California’s newly passed AI safety bill, SB 53, could represent a meaningful regulatory check on the largest artificial intelligence companies, thanks to its targeted approach and focus on accountability. After receiving final approval from the state senate, the bill now heads to Governor Gavin Newsom, who must decide whether to sign it into law or veto it. This comes after Newsom rejected a broader AI safety measure, SB 1047, last year—largely due to concerns about stifling innovation, especially among startups. SB 53 is intentionally narrower, aiming to regulate only AI developers with more than $500 million in annual revenue from AI products. That effectively targets industry giants like OpenAI, Google DeepMind, and Meta, while sparing smaller startups that lack the same scale and resources. The bill introduces several key requirements. It mandates that large AI companies publish detailed safety reports for their models, including assessments of potential risks and mitigation strategies. In the event of a significant AI incident—such as a system failure or harmful output—the company must report it to the state government. Additionally, the law protects employees who raise safety concerns, allowing them to report issues to regulators without fear of retaliation, even if they’ve signed non-disclosure agreements. This employee protection clause is particularly significant, as it addresses a growing concern within AI labs: the tension between corporate secrecy and public safety. By creating a formal channel for internal whistleblowers, SB 53 gives workers a way to speak up when they believe a model could cause harm. The bill’s focus on California is no coincidence. The state is home to the majority of the world’s leading AI companies, from OpenAI’s West Coast operations to Google’s research hubs and countless startups in Silicon Valley. This makes California a natural testing ground for AI regulation, and its laws often set a precedent for other states. While SB 53 has fewer sweeping provisions than its predecessor, it’s not without complexity. Critics may argue it’s still too lenient, especially with exemptions for smaller firms and certain types of AI applications. But supporters, including Anthropic, see it as a balanced step forward—one that targets the most powerful players without undermining innovation in the broader ecosystem. The timing is also crucial. With the federal government taking a hands-off stance on AI regulation—some officials even pushing to block states from enacting their own rules—state-level action like SB 53 may become one of the few avenues for oversight. As federal policy remains uncertain, California’s approach could spark a broader national debate, pitting blue states that prioritize safety against a federal administration that favors industry freedom. Ultimately, SB 53 may not be a complete solution, but it signals a growing recognition that the most powerful AI companies need accountability—and that states, not just Washington, have a role to play in shaping the future of artificial intelligence.