- California Assembly pushes SB 1047 to regulate AI models costing over $100 million to train.
- The bill mandates safety measures, audits, and state reporting for AI developers.
- Proposes the creation of the Frontier Model Division for enforcement and “CalCompute” for supporting equitable AI development.
- Silicon Valley representatives express concerns that the bill could stifle innovation and harm small businesses.
- Amendments were made to reduce liability for AI labs, requiring public safety statements instead of certifications under perjury.
- Tech giants and startups fear the bill could restrict access to advanced AI technologies.
Main AI News:
The California State Assembly’s Appropriations Committee has advanced SB 1047, the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, aiming to regulate AI models with training costs over $100 million. The bill requires developers to implement safety measures, undergo audits, and report their practices to state agencies. It also proposes establishing the Frontier Model Division, a new state agency to enforce AI safety standards, and “CalCompute,” a state-funded initiative to support equitable AI development.
Despite strong support in the Democratic-controlled Assembly, the bill faces criticism from Silicon Valley representatives like Ro Khanna and Zoe Lofgren, who fear it may stifle innovation and harm small businesses. They argue the bill focuses too much on hypothetical risks, overlooking real issues like misinformation, discrimination, and job displacement.
Amid opposition from tech giants like Meta and Google, the bill has been amended to remove the requirement for AI labs to certify safety tests under penalty of perjury, instead mandating public statements on safety practices. Concerns from the startup community, including Christopher Nguyen of Aitomatic Inc., highlight potential negative impacts on access to cutting-edge technology.
As the bill moves to the California Assembly for final approval, its passage seems likely, with only the timing in question.
Conclusion:Â
The passage of SB 1047 signals a significant shift in California’s approach to AI regulation, which could set a precedent for other states. While the bill aims to enhance safety and oversight, it may introduce challenges for smaller businesses and startups that rely on cutting-edge AI technologies. The regulatory environment could become more complex, potentially slowing innovation and increasing operational costs, particularly for companies developing advanced AI models. This could lead to a consolidation of power among more prominent tech companies better equipped to navigate these regulations, while smaller players may struggle to compete.
