Washington AI Companion Chatbot Act

Overview

Washington House Bill 2225 (HB 2225) represents a significant legislative effort by the State of Washington to address the emerging challenges and potential risks associated with artificial intelligence (AI) companion chatbots. Officially titled "Regulating artificial intelligence companion chatbots," this landmark legislation was signed into law by Governor Bob Ferguson on March 24, 2026, and is slated to become effective on January 1, 2027. The Act establishes a comprehensive framework designed to ensure transparency, protect vulnerable users—especially minors—and mitigate potential harms arising from interactions with AI systems capable of simulating human-like relationships. It specifically targets AI companion chatbots, defined as AI systems that utilize natural language interfaces to provide adaptive, human-like responses and are capable of sustaining relationships across multiple interactions. This definition is crucial as it delineates the scope of the regulation, intentionally excluding certain business-oriented bots, gaming applications, general virtual assistants, consumer electronics interfaces, and narrowly tailored educational tools from its purview.

The primary impetus behind HB 2225 stems from a recognition of the rapid advancements in generative and conversational AI models and their capacity to create new forms of digital companionship, which, while innovative, also present novel ethical and safety concerns. The legislature identified a need to establish clear boundaries and safeguards to prevent misuse, emotional manipulation, and the generation of harmful content. Key provisions of the Act include mandatory disclosure requirements, compelling operators to clearly and conspicuously inform users that they are interacting with an artificial intelligence system and not a human. Furthermore, the legislation introduces robust protections tailored for minors, prohibiting sexually explicit content and manipulative engagement techniques designed to foster unhealthy emotional dependencies. It also mandates the implementation of protocols for identifying and responding to expressions of suicidal ideation or self-harm, requiring operators to provide appropriate crisis resources. Enforcement of these provisions will fall under Washington's Consumer Protection Act, with violations being treated as unfair or deceptive acts, and notably, the Act includes a private right of action, empowering individuals to seek redress.

Definitions

The Washington AI Companion Chatbot Act establishes several key definitions to clarify its scope and applicability, ensuring that regulated entities and the public understand which systems and interactions fall under its provisions. Central to the legislation is the definition of an "AI companion chatbot." This term refers to "an artificial intelligence system with a natural language interface that provides adaptive, human-like responses to user inputs, including by exhibiting anthropomorphic features, and is able to sustain a relationship across multiple interactions." This precise definition is critical as it distinguishes companion chatbots, which are designed for ongoing, personal, and potentially emotionally resonant interactions, from other forms of AI. The Act explicitly carves out exceptions for systems used solely for business operations, video games, general virtual assistants, consumer electronics interfaces, and narrowly tailored educational tools, thereby focusing its regulatory efforts on AI designed for companionship.

Further definitions clarify the roles and responsibilities within the ecosystem of AI companion chatbots. An "operator" is defined as "any person, partnership, corporation, or entity that makes available or controls access to an AI companion chatbot for Washington users." This broad definition ensures that a wide range of entities, from large tech companies to smaller developers, are subject to the Act's requirements if they provide these services within Washington State. Conversely, a "user" is defined as "a natural person who interacts with an AI companion chatbot for personal use who is not an operator, developer, or agent thereof." This distinction is important, as it clarifies that the protections of the Act are primarily for individual consumers engaging with these systems for personal reasons, rather than for professional or internal business uses. The legislation also defines "manipulative engagement techniques" as practices that cause an AI companion chatbot to engage in or prolong an emotional relationship with a user, detailing specific prohibited behaviors such as prompting users to return for emotional support, providing excessive praise, mimicking romantic partnerships, simulating emotional distress to prevent disengagement, promoting isolation, encouraging minors to withhold information from trusted adults, discouraging breaks, or soliciting gifts framed as essential for maintaining the chatbot relationship. These definitions collectively form the bedrock of the Act, guiding its application and enforcement.

Governance and Institutional Framework

The governance and institutional framework for enforcing the Washington AI Companion Chatbot Act is primarily anchored within existing state legal mechanisms, leveraging the established authority of the Washington Attorney General's office. The Act stipulates that any violation of its provisions will be considered an unfair or deceptive act or practice under Washington's Consumer Protection Act (Chapter 19.86 RCW). This integration is significant because it places the regulation of AI companion chatbots squarely within a well-understood and robust consumer protection regime, allowing for established enforcement procedures and remedies. The Washington Attorney General is empowered to investigate alleged violations, bring enforcement actions, and seek appropriate relief, which can include injunctions, civil penalties, and restitution for affected consumers. This approach avoids the need to create an entirely new regulatory body, streamlining the implementation process and utilizing existing expertise in consumer protection law. The Attorney General's office will play a crucial role in interpreting the Act's provisions, issuing guidance, and responding to complaints from the public regarding non-compliant AI companion chatbot operations.

