OpenAI to regularly disclose AI misbehavior, warns safety challenges remain OpenAI said on Wednesday it would begin regularly publishing reports on unexpected or unauthorized AI behavior, while ...