Why California’s SB 53 might provide a meaningful check on big AI companies (techcrunch.com)

California’s legislature gave final approval to SB 53, a narrower AI safety bill by state senator Scott Wiener that now heads to Governor Gavin Newsom. Unlike last year’s vetoed SB 1047, SB 53 targets only large AI developers—those generating more than $500 million from their models—aiming squarely at firms like OpenAI and Google DeepMind while sparing smaller startups. The bill would require covered companies to publish safety reports, notify the state about incidents, and provide a government channel for employees to raise safety concerns without retaliation even when NDAs exist. Anthropic’s endorsement and the narrower scope make passage more plausible than last year’s effort. For the AI/ML community, SB 53 represents a potentially meaningful, state-level check on major labs that could drive greater transparency and operational accountability around model risks. Technical implications include mandatory incident disclosure and public safety documentation that could influence development practices, external audits, and risk mitigation priorities at large firms. The bill also sits amid a fraught federal backdrop—recent federal signals discourage state-level regulation and new administration stances could prompt legal or political clashes—so SB 53’s fate and its precedential impact on U.S. AI governance remain consequential.