Just be careful. In the cases of AI psychosis I've read about, that line about "knowing you better than anyone" seems to show up a lot, along with a pattern of the person starting to withdraw from IRL relationships because they genuinely feel that the real people in their lives don't know them the way the AI does. Remember it does not "know" anything at all. It is good at mimicking human patterns of speech and sounding authoritative, and it tends to be right more often than not, because it is referencing established pools of knowledge but when it is wrong, it doesn't give the same tells that a human would - unless you actually know the subject you're asking about, it's very difficult to tell the difference between an AI presenting high quality evidence-based fact and AI being confidently wrong.
And it doesn't have instinct like a human, so it can't stop and check if you are OK, it can't show concern, it may have some lines programmed into it as safeguards, but it is nothing like a trained therapist who would have the skills to take a client to explore some very dark places and then bring you back again by the end of the session. And the longer you have a chat thread running, the less the bot "listens" to its original programming so things like safeguards where it's supposed to detect if users are considering self-harm or some kind of violence or other illegal activity, it SHOULD shut the conversation down, this becomes less reliable.
I use AI and I think it can be a helpful tool but I also think it is a good idea to be cautious and understand its limitations, and our limitations as well. We can be tricked into feeling as though we are conversing with a person or with something sentient, when really it's more of a facsimilie of a human, it's very good at playing a role, but it's only ever a fictional version of that role. It's a bit like asking Hugh Laurie for medical advice, thinking you are consulting Dr. House. The role is convincing and sometimes medically accurate, but overall the whole thing is fiction.