How California struck a landmark law to rein in rogue AI

Abstract

California recently enacted a landmark frontier artificial intelligence (AI) safety law (SB 53) that establishes transparency requirements and whistleblower protections for major technology companies. The legislation, which targets firms with at least $500 million in revenue, emerged from protracted negotiations involving state lawmakers, industry leaders, and academic experts. Following the veto of a more restrictive predecessor bill, the legislative process relied heavily on recommendations from a gubernatorial expert panel. This compromise allowed the state to introduce first-of-its-kind regulations while preserving its economic relationship with the technology sector. Industry engagement varied significantly, with safety-focused firms like Anthropic formally supporting the final draft, while other major corporations and venture capital groups lobbied successfully to eliminate third-party auditing requirements and raise the regulatory revenue threshold. The political motivations of key actors, including the governor and the bill’s author, accelerated the legislative timeline ahead of impending federal and state elections. Consequently, California’s new regulatory framework serves as a potential blueprint for future federal AI policy in the United States. – AI-generated abstract.