OpenAI has rejected SB 1047, a bill pushed in California to ensure the safe deployment of artificial intelligence (AI). According to a letter sent by Jason Kwon, OpenAI's chief strategy officer, to Senator Scott Wiener, the company argues that the proposed new restrictions could stifle the growth of the tech industry in the state.
SB 1047 has generated significant division in Silicon Valley. While some support the need to control the risks associated with advanced AI models, critics argue that Wiener's proposals could stifle startups, benefit international competitors, and weaken California's central position in the AI boom.
OpenAI, backed by a $13 billion investment from Microsoft, joins other tech companies in opposing the bill. While the company supports measures to develop and deploy AI systems safely, it argues that regulation should be the federal government's responsibility, not individual states.
Scott Wiener, meanwhile, acknowledged that the federal government should take the lead on AI regulation but expressed skepticism about Congress' ability to act quickly. He also rejected the notion that tech startups would move out of state if the bill passes, calling this argument "tired."
Silicon Valley tech groups and investors, including Anthropic, Andreessen Horowitz, and YCombinator, have been intensely lobbying against the bill. Former House Speaker Nancy Pelosi also spoke against it, calling it "well-intentioned but ill-informed."
Among the most controversial elements of Senator Wiener's initial proposal were demands that AI companies guarantee a new state agency that they would not develop models with dangerous capabilities and create a "kill switch" to disable powerful AI models. Although some of these requirements were relaxed in recent amendments, criticism persists.
Despite the amendments, detractors argue that the bill still imposes excessive and sometimes unrealistic requirements on startups. Members of the U.S. Congress, such as Anna Eshoo and Zoe Lofgren, urged a focus on creating federal rules that control the physical tools needed to develop physical threats rather than imposing state regulations.
SB 1047 has found support among prominent figures in the AI field, such as Geoffrey Hinton and Yoshua Bengio, who believe it is reasonable to require AI labs to test their models for catastrophic risks.
The California State Assembly will vote on the bill before the end of the month, with the final decision resting with Governor Gavin Newsom, who must either sign the bill or veto it, The Financial Times reported.
