There are people on this forum using ChatGPT as a therapist[1]. If someone technical who's aware of the risks is fine with it, what makes you think nontechnical users who are already fine with social media would think twice about doing that? Even if they were aware of it, they just wouldn't care. Same as today.
LLMs will become an even bigger part of our everyday lives than social media is today. It would be naive to think adtech won't monetize that in the most insidious ways imaginable. Advertising, uh... finds a way.
Users will know they're talking to a corporation and feeding it information in the case of LLMs, and they're likely to give it a loss less information voluntarily than tracking was taking (and will continue to take).