California’s Senate Bill 53 (SB 53) is leading the charge for a new level of transparency in artificial intelligence. It further strengthens safety monitoring to safeguard users and build public confidence. Announced by Senator Scott Wiener, the bill targets the world’s largest AI model developers, aiming to impose unprecedented requirements on their operations. As the debate around AI governance intensifies, key players in the tech industry, including Anthropic, have stepped forward to endorse the initiative.
SB 53 would address the most catastrophic risks associated with AI. It narrowly focuses these risks as events that may result in a minimum of 50 deaths or property damage exceeding $1 billion. The bill emerged following recommendations from an expert policy panel convened by Governor Gavin Newsom, which was co-led by renowned AI researcher Fei-Fei Li. This new nationwide initiative represents a positive step toward helping to safeguard the rapidly evolving world of artificial intelligence.
Anthropic, a company co-founded by Jack Clark, has publicly endorsed SB 53, highlighting the urgency of establishing safety measures before federal regulations materialize. Clark noted that the technology industry will develop powerful AI systems in the coming years and cannot afford to wait for federal action.
“While we believe that frontier AI safety is best addressed at the federal level instead of a patchwork of state regulations, powerful AI advancements won’t wait for consensus in Washington.” – Anthropic
The bill as proposed provides obvious and important state law requirements for developers of AI models. It establishes serious financial penalties for traverses that fail to comply. SB 53 narrowly prohibits using genAI models to develop biological weapons and conduct cyberattacks. Our legislation puts public safety first—in fact, it’s our only priority.
Despite this burgeoning support, strong opposition has formed from a few key players. Chris Lehane, OpenAI’s chief global affairs officer, has been in regular contact with Governor Newsom. At the time, he lobbied him to not implement any rules that might drive fledgling companies out of California. Significantly, he stressed putting a collaborative environment first to allow for innovation.
Miles Brundage, a prominent voice in AI policy, responded critically to Lehane’s concerns, labeling his letter as “filled with misleading garbage about SB 53 and AI policy generally.” These divergent reactions are emblematic of the complex and often combative environment surrounding the regulation of artificial intelligence within the state.
In a huge victory for students, California legislators just amended SB 53. They walked back a provision that would require developers of AI models to undergo third-party audits. This change represents a compromise aimed at addressing some concerns raised during discussions about the bill’s feasibility and impact on innovation.
For now, Governor Gavin Newsom has not taken a clear position on SB 53 either way. His silence is especially puzzling considering his past veto of Senator Wiener’s earlier AI safety bill, SB 1047. As these discussions move forward, many everything SB 53 will be watching with hope to see it become law.
Dean Ball, a local T4A analyst who had kept a close eye on the whole legislative process, shared national advocates’ excitement at seeing SB 53 pass. He argues that the bill reflects an important first step toward setting strong governance frameworks for artificial intelligence.
“The question isn’t whether we need AI governance — it’s whether we’ll develop it thoughtfully today or reactively tomorrow. SB 53 offers a solid path toward the former.” – Anthropic
In any case, California is continuing down the path of innovative and historic legislation. What’s more, this could have significant implications for AI regulation throughout the rest of the country. The continued back and forth between a wide range of stakeholders signals a key inflection point in the unfolding story of AI safety policy.