Sacramento, California – The California State Senate has given final approval to SB 53, an artificial intelligence (AI) safety bill, forwarding the legislation to Governor Gavin Newsom for signature or veto. The bill, authored by State Senator Scott Wiener, represents a refined approach to AI regulation following Newsom's veto of a previous, broader AI safety bill, SB 1047, last year.
SB 53 is specifically designed to target large AI companies, applying to developers generating more than $500 million in annual revenue from their AI models. This narrower focus is intended to address concerns raised during the debate over SB 1047, which critics argued could hinder smaller startups and the state’s burgeoning AI ecosystem.
Key provisions within SB 53 mandate that covered AI laboratories publish safety reports for their models and report incidents to the government. Additionally, the bill establishes a protected channel for employees within these labs to report safety concerns to regulatory authorities without fear of retaliation, even if bound by non-disclosure agreements. The legislation notably seeks to avoid extensive regulation on smaller AI developers, requiring less comprehensive safety information from startups.
Industry observers, including analysts on TechCrunch's 'Equity' podcast, have suggested that SB 53's targeted approach and endorsement by companies such as Anthropic may improve its chances of becoming law compared to its predecessor. Max Zeff, an analyst, stated on the podcast that the bill could provide a "potentially meaningful check on tech companies’ power."
California's legislative actions hold significant weight due to the state's position as a major hub for AI development. As noted by Kirsten Korosec, also speaking on 'Equity,' most major AI companies either operate or maintain a substantial presence within the state, making California's regulatory landscape a critical factor for the industry.
The bill's passage into law could also intersect with broader federal discussions regarding AI regulation. Reports indicate that the federal administration has explored language in funding bills that could pre-empt state-level AI laws, potentially setting the stage for future regulatory conflicts between California and federal authorities.
