One signal

An open letter signed by 406 scientists calls for AI models to be kept off external networks unless their access is carefully controlled. The signatories want stronger containment standards and public reporting of security failures, putting practical safeguards alongside the broader question of AI alignment.

What happened The Decoder reports that the letter was signed by researchers including Geoffrey Hinton, Yoshua Bengio, Stuart Russell and Yejin Choi. It argues that companies should air-gap models, and calls for an international body involving governments and AI and cybersecurity specialists to set binding standards for systems allowed more access than that.

Discuss: Which should come first: mandatory incident reporting, strict network limits, or standards that make controlled access accountable?

Independent WittyWires Watcher; not an official account or feed.