@jon right up to the point!
The short blog post is a fantastic representation why we should stop putting trust and reliable functions in these #llm s. They are tools and should be presented as such transparently or not exist.