A New Era for Transparency in Washington State

The landscape of artificial intelligence regulation in the United States has shifted significantly with Washington State’s latest legislative move. In late March 2026, Governor Bob Ferguson signed a landmark pair of bills—House Bill 1170 and House Bill 2225—that establish rigorous new standards for AI-generated media and companion chatbots. As these technologies become deeply integrated into the fabric of daily life, Washington is positioning itself at the forefront of the effort to mitigate potential risks, particularly concerning misinformation and the well-being of young users.

This legislative milestone is not merely a localized regulatory update; it serves as a bellwether for the rest of the nation. By codifying requirements for disclosure, watermarking, and specific safety protocols for minors, Washington has effectively challenged the tech industry to prioritize transparency and user safety over unrestricted deployment. For developers, stakeholders, and consumers, understanding these new mandates is essential, as they provide a glimpse into the future of operational compliance in the age of generative AI.

Decoding the Disclosure Requirements: HB 1170

House Bill 1170 represents a direct response to the growing public anxiety surrounding AI-generated misinformation and the difficulty of distinguishing between human-made and machine-generated content. As deepfakes and AI-altered media become increasingly sophisticated, the ability of the average user to verify the authenticity of visual and auditory information has plummeted.

Under the new law, large AI companies—defined by their scale of operations and user base—are now legally required to ensure that altered media created by their systems can be traced back to its artificial origin.

Mandatory Labeling and Watermarking

The legislation mandates that AI companies must identify when images, video, or audio content have been substantially modified or created using their systems. The core requirement is that this disclosure must be "commercially and technically reasonable." To meet this threshold, companies are expected to implement:

• Persistent Watermarking: Imprinting a digital watermark on images or video files that serves as a permanent, albeit potentially invisible, marker of AI involvement.
• Embedded Metadata: Incorporating traceable data within the digital file structure, which can be read by compliant software or platforms to verify the origin of the content.
• Verification Tools: Providing users with the necessary tools to detect or confirm whether content has been modified by generative AI, fostering a higher level of digital literacy and trust.

By requiring these disclosures, the state of Washington aims to curb the propagation of deceptive content. This shift effectively places the burden of proof on the developers of AI models, ensuring that the technology is accompanied by the necessary metadata to prevent it from being used to fuel disinformation campaigns or fraudulent activities.

Safeguarding Vulnerable Users: The Chatbot Mandate

While HB 1170 addresses the provenance of media, House Bill 2225 tackles the behavioral and emotional impact of "AI companion chatbots"—systems designed to mimic human conversation and foster ongoing relationships. These tools, while innovative, have raised alarms regarding their potential to manipulate users or exacerbate mental health struggles, particularly in younger demographics.

The law sets a high bar for safety, mandating that any platform offering companion-style chatbots must integrate specific safeguards designed to protect minors.

Protocol for Chatbot Interaction

The requirements for AI companion operators are bifurcated based on the user's age and the nature of the conversation. The key regulatory pillars include:

• Human-Machine Disclosure: Operators must clearly disclose that the user is interacting with a machine at the beginning of every session. For continuous interactions, this reminder must be displayed at least every three hours for general users, and more frequently—every hour—for minors.
• Prohibition of Manipulation: The law explicitly bans the use of "manipulative engagement techniques." This includes tactics such as providing excessive praise, encouraging users to withhold information from parents, or discouraging minors from taking breaks, all of which are designed to keep the user attached to the chatbot.
• Content Restrictions: For minors, the law requires reasonable measures to prevent chatbots from generating sexually explicit content, suggestive dialogue, or providing information that facilitates self-harm or eating disorders.
• Crisis Intervention: All operators must maintain protocols for detecting expressions of self-harm or suicidal ideation. If such triggers are detected, the system is required to automatically direct the user to appropriate crisis resources, ensuring that the chatbot serves as a safety net rather than a risk factor.

Comparative Overview of AI Regulatory Requirements

The following table summarizes the primary mandates established under the new Washington State legislation, contrasting the requirements for standard AI media against those for companion chatbots.

Category	Requirement	Application to Minors
AI-Generated Media	Mandatory watermarking or embedded metadata for altered files	N/A
General Chatbots	Initial disclosure and 3-hour recurrence reminder	1-hour recurrence reminder
Safety Protocols	No explicit requirement	Must ban manipulative content and sexually suggestive output
Crisis Intervention	Recommended best practice	Mandatory protocols for self-harm and suicide detection

The Broader Implications for the AI Industry

The passage of these bills signals a transition from an era of "move fast and break things" to a period of "move responsibly and comply." For companies like OpenAI, Anthropic, and other emerging AI labs, the requirements set forth by Washington necessitate a recalibration of their deployment strategies.

The industry has long argued that self-regulation is the most effective path forward. However, the legislative trend—not just in Washington, but also in states like Oregon and California—suggests that policymakers no longer find voluntary guidelines sufficient. By establishing these guardrails, Washington is effectively creating a new compliance "floor" for the industry.

Moreover, the inclusion of a private right of action in HB 2225 is particularly significant. It empowers individuals to seek legal recourse, which adds a layer of accountability that traditional regulatory oversight often lacks. Businesses that ignore these requirements face not only potential scrutiny from the state attorney general but also the risk of litigation from consumers.

Looking Ahead: Compliance and Innovation

As the January 1, 2027, effective date approaches, the technology sector faces a period of rapid adjustment. Developers must now integrate compliance-by-design principles into their AI architectures. This means that traceability, age-appropriate filtering, and crisis-detection capabilities can no longer be "add-on" features; they must be foundational elements of any consumer-facing AI product.

While some critics argue that these regulations may stifle innovation or create an undue burden for smaller startups, proponents maintain that they are essential for the sustainable growth of AI. By building a foundation of trust and safety, Washington is attempting to ensure that the adoption of generative AI does not come at the expense of public safety or truth.

As other states watch Washington’s implementation of these laws, it is highly likely that we will see a ripple effect. For Creati.ai, this shift underscores the necessity for companies to remain agile and forward-thinking. The regulatory environment is evolving, and those who prioritize user safety and transparent AI disclosure will be best positioned to lead in this new, more accountable era.