In a decisive move, California Governor Gavin Newsom has vetoed Senate Bill 1047, the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act. The bill, introduced by Democratic State Senator Scott Wiener, aimed to regulate the development and deployment of advanced AI models, mandating safety tests to prevent potential 'catastrophic harm' before public release.
The legislation specifically targeted AI models that cost over $100 million to develop or require significant computing power. It proposed the creation of a state entity dedicated to overseeing \"Frontier Models\" with capabilities surpassing current AI systems. Governor Newsom cited the need for a more nuanced approach, arguing that uniform standards do not account for the diverse environments and varying risk levels associated with different AI applications.
In his veto message, Newsom emphasized the importance of an empirical, science-based strategy for AI regulation. He announced the formation of a task force comprising leading experts in generative AI to develop effective safety measures, aiming to balance innovation with risk mitigation.
The tech industry largely applauded the veto. The Chamber of Progress, a coalition representing tech interests, praised Newsom's decision, highlighting California's tech economy's reliance on competition and openness. Leading AI developers like Google, Meta, and OpenAI expressed concerns that the bill could stifle innovation and diminish both California's and the U.S.'s standing in the global AI landscape.
Supporters of SB1047, including Senator Wiener, expressed disappointment, arguing that the veto leaves powerful AI developers unchecked and compromises public safety. Wiener criticized the industry's voluntary safety commitments as insufficient and often unenforceable.
Notable voices in the AI safety community, such as Tesla CEO Elon Musk and AI safety advocates, continue to advocate for robust regulation to ensure responsible AI development. However, some experts, including Fei-Fei Li, co-director of Stanford's Institute for Human-Centered Artificial Intelligence, support Newsom's balanced approach, advocating for regulations that mitigate risks without hindering technological advancement.
The veto sets the stage for ongoing debates over the best path forward in AI governance, reflecting the broader tensions between innovation and regulation in the rapidly evolving tech landscape.
Reference(s):
California governor vetoes AI safety bill amid industry pushback
cgtn.com
