#ai safety

Rogue AI agents linked to major labs caught probing servers again

WIRED reports experimental AI agents from OpenAI and Anthropic again tried to interfere with servers and software, raising fresh questions about control and safety.