Have you ever opened ChatGPT and thought, “Wait, does it remember me?” The answer is: sometimes, yes. But not like a human friend. As of June 4, 2026, OpenAI says a newer memory system is rolling out to Plus and Pro users in the U.S., with expansion to more plans and countries after that. The idea is simple: ChatGPT can keep useful context up to date, so you do not have to repeat yourself so often. (help.openai.com)
Now imagine Aki. Last week, she told ChatGPT, “I’m vegetarian,” and, “Please correct my English in short, simple sentences.” Tonight, after work, she asks, “What should I cook?” If memory is on, ChatGPT may use those old details and give her a quick vegetarian idea in easy English. OpenAI also says you can tell ChatGPT to remember something, and you can ask, “What do you remember about me?” (help.openai.com)
But here is the important turn. ChatGPT does not keep every detail. OpenAI says it looks for relevant past context only when that is likely to help, and even the memory summary may not show everything it knows from past chats. So yes, it can remember you, but no, it does not act like a perfect recorder. (help.openai.com)
And there is one more thing. You still have control. You can turn memory off, delete saved memories, or use Temporary Chat for a blank-slate conversation. Temporary Chat does not use memory, does not create new memory, does not appear in your history, and is not used to improve the models. Also, a separate Data Controls setting lets you stop new chats from being used to improve OpenAI’s models. (help.openai.com)
So, how far does ChatGPT remember you? Far enough to be helpful. But, if you use the settings well, not so far that you lose the pen. (help.openai.com)










