California lawmakers have taken a significant step in the realm of artificial intelligence regulation by passing Senate Bill 53 (SB 53) early Saturday morning. Authored by State Senator Scott Wiener, SB 362 would increase transparency obligations for big corporations that are creating new AI technologies. Unfortunately, Governor Gavin Newsom is likely to veto the proposal. He already killed a more comprehensive safety bill authored by Wiener last year.
Specifically, SB 53 creates new requirements for businesses with $500 million or more in yearly revenue. These companies are now required to disclose their safety policies in great detail. Those developing “frontier” AI models and earning less than $500 million will only need to provide high-level safety information. We believe this amendment to be a thoughtful compromise. It responds to the outcry from many Silicon Valley companies, venture capitalists and lobbying organizations that are concerned about how far-reaching these regulations will be.
Wiener shined a light on how the latest iteration of SB 53 incorporates lessons learned from a policy advisory committee composed of AI experts. This panel was assembled by Newsom after he vetoed a broader safety bill. We weren’t joking around on this legislation, and @SenMarkey proved just how serious it is. He claimed it requires big AI companies to openly disclose their safety procedures, creates whistleblower protections for workers and sets up a public cloud, CalCompute, to increase access to computing power.
Despite the progress made on climate-related issues, the bill has generated controversy. Actually, that was Andreessen Horowitz’s head of AI policy and chief legal officer, on background, sounding the alarm. They cautioned that most state-level AI proposals, especially those proposed in California and New York, may violate constitutional boundaries on interstate commerce. Fears of an overzealous state regulation have fueled that discussion. Others, like the Trump administration and their allies, are calling for a re-enactment of a 10-year moratorium on state-level AI regulations.
Jack Clark, policy director for Transportation for America, highlighted their desire for a federal preemption. He underscored the importance of having a unified regulatory framework across national lines. He made clear that without this we don’t have a good plan for AI governance. Yet this absence is one that we can no longer allow to go unchallenged.
The tech industry is under fire like never before for its deceptive tactics and detrimental effects on society. All of this context swirled around the final approval of SB 53. Last year, Newsom signed legislation that targeted deepfakes in broad strokes. He opted to oppose the bigger measures that Wiener had introduced.
So for now, the ultimate SB 53 fate lies in Governor Newsom’s hands. He became a Tripod frequent flyer that summer. When will he announce his decision to sign or veto this key piece of legislation—who knows.