California State Senator Scott Wiener has rewritten his artificial-intelligence proposal, SB 53, to compel the world’s largest AI developers to publish detailed safety and security protocols and to disclose any significant incidents involving their systems. The amendments, unveiled on 9 July, revive transparency requirements that were stripped from Wiener's earlier bill, SB 1047, which Governor Gavin Newsom vetoed last year amid industry opposition. Unlike its predecessor, SB 53 regulates companies rather than models exceeding a computing-power threshold. It introduces whistleblower protections for employees who believe their employer’s technology poses a “critical risk”—defined as the potential to cause more than 100 deaths or at least $1 billion in damage—and leaves the level of civil penalties to the state attorney general. The measure also would fund CalCompute, a state-run cloud cluster designed to give startups and researchers low-cost access to high-performance computing. If enacted, the legislation would make California the first U.S. state to impose binding transparency mandates on frontier developers such as OpenAI, Google, Anthropic and xAI. AI scientist Yoshua Bengio praised the proposal as a balanced step toward mitigating irreversible harms, while Wiener has opened talks with major technology firms to ease concerns that the rules could slow innovation. SB 53 now heads to the Assembly Committee on Privacy and Consumer Protection. It must clear several additional committees and floor votes before returning to Newsom’s desk later this year.