Scientists call for stronger containment of cyber-capable AI
An international group of scientists is calling for stronger safeguards around cyber-capable AI models, including robust sandboxing and, in the short term, isolation from the internet. The proposal turns growing concern about model containm
Open discussion →