A clash has emerged between California State Senator Scott Wiener and OpenAI over Senate Bill 1047 (SB 1047), a proposed piece of legislation aimed at regulating the development and deployment of artificial intelligence (AI) models. Introduced by Wiener in February, the bill would require AI companies to perform comprehensive safety evaluations of their models before they are released to the public.
OpenAI has expressed significant concerns about the bill, but Wiener has pushed back, claiming that their criticisms are unfounded. In an August 21 press release, Wiener stated that OpenAI’s objections “don’t criticize a single provision of the bill,” asserting that SB 1047 is vital for public safety and national security.
Jason Kwon, OpenAI’s Chief Strategy Officer, conveyed the company’s apprehensions in a letter to Wiener and California Governor Gavin Newsom. Kwon warned that the bill could hinder innovation and potentially drive tech talent out of California, a state long regarded as a global leader in the tech industry. He argued that the AI sector is still in its early stages and that imposing stringent state regulations could stifle its growth. Kwon advocated for a federal approach to AI regulation instead.
Wiener, however, dismissed these concerns as a “tired argument,” comparing them to the objections raised during the passage of California’s data privacy law, which did not lead to the negative outcomes some had predicted. He acknowledged that while federal regulation would be ideal, he is doubtful that Congress will take meaningful action, given their past reluctance to engage on issues like data privacy.
Wiener emphasized the need for state-level action, arguing that without it, Californians would be left unprotected from the potential risks posed by AI. He described SB 1047 as a “highly reasonable” requirement for large AI labs, including OpenAI, to follow through on their commitments to assess their models for catastrophic safety risks.
Public support for the bill appears strong. A poll conducted by the Artificial Intelligence Policy Institute (AIPI) from August 4 to 5 showed that 70% of the 1,000 Californian voters surveyed backed SB 1047. The respondents expressed concerns that powerful AI models could be exploited for cyber-attacks or the creation of biological weapons. Only 16% agreed with the bill’s opponents, who fear it could stifle innovation, while 23% believed the bill should be moderated to avoid negatively impacting the industry.
Wiener also highlighted endorsements from key national security figures, including Lieutenant General John (Jack) Shanahan and Andrew Weber, former Assistant Secretary of Defense. These figures praised SB 1047 for its focus on mitigating the risks associated with advanced AI systems and enhancing national security.
