California has released a comprehensive new report outlining a forward-looking policy framework for managing the development and risks of frontier artificial intelligence (AI), marking a significant step in shaping AI governance in the U.S. and beyond.

Commissioned by Governor Gavin Newsom and led by top academics from Stanford and UC Berkeley, The California Report on Frontier AI Policy offers an evidence-based roadmap for lawmakers and regulators navigating the complexities of cutting-edge AI systems.

The report, developed by the Joint California Policy Working Group on AI Frontier Models, draws on historical case studies, expert consensus, and empirical analysis to inform its recommendations.

Here are five key takeaways from the report:

1. Balance Innovation with Risk Mitigation 

•  California’s policy approach centers on “trust but verify”, aiming to enable transformative benefits from AI in sectors like health, education, and clean technology, while ensuring safeguards are in place to avoid potentially irreversible harms. 
• Early policy choices are seen as critical, with the report stressing that inaction could lock in unsafe norms.

1. Transparency as a Cornerstone

•  The report calls for greater transparency from AI developers, including disclosures about training data, safety practices, and downstream impacts. 
• It proposes third-party evaluations, whistleblower protections, and public reporting systems to fill current information gaps and build public trust.

1. Evidence-Based Policymaking under Uncertainty

• With AI evolving faster than scientific consensus can keep up, the report advocates for dynamic, evidence-generating policies. 
• It urges the use of simulations, adversarial testing, and case comparisons (from the internet and energy sectors) to inform regulatory decisions before harms occur.

1. Adverse Event Reporting and Regulatory Gaps

•  Inspired by safety monitoring in other industries, the report recommends creating AI-specific adverse event reporting systems. These would track post-deployment risks and feed into updates for enforcement agencies. 
• The authors caution that current laws may not fully address the unique risks posed by foundation models.

1. California as a Global Policy Leader

•  As home to many top AI labs, California is uniquely positioned to shape global norms.
• The report notes that effective state-level policies can both influence international standards and strengthen U.S. competitiveness, especially amid rising concerns over AI’s impact on national security, labor markets, and societal stability.

The report stops short of endorsing any specific legislation but provides foundational principles for California’s future AI laws and regulatory actions.

Source: Governor of California Report PDF