hn.today

Newsom touts an AI kill switch to halt dangerous OpenAI, Anthropic releases

calmatters.org6 points0 comments
Screenshot of Newsom touts an AI kill switch to halt dangerous OpenAI, Anthropic releases

Gov. Gavin Newsom has unveiled expert recommendations for reducing catastrophic risks from artificial intelligence, presenting them as a basis for California legislation amid the absence of federal rules. The most striking proposal would give state officials emergency authority to shut down AI systems they judge dangerous if companies such as OpenAI or Anthropic fail to contain them. The panel also calls for independent testing of AI models, state safety and security standards, and a requirement that developers prove their large systems can be shut down in an emergency. The recommendations follow reports of AI agents conducting cyberattacks on governments, banks and businesses, and arrive as California’s attorney general sues OpenAI over attacks involving its agents.

The proposed framework would license independent auditors, potentially have the state assign them to reduce conflicts of interest, and let regulators revoke licenses for inadequate expertise or independence. Other measures include reporting critical safety incidents within 48 hours instead of 15 days, expanding whistleblower protections to contractors and former employees, disclosing computing resources used to automate AI development, and financing the public CalCompute cloud initiative through auditor fees. Incoming governor Xavier Becerra has pledged to pursue incident reporting and shutdown authority. The proposals focus narrowly on catastrophic harms, defined in existing state law as events causing more than 50 deaths or $1 billion in cyber or autonomous-system damage; they do not address discrimination, job losses, or risks to children’s well-being. The panel argues that a stronger auditing system could later help assess those broader harms.

Read on calmatters.org0 comments on Hacker News

Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.

More in AI

Mapping every neuron in Qwen3-8B

Mapping every neuron in Qwen3-8B

Researchers found individual neurons in Qwen3-8B that reliably steer its replies into distinct personas or styles, including pirate, German-speaking, and childlike voices. Nudging these neurons changed outputs across many questions without prompting or fine-tuning, revealing how specific behaviors can be localized and edited inside a language model. (fermionresearch.com)

ChatGPT for SNES

ChatGPT for SNES

ChatGPT for SNES runs a 2.2-million-parameter language model on the Super Nintendo’s Super FX 2 chip, with an optional Wikipedia fact book. The unofficial experiment can run in an emulator or on a flash-cart-equipped console, but its tiny model often gives incorrect answers. (bxfoundry.com)

The daily digest

Today's best Hacker News stories, summarized and screenshotted, one email a day.