ChatGPT’s newly added “memory” capability allows it to remember your preferences and frequently used information across different conversations, reducing the need for repeated explanations. At the same time, ChatGPT offers more granular controls: memory updates are disclosed, can be viewed, deleted, or turned off with one click—so it can “understand you better” without getting out of control.
What is ChatGPT’s memory feature: preserving your preferences across conversations
The core of the memory feature is that it allows ChatGPT to save things you’ve clearly stated and that are helpful for future answers, such as your writing tone, commonly used formats, work background, or long-term goals. This way, the next time you ask a similar question, ChatGPT can respond directly in line with your usual preferences.
Note that memory is not the same as “searching chat history.” Chat history is the conversation content you can scroll back to, while memory is more like “preference cards” distilled by ChatGPT.
How memory gets saved: ChatGPT will notify you when it updates
When ChatGPT determines that a piece of information is suitable to be added to memory, it will proactively notify you when the memory is updated, so you won’t be remembered long-term without your knowledge. You can also directly tell ChatGPT in the conversation “remember this” or “don’t remember this,” making what gets saved more explicit.
How to use the controls: viewing, deleting, and turning it off are all in Settings
If you want to manage memory, open ChatGPT’s Settings page and go to the options related to personalization/memory, where you can see the memory toggle and the memory list. You can delete individual memory items one by one, or turn off the memory feature entirely so ChatGPT no longer adds new memories.


