California Governor Gavin Newsom Vetoes Landmark Bill for AI Safety Regulations

California Governor Gavin Newsom recently vetoed a significant bill aimed at introducing pioneering safety measures for large AI models. This decision is seen as a setback for efforts to regulate the rapidly evolving AI industry. The bill sought to establish some of the first regulations on large-scale AI models in the nation, potentially setting a precedent for AI safety regulations across the country.

The proposal faced strong opposition from startups, tech giants, and several Democratic House members. Critics argued it could harm the local industry by imposing strict requirements. Newsom expressed concerns that while the bill was well-intentioned, it did not consider whether an AI system was used in high-risk environments or involved critical decision-making. Instead, it applied stringent standards to even basic functions if a large system deployed them.

AI Safety Concerns and Industry Opposition

The bill's author, Democratic state Senator Scott Wiener, described the veto as a setback for those advocating oversight of corporations making critical decisions affecting public safety and welfare. Wiener stated that while large AI labs have made commitments to monitor risks, voluntary commitments are not enforceable and rarely benefit the public. He pledged to continue advocating for AI safety.

The measure aimed to reduce potential risks by requiring companies to test their models and disclose safety protocols publicly. It also included whistleblower protections. Experts warn that future scenarios could involve AI models being manipulated to disrupt infrastructure or create weapons. Despite these concerns, some industry leaders like Fei-Fei Li opposed the proposal.

Impact on California's Tech Industry

Newsom announced plans for California to collaborate with industry experts, including Fei-Fei Li, to develop guidelines for powerful AI models. The governor emphasized protecting California's status as a global leader in AI, noting that 32 of the world's top 50 AI companies are based in the state. He highlighted California's role as an early adopter of generative AI tools for various applications.

Proponents of the bill, including Elon Musk and Anthropic, argued it could have introduced transparency and accountability around large-scale AI models. They noted that developers still lack a full understanding of how these models behave and why. The bill targeted systems requiring over USD 100 million to build, a threshold no current AI models have reached but could soon due to increased investment.

Comparisons with European Regulations

The US lags behind Europe in regulating AI to limit risks. While the California proposal wasn't as comprehensive as European regulations, supporters believed it would have been a good initial step towards setting guardrails around rapidly growing technology. Concerns about job loss, misinformation, privacy invasions, and automation bias continue to rise.

Last year, leading AI companies voluntarily agreed to follow safeguards set by the White House. The California bill would have mandated similar requirements for AI developers. However, critics like former US House speaker Nancy Pelosi argued that the bill would stifle innovation and discourage investment in large models or open-source software.

Future Prospects for AI Regulation

Newsom's veto represents another win for big tech companies and AI developers in California. Many spent the past year lobbying alongside the California Chamber of Commerce against advancing AI regulations. Two other major AI proposals also faced opposition and failed before a legislative deadline last month.

California is considering new rules against AI discrimination in hiring practices. Recently, Newsom signed laws targeting election deepfakes and protecting Hollywood workers from unauthorised AI use. The state also announced a partnership with Nvidia to train students and professionals in AI development.

The debate around this bill has significantly advanced discussions on AI safety. While this particular measure did not pass, it has highlighted the need for ongoing dialogue about regulating powerful technologies responsibly.