Committee considers bill that would require large AI providers to offer detection tools and label synthetic images, audio and video

Committee staff described House Bill 1170 as a narrower, consumer‑facing measure that would require large generative AI providers to provide detection tools and disclosures for synthetic image, audio and video content.

Emily Pool, nonpartisan staff, summarized the bill’s main requirements: any covered provider with more than 1,000,000 monthly visitors or users and that is publicly accessible in Washington must make available an AI detection tool at no cost; the tool must allow users to assess whether content was created or altered by the provider’s system, be publicly accessible and permit use without visiting the covered provider’s website. Covered providers must also offer users an option to include a manifest or an easily perceived disclosure on any image, video or audio the provider’s system creates or alters. To the extent feasible, the bill requires a latent disclosure — metadata readable by the provider’s tool — that includes the covered provider’s name, time and date of creation and a unique identifier. Violations would be treated as unfair or deceptive acts under the Consumer Protection Act.

Representative Clyde Shavers, the prime sponsor, described HB 1170 as a “narrowed approach, a common sense approach” focused on large providers and on visual and audio media rather than text. Tom Kemp, who advised the California author of a similar bill, said the bill is aligned with the AI Labeling Act and federal and EU efforts and that technical mechanisms and industry standards (such as content provenance efforts) make implementation feasible. Kemp and other proponents said the bill is intended to give consumers tools — including an API for detection checks — to determine whether content is synthetic.

Industry and civil‑liberties groups voiced objections. Robert Singleton of Chamber of Progress and other industry witnesses argued the bill may require technology that does not yet exist to the required standard and that broad definitions could sweep in many consumer experiences. TechNet and trade associations objected to the bill’s private‑enforcement pathway via the Consumer Protection Act and urged additional trade‑secret safeguards and clearer scope. John Coleman of the Foundation for Individual Rights and Expression (FIRE) said compelled disclosures raise First Amendment concerns and warned that mandatory, permanent labels could stigmatize lawful speech while offering circumventing routes to bad actors.

Committee members exchanged questions with staff and witnesses about threshold levels, the efficacy of proposed detection tools, whether the bill’s requirements are technically feasible, and whether the Consumer Protection Act is an appropriate enforcement vehicle. Witnesses said the bill targets the largest providers but acknowledged technical tradeoffs: detection tools tied to providers’ own metadata would be easier to implement than a general‑purpose AI detector, but no detection regime is foolproof.

No committee vote was taken at the hearing. Members asked for further stakeholder engagement and signaled plans for an AI work session to examine technical feasibility and enforcement options.

Ending: The committee closed the hearing on House Bill 1170 and announced plans for a technical AI work session to follow.