AI-generated summary, not yet reviewed by Better Utah staff. Please consult the bill text.

Utah's proposed AI Transparency Act would apply to "frontier developers"—companies that have used massive amounts of computing power to train advanced AI systems—and would require the largest of these companies, those with at least $500 million in annual revenue, to publicly post detailed safety plans explaining how they identify and manage risks of catastrophic harm, such as AI contributing to weapons attacks or major cyberattacks, as well as separate child protection plans if they operate chatbots likely to be used by minors. Before releasing a new or significantly updated AI model, these large developers would have to publish summaries of their risk assessments, report safety incidents to Utah's Office of Artificial Intelligence Policy within 15 days (or notify law enforcement within 24 hours for incidents posing imminent danger), and avoid making false or misleading statements about these risks, with violations subject to civil penalties of up to $1 million for a first offense and $3 million for repeat offenses. The bill also requires large developers to set up internal, anonymous channels for employees to report safety concerns, protects those employees from being fired, demoted, or otherwise punished for reporting concerns either internally or to the state, and allows retaliated-against employees to sue for reinstatement, double back pay, and other damages. Finally, the bill creates a restricted state account funded by penalty money to support enforcement and directs the state AI policy office to issue annual public reports and legislative recommendations on regulating these systems.