In a significant move that stirs the ongoing debate over artificial intelligence regulations, California Governor Gavin Newsom has vetoed the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act (SB 1047) today. This decision has raised eyebrows statewide and beyond, igniting a fierce conversation about the balance between safety and innovation in the tech industry.

In his veto message, Governor Newsom highlighted his concerns regarding the potential negative impact of SB 1047 on AI companies and noted that the bill could be excessively broad. “While well-intentioned,” he wrote, “SB 1047 does not take into account whether an AI system is deployed in high-risk environments or involves critical decision-making.” He argued that the restrictive measures of the bill targeted even basic AI functions without properly assessing the risks involved, suggesting that it could inadvertently mislead the public into thinking that stringent regulations alone could mitigate the very real threats posed by advanced AI technologies.

The Governor expressed his belief that smaller, specialized AI models could be equally or even more dangerous than those the bill aimed to regulate. He emphasized the need for empirical analysis of AI systems and their trajectories rather than a one-size-fits-all regulatory approach. “I don’t believe the state should settle for a solution that is not informed by an empirical trajectory analysis of AI systems and capabilities,” he stated.

The veto has sparked backlash from supporters of SB 1047, including its primary author, Senator Scott Wiener, who took to social media platform X (formerly Twitter) to describe the veto as a “setback” for public safety oversight regarding powerful AI technologies. Wiener stressed the ongoing challenges of regulating massive tech companies in a landscape where U.S. policymakers are seemingly stuck in gridlock. “This veto leaves us with the troubling reality that companies aiming to create extremely powerful technology face no binding restrictions,” he lamented.

Originally introduced in late August, SB 1047 aimed to establish the strictest legal framework for AI regulation in the U.S. to date. It would have applied to AI companies with models costing over $100 million to train or more than $10 million to fine-tune, mandating safeguards like a “kill switch” and protocols to minimize risks such as cyberattacks and public health emergencies.

Throughout its journey, the bill underwent significant revisions. Key amendments included the removal of proposals for a new regulatory body and granting the state’s attorney general authority to sue developers for potential violations before incidents occurred. Despite the modifications leading to reduced opposition from certain companies, many in the tech industry remained critical. For example, OpenAI's chief strategy officer voiced concerns that the bill would inhibit progress, advocating instead for federal-level regulation. In contrast, Anthropic’s CEO noted after amendments that the bill had improved sufficiently to favor its adoption.

Tech giants, including Amazon, Meta, and Google, expressed relief at the veto. They warned that the law would hinder innovation and economic growth, clashing with California's reputation as a hub for tech advancement. Meta publicly stated that the bill would “stifle AI innovation, hurt business growth, and break the state’s long tradition of fostering open-source development.”

Politicians and notable figures across the political spectrum have weighed in—former House Speaker Nancy Pelosi and San Francisco Mayor London Breed opposed the bill, while supporters included high-profile individuals like Elon Musk and several prominent Hollywood figures.

As the dialogue around AI regulation continues, it remains evident that the complexities of governing such rapidly evolving technology require careful deliberation. Meanwhile, the federal government is also exploring its own regulatory frameworks, having proposed a $32 billion roadmap that addresses various influence areas of AI, including elections, national security, and copyright issues. The dance between regulation and innovation is just beginning, and all eyes will be on how this narrative unfolds in coming months.

Stay tuned—this thrilling saga of AI governance is far from over!