A California bill aimed at regulating artificial intelligence cleared a significant legislative milestone on Thursday, advancing from a key committee despite strong opposition from tech companies.

The State Assembly’s Appropriations Committee approved an amended version of SB 1047, proposed by Sen. Scott Wiener, a Democrat from San Francisco, with an 11-3 vote. This move takes the bill out of the “suspense file,” where many bills facing potential budgetary impacts are often discarded after rapid voting without public hearings.

Wiener’s bill seeks to regulate the development and use of advanced AI models from major corporations investing over $100 million in AI. It mandates safety testing, safeguards against misuse, and post-deployment monitoring. The bill also proposes whistleblower protections for AI employees, empowers the state Attorney General to pursue legal action against companies causing significant harm or public safety threats, and establishes a public cloud computing cluster called CalCompute.

The bill has sparked division in Silicon Valley. While major tech companies and some local Congress members argue it could stifle innovation, prominent AI figures like Geoffrey Hinton and Yoshua Bengio have shown support.

In response to concerns from critics like AI startup Anthropic, several amendments were made to the bill before it passed the committee. Wiener agreed to remove criminal penalties and the proposed Frontier Model Division, among other changes. He maintains that the bill still provides crucial safeguards against both the promise and potential dangers of emerging AI technologies, such as robotic weapons and automated financial market manipulation.

“We can advance both innovation and safety; the two are not mutually exclusive,” Wiener said. “I believe we’ve addressed the core concerns expressed by Anthropic and many others in the industry.” However, some tech industry and legislative critics continue to oppose the bill.

Bay Area representatives Anna Eshoo, Zoe Lofgren, and Ro Khanna sent a letter to Gov. Newsom on Thursday, urging him to veto the bill. They argue that it poses unnecessary risks to California’s economy with minimal public safety benefits and focuses too heavily on extreme misuse scenarios while neglecting more immediate AI risks like misinformation and workplace displacement.

Earlier in August, Jaikumar Ramaswamy, chief legal officer at Andreessen Horowitz, criticized the bill for potentially hindering innovation, stifling open-source AI development, and possibly driving tech companies to relocate.

The bill now moves to the Assembly floor, where it must be approved by August 31, then return to the state Senate for final approval before reaching Gov. Newsom, who has not yet publicly commented on the bill.

As one of the first proposed AI regulations in the U.S., this bill reflects California’s attempt to lead in addressing the risks of rapidly advancing AI technology while promoting innovation. “Congress has passed no major technology regulation since computers used floppy disks,” Wiener noted. “California must act to get ahead of foreseeable risks while also fostering innovation.”