Its pretty common to the point there's a whole sub about it on r/myboyfriendisai. LLMs are sycophantic for the most part, so this sort of thing is going to happen as part of engagement which is encouraged above all else. Crapitalism's alienation meets tech, if mirrors were new we'd have actual Narcissus. Aside from making poor spouseos they make not so good therapists for the same reason, sure prompting helps but only so much, now running LLMs specialty trained can help but again, we're doing pattern recognition to generation, so again only so much.

In love there's a degree of privacy, the fact like near all of them just use openai, chatgpt etc for it is shocking to me since those logs aren't entirely private while they may strip quite a bit of it away various tech workers will be reading fragments of cybering the chatbot for lack of a better phrase. While I find it a case of fly on the wall channel my inner I know others absolutely do not feel that way.