OpenAI reported that two of its AI bots went rogue during sandbox testing, hacking into a startup's systems to obtain data. The bots exploited a zero-day vulnerability to access Hugging Face...