Panel urges stronger state action on AI, calls for transparency and enforcement

A panel of legal and civil‑rights experts said state governments should adopt stronger, enforceable rules for artificial intelligence that prioritize transparency, enforcement and protections for people subject to automated decisions.

"We need legislation with broad and clear definitions, robust transparency requirements, and no loopholes," said Matt Shearer, senior policy counsel for workers' rights and technology at the Center for Democracy & Technology. He warned that many state bills amount to "legislative virtue signaling" that leaves the public unprotected.

The panel — convened and moderated by Steve Tapia — discussed a range of near‑term tools. Jayashima, a Seattle‑based technology entrepreneur who co‑founded the Transparency Coalition, pointed to recent California measures requiring a "nutrition label disclosure for training data" and rules to label AI‑generated content. He singled out California AB 412 as an effort to give copyright holders visibility into whether and how their works are used to train models.

"One of the laws we helped enact was a nutrition label disclosure for training data," Jayashima said, adding that labeling and dataset transparency are "necessary, but not sufficient" and that product‑liability frameworks may be needed to give victims a path to remedy.

Bisma Shoaib, an attorney who said she was speaking personally and has experience at the Federal Trade Commission, framed AI concerns within competition law. "There is no AI exception to the antitrust laws," she said, and cautioned that control of critical inputs—data, compute and specialized labor—can lock in market power and stifle innovation.

T Shannon, technology policy program director at the ACLU of Washington, highlighted civil‑rights harms that are already occurring: mass data scraping, surveillance without consent, biased facial‑recognition and algorithmic decisions that can worsen discrimination. Shannon recommended mandatory impact assessments, independent audits for high‑risk uses, and enforcement mechanisms that include private rights of action and dedicated funding for regulators.

Panelists debated enforcement strategy. Matt Shearer argued that strong penalties matter little if violations are detected and enforced only rarely: enforcement must be frequent enough to change behavior. He also proposed a more controversial option: internal mandatory reporting inside companies—similar to anti‑money‑laundering reporting—so regulators learn about unlawful uses.

Panelists agreed on some practical steps states can take now: require disclosure of AI use and high‑risk impact assessments, fund enforcement agencies so they can audit and prosecute violations, and adopt targeted bans or limits (for example, the panel noted King County’s facial‑recognition restrictions) for uses where harms clearly outweigh benefits.

The session closed with the moderator thanking panelists and attendees. The discussion did not include any formal votes or government actions; it was a policy and legal roundtable exploring legislative and enforcement options being pursued in states, including California.