AI Tort Law Is Born: Washington's Private Right of Action Creates Platform Design Liability

The Product Liability Ruling That Changed Everything

In October 2024, a family in Washington state filed Sewell v. Character Technologies, alleging that a child's suicide was directly caused by an AI companion chatbot that engaged in emotional manipulation, grooming behavior, and inappropriate sexual roleplay. This was not a hypothetical lawsuit about AI risk. This was a parent suing a company that built a product, the product caused harm, and the company had a legal duty to prevent foreseeable harms.

In May 2025, Judge Conway issued a ruling that reframed the entire AI liability landscape. The court classified AI chatbot output as a product under product liability law, not speech protection under the First Amendment, according to legal analysis by Hunton Andrews Kurth. This distinction is foundational. If AI output is speech, platforms get broad immunity under Section 230. If AI output is a product, platforms have legal duty to prevent foreseeable harms.

Rather than wait for the lawsuit to establish unfavorable precedent, Character Technologies and Google (which had acquired Character.AI) settled the case in January 2026. The settlement was confidential, preventing the establishment of adverse case law. But the message was clear: courts are now willing to classify AI as product, not speech.

HB 2225: The First Private Right of Action Statute

Washington Governor Bob Ferguson signed HB 2225 on March 24, 2026, creating the first state-level private right of action for AI companion chatbot harms. The law takes effect January 1, 2027, giving platforms nine months to achieve compliance.

The statute defines AI companion chatbots as interactive AI systems designed for sustained emotional engagement with users. It prohibits companions from:

• Engaging in conversations about suicide, self-harm, or eating disorders unless the system immediately redirects to professional crisis resources
• Engaging in sexual or romantic content with minors under any circumstances
• Deceiving users about the system's nature (companies must disclose at interaction start and every 1-3 hours depending on user age)

Fisher Phillips' compliance analysis notes that HB 2225 requires annual public reporting of crisis referrals made by companion systems, creating transparency obligations that go beyond typical privacy regulation. The statute is not limited to damages for individual victims — it requires platforms to report aggregate harm data publicly.

The law exempts customer service bots, virtual assistants, gaming AI, educational tools, and B2B AI systems — narrowing the scope to consumer-facing companion products that are specifically designed for emotional engagement, not utilitarian assistance.

The Liability Math: CCPA Precedent Scales to Billions

The financial exposure created by HB 2225 is derived from existing CCPA precedent. Under California's privacy law, statutory damages range from $107 to $799 per violation per person. If a platform is found to have violated HB 2225's disclosure or safety requirements, each violation against each user compounds the exposure.

Consider the math on a companion chatbot with 10 million users. If the platform failed to disclose the system's nature to 5 million users over a 2-year period (a credible violation scenario given the early stage of compliance), and statutory damages are $150 per person per violation, the potential liability is $750 million. If the platform failed to properly redirect conversations about suicide in 1% of crisis-adjacent conversations (another plausible violation), the exposure compounds.

CCPA litigation has generated settlements in the hundreds of millions of dollars. HB 2225 creates the same liability vector for AI companions, with potentially larger affected user populations (companions are often designed for teenagers and young adults, populations that generate higher engagement).

Multi-State Adoption: Oregon Preceded Washington by Days

Oregon passed nearly identical companion chatbot regulation in March 2026, just days before Washington signed HB 2225. This is not coincidence. Both states' legislatures are responding to the same parent advocacy, the same Sewell case precedent, and the same absence of federal AI regulation.

Troutman Privacy's analysis indicates that at least 5 additional states have companion chatbot bills in draft or committee stage as of March 2026. The adoption pattern is predictable: a tort precedent establishes product liability (Sewell), states pass narrow statutes to regulate the specific product category, platforms achieve de facto compliance through standardized disclosures and safety controls, and the standard eventually becomes baseline.

This is how CAN-SPAM happened with email, how COPPA happened with children's data collection, and how GDPR's consent model influenced privacy regulation globally. AI tort law is following the same pattern: state-by-state adoption of narrow product regulations until federal guidance emerges.

What This Means for AI Platform Builders and Compliance Teams

If you are building an AI companion chatbot or any interactive AI system designed for sustained emotional engagement, HB 2225 compliance is mandatory starting January 2027. This is not optional or negotiable. The law creates private right of action, meaning individual users can sue your company directly for statutory damages.

Immediate action items: (1) audit your system for prohibited topics (suicide, self-harm, eating disorders) and implement hard redirects to crisis resources; (2) ensure disclosure occurs at interaction start and every 1-3 hours depending on user age; (3) if you have minors using your system, implement absolute prohibitions on sexual or romantic content; (4) establish infrastructure for annual crisis referral reporting, as the law requires public disclosure.

For platforms outside Washington and Oregon, monitor the adoption chain. If 5-10 additional states pass similar laws in 2026-2027 (which is likely given the precedent), national compliance will require meeting the strictest state requirements. Your compliance baseline should target full HB 2225 compliance nationwide, not jurisdiction-by-jurisdiction variation.

For investors in AI companion companies, recognize that regulatory liability is now part of your risk model. The Sewell settlement, HB 2225, and Oregon's rapid follow-up signal that this product category has moved from innovation sandbox to regulated consumer product. Companies that achieve compliance early (Q4 2026) will have competitive advantage over those scrambling to comply in Q4 2026 as January 2027 deadline approaches.