i don't personally see messages with an LLM as being different from, say, terminal commands. it's a machine interface. it sounds like you're anthropomorphizing the chat bot, if you're talking to it like you would a human then i would be more worried about the implications that has for you as a person.