News Summary

Anthropic, a prominent AI company, has officially endorsed California’s proposed AI safety bill, SB 53. This bill, introduced by state senator Scott Wiener, aims to impose stringent transparency requirements on major AI developers such as OpenAI, Anthropic, Google, and xAI. Despite facing opposition from significant tech entities like the Consumer Technology Association and the Trump administration, Anthropic’s endorsement represents a substantial win for the bill. SB 53 mandates that AI model developers create safety frameworks and publish public safety reports before deploying advanced AI models. It also provides protections for whistleblowers with safety concerns. The bill addresses extreme AI risks, such as preventing AI from assisting in biological weapons creation or cyberattacks. Although California’s Senate has approved a prior version, the bill still requires a final vote before reaching the governor. Governor Gavin Newsom has not yet expressed his stance, although he previously vetoed a similar bill. The tech industry argues that AI regulation should be federal to avoid stifling innovation, yet Anthropic’s co-founder Jack Clark asserts that immediate action is necessary. The bill specifically targets major AI companies with revenues exceeding $500 million, establishing legal safety obligations with financial penalties for non-compliance. Despite the challenge, experts view SB 53 as a measured approach compared to previous proposals.