E2SHB 1170
SignedHouse
AI content notices
Informing users when content is developed or modified by artificial intelligence.
How does a bill become law?
- Introduced: The bill is filed and assigned a number.
- Committee: A subject-matter committee holds hearings, takes public testimony, and decides whether to advance the bill.
- Floor Vote: The full chamber (House or Senate) debates and votes on the bill.
- Opposite Chamber: The bill repeats the committee and floor vote process in the other chamber.
- Governor: The Governor reviews the bill and decides whether to sign or veto it.
- Signed: The bill has been signed into law.
AI Analysis
This bill requires large AI companies to make detection tools available and to embed visible and hidden tags in AI-generated images, video, or audio to identify them as AI-created. It also bans collecting personal data through the detection tools and gives the state Attorney General authority to enforce violations under consumer protection laws.
- Large AI providers (with over 1 million monthly users) must offer a free, publicly accessible tool to detect AI-generated content and show system-level origin data (but not user-identifying data).
- AI providers must collect user feedback on the detection tool and use it to improve accuracy.
- AI providers must include a visible (‘manifest’) disclosure in AI-generated content that clearly states it is AI-generated, and a hidden (‘latent’) disclosure with system name, version, timestamp, and unique ID—both must be permanent or hard to remove.
- If an AI provider licenses its system to another company, it must require that licensee keep the disclosure features intact—or the provider must revoke the license within 96 hours if the licensee removes them.
- The law does not apply to video games, TV, streaming, movies, or interactive media experiences.
Who is affected
- Large AI developers (covered providers) — Must provide AI detection tools and include visible or hidden tags in AI-generated content; must stop licensing systems if licensees remove required disclosures.
- General public and content users — Can use free tools to check if digital content was AI-generated and see system-level origin info; can choose to add visible AI labels to their own AI-generated content.
- Third-party software developers and licensees — May need to ensure contracts with AI providers include requirements to preserve AI disclosure features in licensed systems.
Pro/Con Analysis
Stronger case for benefits
Potential Benefits (5)
Mandating free, publicly accessible AI detection tools with API support and user feedback integration empowers everyday Washingtonians to independently verify the authenticity of digital content—reducing exposure to deepfakes, misinformation, and AI-generated scams, especially for non-technical users.
Public SafetyPeopleRef: Sec. 2(1)(a)-(f), Sec. 2(2)The requirement for both visible and latent disclosures in AI-generated content ensures that users can detect synthetic media even without specialized tools—protecting against impersonation, fraud, and disinformation in social media, news, and civic communication.
Public SafetyPeopleRef: Sec. 3(1), Sec. 3(2)The explicit ban on collecting personal information or personal provenance data through the detection tool—and the prohibition on retaining submitted content beyond necessity—provides meaningful privacy safeguards for users, reducing surveillance risk and misuse of sensitive user data.
Rights & LibertiesPeopleRef: Sec. 2(1)(c), Sec. 2(3)(a)-(c)Enforcement under the Consumer Protection Act (chapter 19.86 RCW) gives the Attorney General strong legal tools to penalize violations, creating a credible deterrent against bad-faith AI deployment and enabling class-action-style enforcement for widespread harms.
Public SafetyPeopleRef: Sec. 5The licensing provisions require downstream licensees to preserve disclosure features or face license revocation—this incentivizes industry-wide adoption of consistent AI labeling standards and may encourage responsible AI development among third-party integrators.
Business & EmploymentPeopleRef: Sec. 3(3)(a)-(c)
Potential Concerns (5)
The requirement for detection tools to impose 'reasonable limitations' on access to prevent 'demonstrable risks' creates ambiguity that could be exploited to restrict access to AI detection tools—potentially undermining transparency and enabling covert manipulation of AI-generated content. While intended for security, vague 'demonstrable risks' thresholds may allow providers to arbitrarily deny access, especially to marginalized or high-traffic user groups.
Public SafetyPeopleRef: Sec. 2(1)(d), Sec. 2(1)(e), Sec. 2(1)(f)The latent disclosure includes a unique identifier per piece of content, which—while not personally identifying—could be cross-referenced with other data (e.g., platform logs, metadata from other tools) to re-identify users or track behavior patterns over time. This creates a latent surveillance risk, especially for users in contexts where AI use is stigmatized (e.g., whistleblowers, journalists, or vulnerable populations).
Rights & LibertiesPeopleRef: Sec. 3(2)(a)(iv)The mandate that AI providers revoke licenses within 96 hours if licensees remove disclosures imposes operational and contractual burdens on both large providers and third-party licensees—particularly small-to-mid-sized software firms that lack legal teams to negotiate or enforce such clauses. This could reduce innovation in AI-integrated tools and discourage third-party integrations.
Business & EmploymentLean peopleRef: Sec. 3(3)(a), Sec. 3(3)(b)The 1 million monthly user threshold excludes many emerging AI startups and open-source projects that may not reach that scale but still generate harmful or misleading content—potentially creating a regulatory gap where mid-tier AI tools operate without disclosure obligations, increasing risk of misuse.
Business & EmploymentLean peopleRef: Sec. 1(2) (definition of 'covered provider' as >1M monthly users)The broad exemptions for video games, TV, and interactive media mean consumers may be exposed to undetected AI-generated content in high-engagement, emotionally immersive environments—where misleading or manipulative AI content could have outsized psychological or behavioral effects, especially on youth.
Public SafetyLean peopleRef: Sec. 4 (exemption for video games, TV, streaming, interactive media)
Who Is Most Affected
Large AI providers (e.g., OpenAI, Anthropic, Meta) must invest in detection infrastructure and compliance processes. While this increases operational costs, it may also strengthen trust and market legitimacy—especially for consumer-facing products. The 1M-user threshold likely excludes most small players, giving incumbents a regulatory moat.
Most everyday Washingtonians benefit from increased transparency and reduced exposure to AI-generated misinformation, especially in civic, educational, and personal contexts. However, some may face friction in creative workflows or content sharing if they frequently use AI tools.
Third-party developers and SaaS platforms integrating AI models must now ensure contract terms and technical integrations preserve disclosure features. This adds legal and engineering overhead, especially for small firms, but may also create demand for compliance tools and auditing services.
The Attorney General gains new authority to enforce AI transparency, expanding the scope of consumer protection law into digital media. This strengthens state capacity to respond to AI-related harms but may require new staffing or expertise.
Media creators, educators, and journalists may benefit from clearer provenance of AI content, reducing risk of misattribution or reputational harm. However, those using AI for creative or pedagogical purposes may face added friction or confusion if users distrust AI-labeled content.