California is again testing how far a state can go in demanding transparency from the companies building the most powerful AI systems. New amendments to SB 53 would require the world’s largest AI companies to publish safety and security protocols and issue reports when safety incidents occur.
If signed into law, California would become the first state to impose meaningful transparency requirements on leading AI developers, likely including OpenAI, Google, Anthropic, and xAI. The bill is a new attempt from California State Senator Scott Wiener, whose earlier AI bill, SB 1047, included similar safety-reporting requirements before it was ultimately vetoed by Governor Gavin Newsom.
A narrower return to AI transparency rules
SB 53 follows a bruising fight over SB 1047. Silicon Valley opposed that earlier bill strongly, and Governor Gavin Newsom ultimately vetoed it. Afterward, California’s governor called for a group of AI leaders, including the leading Stanford researcher and co-founder of World Labs, Fei-Fei Li, to form a policy group and set goals for the state’s AI safety efforts.
That policy group recently published final recommendations. The report cited a need for “requirements on industry to publish information about their systems” so California could build a “robust and transparent evidence environment.” Senator Wiener’s office said the new SB 53 amendments were heavily influenced by that report.
The political goal is clear: keep pressure on major AI labs to explain how they manage safety risks, while avoiding the parts of SB 1047 that drew the strongest opposition. SB 53 aims to create transparency requirements for the largest AI developers without thwarting the rapid growth of California’s AI industry.
“The bill continues to be a work in progress, and I look forward to working with all stakeholders in the coming weeks to refine this proposal into the most scientific and fair law it can be,” Senator Wiener said in the release.
What SB 53 would require
The central requirement in SB 53 is disclosure. The world’s largest AI companies would have to publish safety and security protocols. They would also need to issue reports when safety incidents occur.
Supporters frame that as a basic condition for public trust. Nathan Calvin, VP of State Affairs for the nonprofit AI safety group, Encode, told TechCrunch that groups like his have been raising these concerns for some time.
“Having companies explain to the public and government what measures they’re taking to address these risks feels like a bare minimum, reasonable step to take.”
The bill also includes whistleblower protections for employees of AI labs who believe their company’s technology poses a “critical risk” to society. In the bill, that risk is defined as contributing to the death or injury of more than 100 people, or more than $1 billion in damage.
Another piece of SB 53 is CalCompute, a public cloud computing cluster meant to support startups and researchers developing large-scale AI. That provision places the bill in a broader category than reporting alone: it pairs oversight of large AI developers with infrastructure aimed at other parts of the AI ecosystem.
How SB 53 differs from SB 1047
The most important difference from SB 1047 is what SB 53 does not do. Senator Wiener’s new bill does not make AI model developers liable for harms caused by their AI models.
SB 53 was also designed not to create a burden for startups and researchers that fine-tune AI models from leading AI developers, or for those that use open source models. That distinction matters because earlier AI safety proposals have faced criticism from parts of the technology industry that worry state rules could fall hardest on smaller teams.
In practical terms, SB 53 is a toned-down version of previous AI safety bills. But toned-down does not mean minor. If passed, it could still force major AI companies to publish more information than they currently share.
The bill is now headed to the California State Assembly Committee on Privacy and Consumer Protection for approval. If it passes there, it still needs to move through several other legislative bodies before it can reach Governor Newsom’s desk.
The state-by-state AI fight is widening
California is not the only state weighing this kind of AI safety legislation. On the other side of the U.S., New York Governor Kathy Hochul is considering the RAISE Act, a similar AI safety bill that would also require large AI developers to publish safety and security reports.
The future of state AI laws such as SB 53 and the RAISE Act was briefly uncertain because federal lawmakers considered a 10-year AI moratorium on state AI regulation. That proposal was intended to limit a “patchwork” of AI laws that companies would have to navigate.
But the moratorium proposal failed in a 99-1 Senate vote earlier in July. That failure left states with room to keep advancing their own AI legislation.
Geoff Ralston, the former president of Y Combinator, told TechCrunch that states need to act because there is no serious federal action in sight. He described California’s SB 53 as a thoughtful example of state leadership.
Why the bill matters for major AI labs
So far, lawmakers have struggled to get AI companies aligned behind state-mandated transparency requirements. Anthropic has broadly endorsed the need for increased transparency into AI companies and expressed modest optimism about the recommendations from California’s AI policy group. OpenAI, Google, and Meta have been more resistant to these efforts.
Leading AI model developers usually publish safety reports for their models, but the source article notes that consistency has slipped in recent months. Google did not publish a safety report for Gemini 2.5 Pro until months after the model was made available. OpenAI also did not publish a safety report for GPT-4.1. Later, a third-party study suggested it may be less aligned than previous AI models.
That pattern explains why SB 53 is focused on mandatory transparency rather than voluntary disclosure. If companies publish safety information only when and how they choose, lawmakers and public-interest groups may see a gap between the scale of AI deployment and the evidence available to evaluate it.
For now, SB 53 is still moving through the legislative process. But the direction is significant: California is once again asking whether the largest AI developers should be legally required to show more of their safety work to the public and to government.