California has passed the first U.S. law regulating the activities of major AI companies. The bill, SB 53, signed by Governor Gavin Newsom, requires companies such as OpenAI, Meta, Anthropic, and Google DeepMind to disclose their model safety protocols and report serious incidents, including cases of misinformation or AI acting without human control.
The law also protects internal whistleblowers, allowing them to safely report violations. A special channel has been created for reports to California’s Office of Emergency Services. This makes the bill unique even compared with European regulation, where such cases are not always yet brought to the attention of regulators.
The context for the bill’s passage is telling. Against the backdrop of political debate over “soft” and “hard” AI regulation, companies are actively advancing their interests by supporting preferred candidates. But SB 53 emerged from a long, carefully weighed discussion. For the first time in the U.S., the balance between AI development and public protection has been set out in law. This is not just a step toward regulation, but an attempt to build trust and ensure safety.
