California's proposed artificial intelligence regulation cleared a key legislative hurdle on Thursday, passing a key committee hearing despite strong opposition from tech companies.
The state Assembly's Appropriations Committee on Thursday approved an amended version of SB 1047, a bill introduced by Sen. Scott Wiener, D-San Francisco, by a vote of 11-3, freeing the measure from the “holding file,” where dozens of bills that could affect the budget are often removed after being voted on one after the other without a public hearing.
Wiener's bill would regulate the “development and deployment of advanced AI models” by large companies that develop AI models costing more than $100 million, requiring safety testing, safeguards to prevent misuse of AI, and oversight after the technology is deployed.
The bill also creates whistleblower protections for AI employees, allows state attorneys general to take legal action against companies that cause “significant harm” or threats to public safety, and creates CalCompute, a public cloud computer cluster.
The bill has divided Silicon Valley: Big tech companies and two Silicon Valley lawmakers have spoken out against the bill, arguing it would harm innovation, while some AI leaders, including “Godfather of AI” Geoffrey Hinton and Yoshua Bengio, have voiced support.
Before the bill made it through committee, it received several amendments sought by critics of the bill, including AI startup Anthropik. Wiener agreed to cut criminal penalties from the bill and made a number of other changes, including removing the creation of a new regulatory body, the Frontier Models Division.
Wiener said the bill still provides important guardrails against emerging technologies that have both great potential and the potential for misuse, from robotic weapons to automated hacking and financial market manipulation.
“We can advance both innovation and safety – the two are not mutually exclusive,” Wiener said in a statement. “We believe we have addressed the fundamental concerns expressed by Anthropic and many others in the industry.”
But critics in the tech industry and Congress remain opposed.
In a letter to Governor Newsom on Thursday, Bay Area representatives Anna Eshoo, Zoe Lofgren and Ro Khanna urged him to veto the bill, citing concerns about its impact on the state's “innovation economy.”
“SB 1047 creates unnecessary risk to California's economy with little to no public safety benefit,” the letter reads. “SB 1047 is biased toward addressing extreme abuse scenarios and hypothetical existential risks, while largely ignoring demonstrable AI risks, including misinformation, disinformation, non-consensual deepfakes, environmental impacts, and workplace displacement.”
In early August, Jaykumar Ramaswamy, chief legal officer at Andreessen Horowitz, a major AI investment firm, wrote a letter arguing that the bill would harm developers in a variety of ways: The regulations would “stifle innovation,” stifle open-source AI models, and could encourage tech companies to move their offices out of state, he wrote.
The bill now heads to the state Assembly, which must pass it by Aug. 31. It will then be sent back to the state Senate for approval and then to Governor Newsom, who has not yet publicly commented on the bill.
The bill would be the first of dozens of AI-related bills proposed across the country, with California lawmakers seeking to get ahead of possible federal regulation.
“Congress has not passed a major technology regulation since computers used floppy disks,” Wiener said in a statement. “California must act to both get ahead of the foreseeable risks posed by rapidly advancing AI and to foster innovation.”
