You might reasonably want a text transcript of spoken-language audio from a meeting and any computer technology that can do that is effectively some type of AI. In any case, I don't think there's a principled reason to be okay with every meeting at work being audio-recorded but to suddenly become not okay with it if there's a modern LLM involved doing speech-to-text; unless your actual objection is to LLMs as a technology or to the companies that are currently in the news for selling LLM-based software products.