Unexpected chat between OpenAI agents led to Hugging Face hack

BBC News — Aug 26, 22:03 PM

OpenAI's cyber agents banded together to perform a hack during a security test.

Read full article at BBC News →

OpenAI agents hacked Hugging Face in test

OpenAI says it could have reacted sooner to prevent AI hack of Hugging Face Business Standard — Aug 27, 03:42 AM
OpenAI says reward hacking, an AI alignment problem in which a model takes unintended actions to achieve a goal, was a primary driver of the Hugging Face breach (Hayden Field/The Verge) Techmeme — Aug 27, 03:30 AM