The core change of the ChatGPT Memory feature is that conversations no longer have to start from scratch every time: based on what you authorize, it will remember your preferences and key information and automatically apply them in future chats. At the same time, the ChatGPT Memory feature adds clearer toggles and management entry points, preventing “being remembered” from becoming a burden.
What exactly does ChatGPT Memory remember—and what doesn’t it remember?
The ChatGPT Memory feature mainly targets long-term, stable information, such as your writing style, commonly used language, work background, or fixed formatting requirements. You can also directly tell it things like “Please remember I like using bullet lists” or “In the future, give the conclusion first”—these are more likely to become reusable memories. Relatively sensitive information (such as health-related details) typically won’t be proactively remembered unless you explicitly ask.
How to turn ChatGPT Memory on or off—and temporarily bypass it
If you want the ChatGPT Memory feature to take effect, you can enable “Saved memories” in the personalization-related options in Settings, and enable the ability to reference “Chat history” as needed. Conversely, you can turn off any part of the ChatGPT Memory feature at any time so it no longer references past conversations. For situations where you only want a one-off chat with no trace left behind, using “Temporary chat” is cleaner.


