referral to the 988 lifeline (optional youthline referral for users identified as under 25) and clinical best-practice intervention for users who keep expressing ideation. Protocol details must be published on the operator's website.
Unlawful to operate an AI companion without a protocol for possible suicidal ideation, self-harm or physical harm to others that notifies the user and refers them to crisis services as soon as detected.
Operator may not make the chatbot available unless it maintains a protocol to detect suicidal ideation and self-harm (including eating disorders), refer users to crisis resources such as a suicide hotline or crisis text line, and prevent content that encourages or describes self-harm.