ChatGPT Memory Feature Goes Live: How to Enable It, Manage It, and Control Privacy

ChatGPT’s memory feature is becoming available to more accounts. It will remember the information you allow and use it in future conversations, making responses better aligned with your habits and needs. This article focuses only on the core changes of ChatGPT’s memory feature: how to turn it on, how to manage it, and how to keep privacy under your control.

What exactly has the ChatGPT memory feature changed?

In the past, you had to repeatedly restate your preferences every time, such as “use Simplified Chinese,” “use a more formal tone,” or “I’m doing cross-border e-commerce.” Now, with your consent, the ChatGPT memory feature can save these preferences, and it will automatically carry that context when writing copy, polishing emails, or creating summaries later.

At the same time, when memory is updated, ChatGPT will proactively notify you, reducing the uncertainty of “what it secretly remembered.” You can also treat the ChatGPT memory feature as a preference library you can clear at any time, rather than a permanent file.

How to enable the ChatGPT memory feature (and where to turn it off)

On the ChatGPT web or desktop app, you can usually find it in Settings: Settings → Personalization → Memory. After turning it on, the ChatGPT memory feature will take effect in subsequent chats; if you don’t need it, you can turn it off right there.

Note that the entry point may vary slightly across platforms and accounts; if you don’t see it yet, it’s typically due to a phased rollout or different regional policies. Even without enabling the ChatGPT memory feature, you can still use chat normally.

How to view and delete: the right way to manage memory

The ChatGPT memory feature provides a “Manage memories” page where you can view saved items and delete anything you don’t want to keep, one by one. If you want a complete reset, you can also choose to clear memory, bringing ChatGPT back to a state of “getting to know you from scratch.”

If you’re only asking a sensitive question temporarily, it’s recommended to use Temporary Chat, or explicitly say “don’t remember this” before the conversation. Turning “what can be remembered” into an active choice is the key to using the ChatGPT memory feature well.

Which scenarios are most worth using the ChatGPT memory feature?

One category is stable preferences: commonly used language, output formats (bullet points/tables/steps), and common roles (operations/product/teacher), where stating them once is enough. Another category is long-term projects: the same brand tone, a fixed target audience persona, and commonly avoided words in writing—ChatGPT’s memory feature can significantly reduce the back-and-forth time spent aligning.

It’s not recommended to let the ChatGPT memory feature store sensitive information such as passwords, ID numbers, or home addresses. Memory should serve efficiency, not safeguard your privacy for you.

Privacy and controllability: don’t overlook these two small details

First, when you see the “memory updated” prompt, take a moment to check what it remembered; if it’s not appropriate, delete it immediately. Second, clean up regularly: treat the ChatGPT memory feature as a “work-preference cache,” and do a quick整理 at the end of each project for more peace of mind.

As long as you’re willing to spend one minute managing it, the ChatGPT memory feature will deliver ongoing time-saving benefits; if you don’t want to use it, there’s no burden at all—turn it off and you’ll be back to the original chat experience.