The LLM meeting-summary bot in Teams seems accurate… unless you were in the meeting, and also closely read the summary afterward. It misrepresents what people actually said all the time.