California’s state senate has given final approval to SB 53, a new artificial intelligence (AI) safety bill, sending it to Governor Gavin Newsom for signature or veto. This legislative effort comes after Newsom vetoed a similar AI safety bill, SB 1047, authored by Senator Scott Wiener, last year. However, SB 53 is distinguished by its narrower scope, specifically targeting large AI companies that generate over $500 million in annual revenue from their AI models, a key difference that is believed to enhance its chances of becoming law.

The nuances of SB 53 and its potential impact were discussed on TechCrunch’s flagship podcast, Equity. According to Max Zeff, a TechCrunch colleague, the bill's focus on major AI developers such as OpenAI and Google DeepMind, combined with its endorsement by AI company Anthropic, makes it a more viable piece of legislation. It aims to introduce meaningful regulations, compelling these powerful entities to publish safety reports for their AI models and to formally report any incidents to the government. Furthermore, SB 53 provides a vital channel for employees within these AI labs to report concerns to the government without facing corporate retaliation, even if bound by non-disclosure agreements, marking a significant check on the unchecked power that tech companies have often enjoyed.

Kirsten Korosec, another colleague on the podcast, highlighted the critical importance of California's role in this legislative push. Given that nearly every major AI company either originated in California or maintains a substantial footprint within the state, regulations enacted here carry considerable weight and can set precedents for the broader industry. The state's status as a global hub for AI activity means its legislative actions have far-reaching implications.

While SB 53 is designed to be narrower, this also introduces a degree of complexity, as Max Zeff pointed out. A primary objective was to avoid the controversy surrounding SB 1047, which faced criticism for potentially harming California’s booming startup ecosystem. Consequently, SB 53 includes carve-outs that largely exempt small startups from the most stringent requirements. Smaller startups are still expected to share some safety information but face significantly fewer regulatory burdens, ensuring the bill primarily targets the industry giants.

Beyond the state-level implications, the discussion also touched upon the broader landscape of AI regulation and the evolving federal stance. There is an observation that the federal administration might be moving towards a less regulated approach, potentially including language in future funding bills that could restrict states from enacting their own AI regulations. This suggests a potential future conflict between a federal administration seeking to minimize regulation and blue states like California pushing for more oversight.