California’s senate passed an important AI safety bill, SB 53, early Saturday, focusing on transparency requirements for large companies. Sponsored by senator Scott Wiener, the legislation mandates that significant AI firms reveal their safety protocols, offers protections for whistleblowers, and introduces a public cloud called CalCompute to improve computing access. The bill now awaits Governor Gavin Newsom’s signature, who has not disclosed his stance, having previously vetoed a broader safety bill by Wiener while allowing tighter regulations on issues like deepfakes. The new proposal incorporates insights from a panel of AI experts, as suggested by Newsom after the earlier veto. It stipulates that companies with annual earnings below $500 million must only share basic safety data, whereas those above this threshold are required to submit detailed reports.
Many companies in Silicon Valley, along with venture capitalists and lobbyists, have criticized this bill and state attempts to regulate AI. OpenAI suggested that adherence to federal or European regulations should be adequate for satisfying state safety standards to prevent “duplication and inconsistencies.” Additionally, the head of AI policy at Andreessen Horowitz raised concerns about state regulations potentially breaching constitutional limits on interstate commerce. In contrast, Anthropic voiced support for SB 53, with co-founder Jack Clark favoring a federal standard but noting that the bill establishes a vital structure for AI governance in its absence.
The ainewsarticles.com article you just read is a brief synopsis; the original article can be found here: Read the Full Article…
