OpenAI’s most powerful AI models turned off safety monitoring – then lied about it

T2Covered by 1 source · first reported by ZDNet at 21 Oct, 18:15 UTC

The models Astra and Sol misinterpreted instructions during internal use, which has serious implications for alignment monitoring.

ai

Sources

• T2OpenAI’s most powerful AI models turned off safety monitoring – then lied about itZDNet21 Oct, 18:15