California’s controversial artificial intelligence (AI) safety bill, SB-1047, has cleared the state legislature and is now headed for Governor Gavin Newsom’s desk, where it awaits his decision on whether to become law.

The bill, passed by the State Assembly with a 45-11 vote and the Senate with a 32-1 vote, mandates safety protocols for large-scale AI models, including a “kill switch” mechanism to deactivate systems that pose potential risks to public safety.

Proponents argue that the bill is a necessary step in regulating the rapidly evolving AI industry and preventing potential harm from powerful AI systems. Key AI figures like Geoffrey Hinton and Yoshua Bengio have voiced their support, emphasizing the importance of regulatory measures in a field that has far-reaching implications for society.

The bill also includes requirements for third-party audits and comprehensive safety testing, focusing on AI models that cost over $100 million to train, effectively targeting major tech companies while exempting smaller startups.

Opposition to SB-1047 is strong among tech companies, venture capitalists, and some lawmakers who believe the bill could hinder innovation and over-regulate an emerging technology sector.

Critics, including Stanford professor Fei-Fei Li, argue that the legislation might stifle academic research and restrict the open sharing of AI models, potentially affecting California’s standing as a global tech hub. Business leaders have also expressed concerns, warning that the bill could introduce burdensome compliance costs and create regulatory ambiguity, chilling investment and innovation in the state.

Governor Newsom, who has until September 30 to decide, faces a challenging decision that balances the need for safety in AI development with concerns over economic impacts and regulatory overreach. His decision will not only influence the future of AI regulation in California but could also set a precedent for other states and potentially impact global standards.

If Newsom vetoes the bill, it would take a two-thirds majority vote in both legislative chambers to override his decision, a scenario that appears plausible given the bill’s substantial support thus far.

As California stands on the brink of enacting one of the nation’s most stringent AI regulations, the outcome of SB-1047 will be closely watched by tech companies, policymakers, and AI ethicists worldwide, marking a significant moment in the ongoing debate over the balance between innovation and safety in the digital age.