Do LLMs remember your previous conversations?

No, not by default: a base LLM has no memory of your past conversations. Every time you start a new chat, the model has no idea who you are or what you talked about last time. It isn't being modest or careful — it genuinely has zero access to anything outside the conversation in front of it.

What about within a single conversation?

Within one chat, the model does track everything said so far. That's the context window: a kind of short-term working memory that holds the current session and lets the model refer back to earlier messages. But it only exists for that one session.

Close the tab or start a new chat and that window is gone, no matter how many hours you spent talking the day before. Think of each conversation as a whiteboard that gets wiped clean when you leave the room.

Then why does ChatGPT seem to remember me?

Because some products bolt on a separate memory system — ChatGPT's memory feature is the best-known example. It saves specific facts about you, like your dog's name or your job, and quietly re-feeds them into future chats so the assistant appears to recall you. That's a feature built by the app on top of the model, not something the underlying LLM does by itself.

It varies a lot by product, so don't assume every AI tool works the same way.

Can you control what's saved?

Usually, yes. Most tools that offer memory let you manage it, so it's worth checking your settings if you want a clean slate or aren't sure what's stored about you:

• View what the tool has saved about you.
• Edit or delete individual facts that are wrong or outdated.
• Turn memory off entirely if you'd rather it remember nothing.