Murray urges federal AI safeguards that preserve stronger state rules

Sen. Patty Murray brought Washington state policymakers together in Olympia on October 8, 2026, to press for federal AI safeguards without displacing stronger state protections. The roundtable examined rules already adopted in Washington and the case for national oversight, with Murray rejecting voluntary company promises as an adequate response to AI risks.

The meeting was a push for congressional action, not the introduction of new federal rules. Murray called for basic bipartisan guardrails and the hearings, briefings, industry testimony and debate needed to develop them. She described federal AI regulation as “virtually nonexistent” and argued that lawmakers should act before the technology’s worst risks become reality.

Participants included Beau Perschbacher, a senior policy adviser to Gov. Bob Ferguson; Yuki Ishizuka of the state attorney general’s office; Washington chief technology officer Gretchen Peri; and Transparency Coalition CEO Rob Eleveld. Their discussion covered protections for children, identifying AI-generated media, and ways to hold developers responsible for unsafe products.

Washington’s safeguards are already on the books

Perschbacher outlined legislation Ferguson signed earlier this year. As detailed by The Spokesman-Review, the measures address different problems: recognizing synthetic media, limiting manipulative chatbot behavior, and prohibiting computer-generated sexual content involving minors. These were existing state safeguards presented to Murray, rather than new laws announced at the meeting.

• AI-created or AI-enhanced video, photos and audio must include distinguishable provenance data—information that helps identify the content as created or changed using AI.
• Chatbots targeting children must disclose that they are not human. The state also prohibits those chatbots from using manipulative engagement techniques.
• Another law bars possession, distribution or viewing of computer-generated sexually explicit conduct involving minors.

Perschbacher framed the media requirement around public confidence and deepfakes: whether people can tell what is true. He said the state worked with industry on the legislation. Peri, meanwhile, described Washington’s statewide AI strategy as giving agencies tools, guardrails and shared platforms to deploy AI safely and transparently.

We can’t afford to regulate AI with nothing but a pinky promise. We cannot just wait for the worst risks of this technology to become reality before we take action.

Sen. Patty Murray, at the October 8 Olympia roundtable

A national floor, not a replacement for states

State officials also made the case for federal rules. Ishizuka called strong federal regulation with uniform laws nationwide the best path forward and asked for bipartisan congressional leadership. Eleveld said states can protect user data through legislation, but that protection would be better handled nationally. Murray’s position pairs a federal minimum with continued state authority.

The discussion came one day after Sen. Maria Cantwell released a six-principle AI governance framework. Her proposal calls for enforceable federal safety standards and independent audits before covered advanced AI models are released. It also includes protections for children. Murray said she had not closely examined Cantwell’s proposal, while calling some of its principles “extremely important.”

Outside audits and responsibility for harm

Eleveld used the roundtable to describe the Transparency Coalition’s next priorities. In its account of the meeting, the coalition said he identified third-party AI safety and compliance audits, alongside clearer application of product liability law, as areas of focus for 2027. He said roughly 70% of its current work concerns child-protection bills, including chatbots and deepfake likenesses.

For audits, Eleveld pointed to outside scrutiny already used in finance and cybersecurity. He argued that AI systems should face similarly rigorous checks before deployment. His comparison was an argument for a regulatory requirement, not a description of a new audit mandate adopted at the roundtable.

On liability, he argued that AI models, systems and tools should be clearly treated as products whose makers have a duty of care not to harm consumers. He said product liability laws already exist, but legislative action is needed to clarify their application to AI. The approach he advocated would hold companies responsible for product safety, rather than simply prescribe what they may build.