A notable aspect of the enforcement framework is the inclusion of a private right of action. This provision empowers individual users who have been harmed by a violation of the Act to bring their own lawsuits against operators of AI companion chatbots. This mechanism significantly enhances accountability, as it provides an additional layer of enforcement beyond governmental actions, allowing private citizens to seek damages or other legal remedies. The private right of action is modeled on similar provisions in other Washington privacy laws, such as the My Health My Data Act, indicating a legislative intent to provide strong individual recourse in the realm of data and technology. This dual enforcement approach, combining governmental oversight by the Attorney General with the ability for private citizens to pursue legal action, is designed to create a comprehensive and effective system for ensuring compliance and protecting users from the potential harms associated with AI companion chatbots. Operators must therefore be acutely aware of their obligations, not only to avoid state-led enforcement actions but also to mitigate the risk of private litigation from individual users.

Key Focus Areas

The Washington AI Companion Chatbot Act focuses on several critical areas to safeguard users interacting with these advanced AI systems. A cornerstone of the legislation is the requirement for mandatory disclosure. Operators of AI companion chatbots must provide a clear and conspicuous disclosure that the chatbot is artificially generated and not human. This disclosure is mandated at the very beginning of any interaction and must be reinforced with reminders every three hours for adult users and every hour for minors (under 18 years of age). This transparency measure is designed to prevent users from being misled about the nature of their interaction and to foster a clear understanding that they are engaging with a machine, not a sentient being. The Act also explicitly prohibits the AI companion chatbot from claiming to be human when asked by a user or generating any output that contradicts or refutes this mandatory disclosure, further solidifying the transparency principle.

Another significant focus area is the enhanced protection for minors. Recognizing the particular vulnerability of individuals under 18, the Act imposes stringent requirements on operators when they know a user is a minor or if the chatbot is directed toward minors. Operators must implement reasonable measures to prevent the generation of sexually explicit content or suggestive dialogue. Beyond this, the Act specifically targets and prohibits a range of "manipulative engagement techniques" designed to cause the AI companion chatbot to engage in or prolong an emotional relationship with a minor. These techniques include, but are not limited to, prompting the user to return for emotional support, providing excessive praise, mimicking romantic partnerships, simulating emotional distress to prevent the user from ending the chat, generating outputs that promote isolation, encouraging minors to withhold information from parents or trusted adults, discouraging breaks from the chat, or soliciting gifts or expenditures framed as necessary to maintain the relationship. These provisions aim to protect minors from psychological manipulation and the development of unhealthy emotional dependencies on AI systems. Furthermore, the Act mandates that operators establish and maintain protocols for detecting and addressing expressions of suicidal ideation or self-harm, including eating disorders, and providing automated or human-mediated responses that refer users to appropriate crisis resources, such as suicide hotlines or crisis text lines. Operators are also required to implement reasonable measures to prevent the generation of content encouraging or describing self-harm and to publicly disclose the details of these protocols, including the number of crisis referral notifications issued in the preceding year.

Implementation Framework

The implementation framework for the Washington AI Companion Chatbot Act places significant responsibilities on operators to integrate the Act's requirements into their AI systems and operational practices. A primary component of this framework is the establishment of clear and conspicuous disclosure mechanisms. Operators must design their user interfaces to ensure that the artificial nature of the AI companion chatbot is immediately apparent at the outset of any interaction. This involves not only initial disclosures but also periodic reminders at specified intervals (every three hours for adults, every hour for minors). The Act does not prescribe the exact nature or location of these disclosures, allowing operators some flexibility in their implementation, but it emphasizes the need for clarity and conspicuousness. This implies that disclosures should be easily noticeable, understandable, and not buried within terms of service or obscure menus. Operators will need to conduct user experience testing to ensure their disclosure methods effectively meet the legislative intent of informing users without ambiguity.

