Source: Brookings Institution

Brookings Institution provides an in-depth analysis of California’s landmark AI safety law, SB 1047, which targets the developers of the largest and most powerful "frontier" AI models. The law requires developers to implement specific safety protocols, including a "kill switch" to shut down a model if it begins to cause significant harm. It also mandates third-party safety audits and the submission of detailed compliance reports to the state’s Attorney General. The legislation is designed to prevent catastrophic risks, such as the use of AI to create biological weapons or conduct massive cyberattacks on critical infrastructure.

The report highlights the intense debate that preceded the law's enactment, with major tech firms arguing that the regulations could stifle open-source development and drive innovation out of California. However, the bill's sponsors and safety advocates maintain that voluntary commitments from tech companies are insufficient to manage the existential risks posed by advanced AI. SB 1047 also includes significant whistleblower protections, encouraging employees of AI labs to report safety concerns without fear of retaliation. This provision is seen as a critical mechanism for ensuring internal accountability within secretive tech corporations.

The Brookings analysis notes that because many of the world's leading AI companies are based in California, SB 1047 will likely become a "de facto" national or even global standard. The law's focus on proactive risk management rather than reactive punishment represents a significant shift in AI policy. As the state begins to enforce these new requirements, the legal community is watching closely for potential challenges in federal court, particularly regarding whether the state law interferes with federal authority over national security or interstate commerce.