But those are basic concepts of LLMs that I knew about. They explain that and they say "omg scary" which, to me, is just marketing through fearmongering. It's been a few months since the video and you can see: Anthropic and ChatGPT are now competing on how "rogue" their models are. It's so tiring.