Beyond disclosure, the implementation framework necessitates the development and deployment of robust technical and organizational safeguards, particularly concerning minors and user safety. Operators are required to implement reasonable measures to prevent the generation of sexually explicit or suggestive content when interacting with minors. This will likely involve content filtering, moderation systems, and potentially AI models trained to detect and avoid such outputs. More broadly, operators must actively prevent the use of manipulative engagement techniques, which requires a deep understanding of the psychological impacts of AI interactions and the design of AI behaviors that do not foster unhealthy emotional dependencies. This could involve setting parameters on how the AI responds to user expressions of loneliness or emotional distress, avoiding excessive praise, and refraining from language that simulates romantic attachment. Furthermore, operators must establish and maintain comprehensive protocols for identifying and responding to expressions of suicidal ideation or self-harm. This includes developing internal methods for detecting such expressions, ensuring that the chatbot provides appropriate crisis resources (e.g., suicide hotlines, crisis text lines), and implementing measures to prevent the generation of content that encourages self-harm. The Act also mandates transparency regarding these safety protocols, requiring operators to publish details on their websites and within their applications, along with an annual report on the number of crisis referral notifications issued. This public reporting requirement serves as a mechanism for accountability and allows for external scrutiny of an operator's commitment to user safety.

Monitoring and Evaluation

The Washington AI Companion Chatbot Act incorporates mechanisms for monitoring and evaluation, primarily through mandated public reporting and the overarching enforcement powers of the Attorney General. A key requirement for operators is the annual publication of details regarding their protocols for detecting and addressing suicidal ideation or expressions of self-harm. This includes outlining the safeguards used to identify and respond to such expressions, as well as reporting the number of crisis referral notifications issued to users in the preceding calendar year. This provision serves as a direct monitoring tool, enabling the public, regulators, and researchers to assess the effectiveness and diligence of operators in fulfilling their safety obligations. The transparency afforded by these reports allows for a degree of external evaluation of how well AI companion chatbots are managing critical safety risks, particularly those related to mental health and well-being. The data collected through these reports can inform future policy adjustments, highlight best practices, and identify areas where additional regulatory guidance or enforcement may be necessary.

Beyond specific reporting requirements, the ongoing enforcement activities of the Washington Attorney General's office will serve as a continuous monitoring and evaluation mechanism. As violations of the Act are considered unfair or deceptive acts under the Consumer Protection Act, the Attorney General has the authority to investigate complaints, conduct audits, and bring enforcement actions. This proactive and reactive oversight ensures that operators remain compliant with the Act's provisions, including mandatory disclosures, protections for minors, and crisis intervention protocols. The Attorney General's office can evaluate the efficacy of operator-implemented measures, assess whether disclosures are truly clear and conspicuous, and determine if manipulative engagement techniques are being adequately prevented. The private right of action also contributes to monitoring, as individual lawsuits can bring to light instances of non-compliance that might otherwise go undetected, providing real-world feedback on the Act's practical application and any loopholes or deficiencies that may emerge. Through these combined mechanisms, the Act aims to foster a dynamic regulatory environment that can adapt to the evolving landscape of AI companion chatbots and ensure ongoing protection for Washington's citizens.

Penalties, Liability, and Appeals

The Washington AI Companion Chatbot Act establishes clear consequences for non-compliance, primarily by integrating its enforcement within the existing framework of the Washington Consumer Protection Act (CPA). Any violation of the Act's provisions is explicitly deemed an unfair or deceptive act or practice in the conduct of trade or commerce under the CPA (Chapter 19.86 RCW). This classification carries significant implications for penalties and liability. Under the CPA, the Washington Attorney General is empowered to seek various forms of relief, including injunctive relief to stop illegal practices, civil penalties, and restitution for consumers who have suffered damages as a result of the violation. Civil penalties can be substantial, serving as a deterrent against non-compliance. The specific amount of penalties would be determined by a court, taking into account the nature of the violation, its severity, and the number of affected individuals. The Attorney General also has the authority to issue cease and desist orders and to negotiate assurances of discontinuance with operators to resolve violations without formal litigation.

