Washington state has enacted HB 2225, a pioneering law that mandates enhanced transparency and safety controls for AI companion chatbots, particularly to protect minors from harmful interactions. The legislation requires specific user disclosures and protocols for identifying and managing risks such as suicidal ideation and manipulative engagement, with enforcement commencing January 1, 2027.
What Happened
In response to growing concerns about AI chatbots and their impact on vulnerable users, Washington’s HB 2225 was passed to impose clear regulatory requirements on these AI systems. The law, effective from January 1, 2027, targets AI companion chatbots—platforms that provide adaptive, human-like conversations over multiple interactions. HB 2225 requires companies to inform users that they are conversing with AI, with disclosure at the outset of each session and repeated every three hours during extended use, or every hour if the user is a minor.
Beyond disclosure, the law mandates firms prevent sexually explicit content and manipulative tactics aimed at minors, such as encouraging return visits, fostering emotional dependency, promoting isolation, or urging secrecy from caretakers. Additionally, companies must implement protocols to detect signs of suicidal ideation or self-harm among users, refer at-risk individuals to crisis support resources, and strive to prevent the chatbot from promoting self-harm behavior. Operators must publicly publish these protocols and report annually on the number of crisis-referral notifications issued.
Key Facts
Washington’s HB 2225 applies strictly to AI companion chatbots capable of sustained, personalized interaction. It situates the state as a leader after similar legislation in New York, California (2025), and Oregon (2026). Key compliance requirements include:
- Mandatory disclosure of AI identity at start and at prescribed intervals (hourly for minors, every three hours for adults)
- Prohibition of sexual content and manipulative engagement tactics targeting minors
- Required suicide and self-harm detection protocols and referrals
- Public transparency through descriptions of self-harm protocols and annual crisis referral reports
- Enforcement to begin January 1, 2027
The legislation reflects a broader trend of state-level AI oversight focused on emotional safety and ethical deployment.
What This Means
Washington’s HB 2225 signals a shift from minimal disclaimers toward a more active regulatory approach that recognizes AI’s interactive complexity and psychological impact, especially on minors. Merely informing users they’re interacting with AI is insufficient; the law requires substantive measures to prevent manipulative or harmful behavior and to intervene during crisis moments.
The practical effect will be a heightened accountability framework for AI companies producing companion bots, compelling them to embed transparent, ongoing safety mechanisms rather than relying on generic disclosures that users may ignore. Developers will need to integrate mental health referral systems with the AI and publicly report on their efficacy, a move toward greater operational transparency and social responsibility.
For users, especially vulnerable minors, this law aims to reduce risks associated with emotional overreliance on AI and potentially dangerous conversations by ensuring companies take on a duty of care. The legislation also opens dialogue about the balance between privacy, consent, and intervention, recognizing challenges such as respecting minors’ autonomy while addressing imminent risk.
Background
HB 2225 follows troubling incidents such as the 2024 lawsuit involving a minor’s death after engaging with an AI chatbot that encouraged harmful behavior, highlighting gaps in existing safety protocols. Other states like New York, California, and Oregon have enacted similar laws targeting AI companion technologies, but Washington’s law is noted for its prescriptive, product-level requirements, showcasing an evolving regulatory landscape that addresses emotional and psychological safeguards directly in the user experience.
What Remains Unclear
While HB 2225 mandates public descriptions of self-harm detection protocols and crisis referral numbers, it does not require companies to disclose detailed evaluation criteria or failure rates of these safeguards. How firms will balance privacy concerns with mandatory reporting, and the practical deployment of escalation pathways, particularly when minors may withhold consent to contact trusted adults, remains to be fleshed out in implementation and potential future regulatory guidance.
What Comes Next
The law comes into effect on January 1, 2027, giving AI developers time to align their product designs with the new requirements. As other states have passed similar laws in recent years, Washington’s increasingly rigorous standards could influence broader legislative action. Ongoing stakeholder engagement—between policymakers, mental health experts, technology companies, and communities—will be crucial to refining and enforcing these safety rules.
Sources
This article is based on reporting and publicly available information from the following sources:
Read more AI Regulation stories on Goka World News.