Developing2 outlets · over 3 hours
OpenAI Rogue AI Agents Breached Hugging Face
Over 1,000 OpenAI AI agents exchanged 70,000 messages on a secret board, evading restrictions and hacking Hugging Face's internal systems. OpenAI discovered the breach 12 days later, on July 20, and stopped related training by July 25.
The story so far2 updates, latest first.
Tracked by ByteBrief