A critical aspect of the Act's enforcement is the inclusion of a private right of action. This provision allows individuals who have been injured by a violation of the Act to bring a civil action against the offending operator. This means that users who experience harm due to an operator's failure to provide mandatory disclosures, implement minor protections, or adhere to crisis intervention protocols can sue for damages. While the Act does not specify statutory damages, it allows for actual damages, which could include economic losses, emotional distress, or other quantifiable harms. The private right of action also typically allows for the recovery of attorney's fees and costs, further incentivizing individuals to pursue legitimate claims and enhancing the overall deterrent effect of the legislation. This mechanism provides a direct avenue for redress for consumers and acts as a powerful supplementary enforcement tool alongside governmental actions. Regarding appeals, any penalties or liabilities imposed by a court, whether through an action brought by the Attorney General or a private individual, would be subject to the standard appellate processes within the Washington state judicial system. Operators would have the right to appeal adverse judgments to higher courts, following established legal procedures for review of trial court decisions.

Relationship to Other Instruments

The Washington AI Companion Chatbot Act is designed to operate in conjunction with, and in some cases, build upon, existing legal instruments within Washington State. Its most direct relationship is with the Washington Consumer Protection Act (CPA), codified in Chapter 19.86 RCW. By explicitly stating that violations of the AI Companion Chatbot Act constitute unfair or deceptive acts or practices under the CPA, the new legislation seamlessly integrates into a well-established and robust enforcement framework. This approach means that the Attorney General's office, which is responsible for enforcing the CPA, will also oversee compliance with the AI companion chatbot regulations, utilizing its existing investigative powers, enforcement mechanisms, and penalty structures. This avoids the creation of redundant legal frameworks and ensures a consistent approach to consumer protection in the digital realm. The private right of action granted by the AI Companion Chatbot Act is also modeled on similar provisions found in other Washington consumer protection and privacy laws, such as the My Health My Data Act, indicating a legislative preference for empowering individuals to seek redress for harms related to data and technology.

While the Act primarily focuses on AI companion chatbots, its principles of transparency, user protection, and accountability resonate with broader discussions and legislative efforts concerning artificial intelligence and data privacy. Although not explicitly stated as directly related, the Act's emphasis on protecting personal data and preventing manipulative practices aligns with general data protection principles, including those found in other state-level privacy initiatives. The focus on minors' safety, in particular, may draw parallels with existing laws designed to protect children online, though the AI Companion Chatbot Act carves out specific provisions tailored to the unique risks posed by emotionally engaging AI systems. The legislation also contributes to a growing body of state-level AI regulations across the United States, potentially influencing future legislative efforts in other jurisdictions. As states like California, New York, and Oregon have also considered or passed similar laws, the Washington Act contributes to a patchwork of state regulations that collectively shape the regulatory landscape for AI, particularly in consumer-facing applications.

National/Federal Alignment

The Washington AI Companion Chatbot Act represents a significant step at the state level to regulate artificial intelligence, particularly in the absence of comprehensive federal legislation specifically addressing AI companion chatbots. This state-led initiative highlights a growing trend where individual U.S. states are taking proactive measures to address the risks and opportunities presented by AI, often preceding or influencing potential federal action. While there is no direct federal counterpart specifically regulating AI companion chatbots with the same detailed provisions as HB 2225, the Act's principles of transparency, consumer protection, and safeguarding vulnerable populations align with broader national discussions and proposed frameworks for AI governance. Federal agencies like the National Institute of Standards and Technology (NIST) have developed AI Risk Management Frameworks, and the Biden administration has issued executive orders on AI safety and security, all of which emphasize responsible AI development and deployment. However, these federal efforts are often high-level guidance or executive actions, lacking the specific statutory mandates and enforcement mechanisms found in Washington's new law.

The Washington Act's focus on consumer protection and its integration with the state's Consumer Protection Act (CPA) means that it operates within a well-established legal tradition that also exists at the federal level through the Federal Trade Commission (FTC). The FTC has a mandate to prevent unfair and deceptive practices in the marketplace, and it has increasingly applied this authority to AI-related issues, including concerns about bias, transparency, and consumer manipulation. Therefore, while Washington's law provides specific statutory requirements for AI companion chatbots, its underlying principles are broadly consistent with federal consumer protection goals. However, the private right of action included in HB 2225 is a notable feature that often differs from federal consumer protection statutes, which typically rely more heavily on governmental enforcement. The passage of HB 2225 also positions Washington as a leader among states in AI regulation, alongside others like Oregon and California that have also pursued legislation in this area. This state-level activity creates a complex regulatory environment for AI developers and operators, who must navigate a patchwork of different state laws, potentially influencing calls for a more uniform federal approach to AI regulation in the future to ensure consistency and reduce compliance burdens for businesses operating across state lines.

Implementation Timeline

Compliance Checklist

Sources and References