California Governor Gavin Newsom on Sunday vetoed a landmark bill aimed at establishing the nation’s first safeguards for large artificial intelligence models.
The decision is a major blow to efforts to rein in a domestic industry that is rapidly evolving with little oversight. invoice would have enacted some of the first regulations Advocates say the move has led to large-scale adoption of AI models in the country, paving the way for nationwide AI safety regulations.
Earlier this month, the Democratic governor told an audience at DreamForce, an annual conference hosted by software giant Salesforce, that California needs to take the lead on AI regulation in the face of federal inaction, but this proposal “It could have a chilling effect on the industry,” he said.
The proposal drew fierce opposition from startups, big tech companies and several Democratic House members, and Newsom said the strict requirements could harm domestic industry.
“SB 1047, while well-intentioned, fails to consider whether AI systems are deployed in high-risk environments, involve critical decision-making, or involve the use of sensitive data,” Newsom said. he said in a statement. “Instead, this bill applies strict standards to even the most basic functions of large systems as long as they are deployed. This is the best approach to protect the public from the real threats posed by technology. I don’t think so.
Newsom announced Sunday that the state will instead partner with industry experts, including AI pioneer Fei-Fei Li, to develop guardrails centered around powerful AI models. Mr Lee opposed proposals regarding AI safety.
The measure, aimed at mitigating the potential risks created by AI, would require companies to test models that could be used to, for example, destroy a state’s power grid or help manufacture chemical weapons. They would be required to publish safety protocols to prevent manipulation. Experts say such a scenario could become possible in the future as the industry continues to advance rapidly. It would also have provided whistleblower protection for workers.
The bill is one of a number of bills passed by Congress this year aimed at regulating and combating AI. deep fake and protect workers. State lawmakers said they must act this year, citing hard lessons learned from California’s failure to rein in social media companies when it might have had the chance.
Supporters of the bill, including Elon Musk and Anthropic, said developers and experts are trying to understand how AI models work and why.
The bill targets systems that cost more than $100 million to build. Current AI models don’t reach that threshold, but some experts say that could change within the next year.
“This is due to a massive scale-up of investments within the industry,” said Daniel Cocotadillo, a former OpenAI researcher who resigned in April after accusing the company of ignoring AI risks. . “This is an incredible amount of power for a private company to control without accountability, and it is also incredibly dangerous.”
The US already lags behind Europe in regulating AI to limit risk. California’s proposal was not as comprehensive as Europe’s regulations, but it would put guardrails around a rapidly growing technology that has raised concerns about job losses, misinformation, privacy violations and automation bias. It would have been a good first step, supporters said.
Many large AI companies voluntarily agreed last year to follow safeguards set by the White House, including testing their models and sharing information. California’s bill would require AI developers to follow similar requirements to those promises, according to its supporters.
But critics, including former U.S. House Speaker Nancy Pelosi, said the bill would “kill California technology” and stifle innovation. That would deter AI developers from investing in large-scale models and sharing open source software, they said.
Mr. Newsom’s decision to veto the bill marks another victory in California for big tech companies and AI developers, many of whom have been pushing the governor and lawmakers over the past year to veto AI regulations. It has lobbied with the California Chamber of Commerce to block the move.
Two other radical AI proposals also faced growing opposition from the technology industry and others, and were scrapped ahead of last month’s legislative deadline. The bill would require AI developers to label the content they generate and prohibit discrimination by AI tools used in hiring decisions.