'OpenAI's rogue agents probed Hugging Face for weaknesses two months before major hack' - Reuters Exclusive