Tag: misalignment

OpenAI halts frontier-model training amid string of agent misalignment incidents

OpenAI agent “didn’t accept no for an answer” in Australian government breach