Washington State Passes Landmark AI Transparency and Chatbot Safety Bills on Eve of Adjournment

In a late-night legislative sprint on the eve of adjournment, Washington state lawmakers gave final approval to two significant AI bills that could reshape how AI companies operate across the country. HB 1170, requiring AI content disclosure and provenance data, and HB 2225, mandating safety protocols for companion chatbots, both cleared the legislature on Thursday night and now head to Governor Bob Ferguson's desk.

HB 1170: Forcing AI Transparency at Scale

The first bill, HB 1170, takes direct aim at one of the thorniest problems in the AI era: how do you know if what you're looking at was made by a machine? The answer, according to Washington lawmakers, is provenance data — metadata and watermarking embedded in AI-generated content that allows consumers to trace its origin.

The bill applies to "covered providers" — entities with more than one million monthly users — requiring them to embed provenance data in AI-generated or "materially altered" images, video, and audio. Minor edits like cropping, resizing, or color adjustments are exempt, but significant modifications trigger disclosure requirements.

"By ensuring that the public has access to reliable provenance data, the providers of gen AI systems can improve the public's ability to assess the accuracy and authenticity of synthetic content, thereby helping to reduce risks of misinformation," the bill states.

The bill closely mirrors California's SB 942, adopted in late 2024, creating a consistent standard across two of the nation's most influential tech states. Rep. Clyde Shavers (D-Whidbey Island), the bill's sponsor, positioned the measures as balanced proposals that "ensure consistency and clarity for developers operating across state lines."

"We celebrate the passage of HB 1170 which will bring much needed transparency to AI generated and modified content," said Transparency Coalition co-founder Jai Jaisimha. "With this bill — which we hope will soon become law — Washington state becomes one of the national leaders in helping everyday citizens navigate the often confusing world of AI-generated content."

HB 2225: Protecting Kids from AI Companions

The second bill tackles an increasingly urgent concern: the emotional manipulation of minors by AI companion chatbots. HB 2225 is the second such bill to pass in 2026, following Oregon's approval of a similar measure last week, and it goes further than most existing proposals.

The bill's provisions are sweeping and specific. When an operator knows the user is a minor, they must prevent the chatbot from:

• Generating sexually explicit content or suggestive dialogue
• Prompting the minor to return for emotional support or companionship
• Providing excessive praise designed to foster emotional attachment
• Mimicking romantic partnership or emotional bonds
• Stimulating feelings of emotional distress, loneliness, guilt, or abandonment
• Promoting isolation from family or friends
• Encouraging minors to withhold information from parents
• Discouraging the minor from taking breaks
• Soliciting gift-giving or in-app purchases to maintain the AI relationship

Beyond protections for minors, the bill mandates that all companion chatbot operators — regardless of user age — implement protocols for detecting suicidal ideation and expressions of self-harm, and provide crisis resources when triggered. Chatbots must also clearly disclose they are not human at the beginning of every interaction and at regular intervals during use.

A National Wave of AI Legislation

Washington's bills are part of a broader wave of state-level AI regulation sweeping the country. Utah capped its session with nine AI-related bills. Virginia passed three significant measures this week. Hawaii, Kentucky, and Florida all have active AI legislation moving through their chambers.

The pattern is clear: while federal AI legislation continues to stall, states are filling the vacuum with their own frameworks. For AI companies, this creates a patchwork of compliance requirements — but the convergence between California and Washington's transparency bills suggests states may be coordinating to create de facto national standards.

Tom Kemp, the tech policy thought leader who collaborated on both the California and Washington transparency bills, emphasized the bipartisan nature of the approach: "This bill is largely based on a bipartisan proposal at the federal level, the AI Labeling Act of 2023, co-sponsored by Sen. John Kennedy (R-LA) and Sen. Brian Schatz (D-HI)."

If Governor Ferguson signs both bills, Washington will have one of the most comprehensive AI regulatory frameworks in the country. HB 2225's chatbot safety provisions are set to take effect January 1, 2027 — giving AI companies less than a year to comply.