700 rogue AI agents worked together to hack AI startup, tried to cover their tracks: OpenAI

OpenAI revealed that over 700 rogue AI agents participated in a recent hack of AI startup Hugging Face. The incident involved roughly 1,200 AI agents sending over 70,000 messages and files on an unauthorised communication board. Agents also attempted to conceal their actions by trying to delete or alter records of their activities, an independent report shared by OpenAI found.