California lawmaker behind SB 1047 reignites push for mandated AI safety reports

California State Senator Scott Wiener has reignited discussions around AI safety with his latest bill, SB 53, which proposes that major AI firms must disclose their safety measures and report incidents that compromise safety. If enacted, this legislation would position California as the first state to enforce significant transparency protocols for leading AI developers, including companies like OpenAI, Google, Anthropic, and xAI. Previously, Senator Wiener introduced a similar initiative, SB 1047, which faced fierce opposition from Silicon Valley and was ultimately vetoed by Governor Gavin Newsom. In response, the Governor established a task force comprising AI experts, including Stanford researcher Fei Fei Li, to formulate recommendations for enhancing AI safety in the state. Their recent report emphasized the necessity for industry transparency regarding AI systems. In a press release, Senator Wiener noted that the changes in SB 53 were heavily influenced by this task force's findings. He expressed his commitment to refining the bill in collaboration with stakeholders to ensure it is scientifically sound and equitable. The focus of SB 53 is to balance meaningful transparency requirements with the need to foster growth within California’s AI sector—something Governor Newsom felt previous efforts had not accomplished. Nathan Calvin, the VP of State Affairs at the nonprofit AI safety organization Encode, remarked that it’s essential for companies to outline their risk mitigation strategies to the public and regulatory bodies as a fundamental step forward. Additionally, SB 53 includes protections for whistleblowers who report critical risks associated with AI technologies, defined as circumstances leading to substantial human harm or significant financial loss. The new bill also aims to establish CalCompute, a public cloud computing infrastructure to assist startups and researchers engaged in large-scale AI development. Unlike its predecessor, SB 53 does not hold AI developers liable for the impacts of their models and is designed to alleviate any potential burden on smaller companies experimenting with AI. Currently, SB 53 awaits approval from the California State Assembly Committee on Privacy and Consumer Protection. If it clears this hurdle, it will move through additional legislative processes before reaching the Governor’s desk. Meanwhile, New York Governor Kathy Hochul is contemplating a parallel bill, the RAISE Act, which would impose similar reporting requirements on large AI firms. The future of state-level AI regulations faced uncertainty recently when federal lawmakers considered a ten-year moratorium on such legislation. However, that proposal was overwhelmingly rejected, clearing the way for states like California to take the lead on AI transparency. Geoff Ralston, the former president of Y Combinator, voiced his support for state-led initiatives, emphasizing that safety in AI development should be a priority. Despite the ongoing challenges in garnering cooperation from AI firms, the push for transparency continues, with SB 53 marking a significant step in the ongoing dialogue about responsible AI development.