California state senator Scott Wiener is trying again to set rules for frontier AI companies. After the defeat of SB 1047 in 2024, his new bill, SB 53, takes a more targeted route: require major AI labs to disclose how they assess severe risks.
The proposal is now on Governor Newsom’s desk, awaiting a signature or veto sometime in the next few weeks. Unlike last year’s fight, this one has not triggered the same level of open conflict from Silicon Valley.
A narrower AI safety bill
SB 53 is built around transparency. If signed, it would create some of the nation’s first safety reporting requirements for AI giants such as OpenAI, Anthropic, xAI, and Google.
The bill would apply to leading AI labs making more than $500 million in revenue. Those companies would need to publish safety reports for their most capable AI models.
The reports would focus on the gravest risks described in the bill: the ability of AI systems to contribute to human deaths, cyberattacks, and chemical weapons. The source article also notes that many AI labs already publish safety reports voluntarily, but that these disclosures are not required and are not always consistent.
SB 53 also includes two additional pieces. It creates protected channels for AI lab employees to report safety concerns to government officials. It also establishes CalCompute, a state-operated cloud computing cluster intended to provide AI research resources beyond Big Tech companies.
Why SB 53 is different from SB 1047
Wiener’s earlier bill, SB 1047, faced a fierce campaign from Silicon Valley in 2024. That proposal would have made tech companies liable for potential harms caused by their AI systems.
Tech leaders argued that SB 1047 would damage America’s AI boom. Governor Gavin Newsom vetoed the bill, echoing similar concerns. Afterward, a popular AI hacker house held an “SB 1047 Veto Party,” where one attendee told TechCrunch, “Thank god, AI is still legal.”
SB 53 is less severe. It does not center on liability in the same way. Instead, it asks the largest AI companies to publicly describe how they test and think about the most serious dangers connected to their models.
That narrower design appears to have changed the politics around the bill. Anthropic endorsed SB 53 earlier this month. Meta spokesperson Jim Cullinan told TechCrunch that the company supports AI regulation that balances guardrails with innovation and said, “SB 53 is a step in that direction,” while also noting that there are areas for improvement.
Former White House AI policy adviser Dean Ball told TechCrunch that SB 53 is a “victory for reasonable voices,” and said he thinks there is a strong chance Governor Newsom signs it.
The federal-versus-state fight
Even with a softer approach, SB 53 still sits inside a larger conflict over who should regulate AI. Many in the tech industry argue that states should leave AI rules to the federal government.
OpenAI made that argument in a recent letter to Governor Newsom, saying AI labs should only have to comply with federal standards. Andreessen Horowitz also published a recent blog post suggesting that some California bills could raise concerns under the Constitution’s dormant Commerce Clause, which prohibits states from unfairly limiting interstate commerce.
Wiener’s answer is that he does not trust the federal government to pass meaningful AI safety regulation. He argues that states must act when federal action is unlikely or inadequate.
The source article describes a broader policy shift in Washington. The Trump administration has moved away from the Biden administration’s focus on AI safety and toward growth. Shortly after taking office, Vice President J.D. Vance appeared at an AI conference in Paris and said: “I’m not here this morning to talk about AI safety, which was the title of the conference a couple of years ago. I’m here to talk about AI opportunity.”
Silicon Valley has welcomed that shift. The article points to Trump’s AI Action Plan, which removed barriers to building the infrastructure needed to train and serve AI models, as well as Big Tech CEOs dining at the White House or announcing hundred-billion-dollar data centers alongside President Trump.
What Wiener says he is trying to solve
Wiener frames SB 53 as an attempt to encourage safe innovation, not stop technology development. He represents San Francisco, which he describes as “the beating heart of AI innovation,” and says he is not anti-tech.
His concern is that the largest technology companies should not be expected to regulate themselves through voluntary commitments alone. In his view, capitalism can create prosperity while also producing harm when no sensible public-interest rules exist.
Wiener also says SB 53 is not meant to address every AI risk. He specifically distinguishes catastrophic risks from other issues such as algorithmic discrimination, job loss, deep fakes, and scams. Other California bills, he notes, address other categories of risk, including engagement-optimization techniques in AI companions.
The focus of SB 53 came, according to Wiener, from people in the AI space in San Francisco, including startup founders, frontline AI technologists, and people building models. He said they told him, “This is an issue that needs to be addressed in a thoughtful way.”
The decision now sits with Governor Newsom
The bill’s fate is now in Governor Newsom’s hands. Wiener’s message to him is that the new proposal was shaped in response to the governor’s earlier veto of SB 1047.
Wiener said Newsom gave a comprehensive veto message, convened a working group, and produced a report that helped guide the drafting of SB 53. His argument is simple: the governor laid out a path, and SB 53 was written to follow it.
For California, the question is whether transparency is a workable first step for AI safety. For the largest AI labs, the question is whether voluntary safety disclosures are about to become a legal requirement.