California Senate Passes AI Safety Bill SB 53 Amid Industry Pushback and Governor Newsom’s Veto Risk

California’s state Senate has passed SB 53, a sweeping AI safety bill that mandates transparency from large AI developers, establishes whistleblower protections for employees, and creates a public cloud infrastructure called CalCompute to expand access to computing resources. The bill now heads to Governor Gavin Newsom, who has not yet signaled his position on the legislation. Senator Scott Wiener, the bill’s author, said SB 53 is designed to ensure that companies building powerful AI systems are accountable for their safety practices. It requires large AI labs to disclose their safety protocols and risk mitigation strategies, with the level of detail depending on company revenue. Specifically, firms generating less than $500 million annually in revenue will only need to submit high-level safety summaries, while those exceeding that threshold must provide more comprehensive reports. The bill was shaped in part by recommendations from a panel of AI experts convened by Newsom after he vetoed a broader AI safety bill last year. That earlier measure, also authored by Wiener, was criticized for applying strict requirements to all large models regardless of their use case or risk level. Newsom had expressed concern that such an approach could impose burdens on companies even when their AI systems were not deployed in high-risk contexts involving sensitive data or critical decision-making. Despite the changes, the bill faces strong opposition from major tech companies, venture capital firms, and industry lobbying groups. OpenAI, while not naming SB 53 directly in a recent letter to Newsom, argued that companies should be considered compliant with state regulations if they meet federal or European Union standards, warning against duplicative and inconsistent rules. Andreessen Horowitz, a prominent Silicon Valley venture firm, has raised constitutional concerns, claiming that state-level AI legislation like SB 53 could infringe on states’ authority to regulate interstate commerce. The firm’s leadership has previously linked tech regulation to political decisions, including their support for Donald Trump’s 2024 presidential campaign. The Trump administration and allies have advocated for a 10-year moratorium on state-level AI regulation. In contrast, Anthropic has publicly endorsed SB 53. Co-founder Jack Clark stated that while the company supports a federal AI framework, the bill offers a strong and practical model for governance that cannot be overlooked. “In the absence of federal action, this creates a solid blueprint for responsible AI development,” he said. As the final decision rests with Governor Newsom, the outcome of SB 53 will signal California’s approach to balancing innovation with public safety in the rapidly evolving AI landscape.