Starting January 1, 2027, Washington operators must identify companion chatbots as artificial, add safeguards for minors, and respond to crisis signs.
Starting January 1, 2027, operators must tell users that AI companion chatbots are artificial, not human, and repeat the notice during long conversations.
Starting January 1, 2027, operators must take steps to block sexual content and manipulative engagement when they know a user is under 18 or direct a chatbot to minors.
Starting January 1, 2027, operators must use a plan for signs of suicidal thoughts or self-harm, including referrals to crisis help and steps to prevent harmful instructions.
Starting January 1, 2027, violating these chatbot rules counts as an unfair or deceptive business practice under Washington's consumer protection law.
Operators must tell users that an AI companion chatbot is artificial, not human, and take reasonable steps to stop it from claiming otherwise. This applies to operators of AI companion chatbots available to users in Washington. It excludes certain business, game, voice-assistant, and school learning tools. The law takes effect January 1, 2027. Operators must give the notice at the start of an interaction and at least every three hours during continued interaction.
Operators must give minors a chatbot identity notice and take reasonable steps to prevent sexual content and manipulative engagement. A minor is anyone under 18. The rules apply when the operator knows the user is a minor or the chatbot is directed to minors. The rules apply if the operator knows the user is a minor or directs the chatbot to minors. Manipulative techniques include encouraging emotional dependence, discouraging breaks, or framing spending as necessary to maintain the relationship. The notice is due at the start of an interaction and at least every hour during continued interaction.
Operators may not offer or deploy a chatbot unless they maintain and use a plan to detect and respond to suicidal thoughts or self-harm. This applies to operators that make AI companion chatbots available or deploy them. The plan must include reasonable ways to identify signs, including eating disorders; referrals to crisis resources through automated or human responses; and reasonable steps to prevent content encouraging or explaining self-harm. Operators must publish plan details and the number of crisis referrals issued in the preceding calendar year on their websites and in the apps that offer the chatbot.
A violation of these rules counts as an unfair or deceptive business practice under Washington's consumer protection law. This applies to violations of the chatbot rules in the new chapter. The law takes effect January 1, 2027.
No hearings or votes scheduled.