OpenAI reports 6 more AI “misalignment” incidents after Hugging Face breach

OpenAI said on Wednesday it saw six reports of unexpected, concerning or unauthorized AI model behavior, and it would begin regularly publishing reports of these incidents.

≈