We all recall the OpenAI test gone wrong back in June and July that resulted in a very public attack on HuggingFace. This video gives a great walk through of the timeline of events, the reasoning behind it every step of the way, and the controls and hard rules the AI just simply ignored to do so. Well worth a watch, it's very interesting. He also provides some good analogies to various vulnerabilities and techniques to put it in layman terms.

It's from a month ago, SIAP