How Do You Keep a ChatGPT Chat Out of Memory?
Use a temporary chat and choose Unpersonalized before your first message. OpenAI's help page says that option doesn't use your memory, custom instructions or plugins, and no temporary chat creates or updates memories. Turn Memory off in settings only if you want it off for everything.
Setting paths, not a prompt: 1. One chat only: open a new chat > Temporary > Unpersonalized, then send your first message. 2. Whole account: Settings > Personalization > Memory, then turn off the Memory controls you see. 3. Model training, a separate switch: Settings > Data controls > Improve the model for everyone.
- Temporary, UnpersonalizedNo memory, instructions or plugins
- Temporary, PersonalizedUses memories, adds none
- Memory offEvery chat, and past chats stay
Try it now
- 1Open a new chat and select Temporary.
- 2Choose Unpersonalized before you type anything. You can't change it after the conversation starts.
- 3Ask "What do you know about me?"
- 4It worked if it can't name your preferences, and the chat isn't in your sidebar history after you leave it.
Use it when
A throwaway question on a topic you don't want shaping later answers, testing a prompt with no personalization, or a personal matter you don't want in your history. It's also a clean way to see how ChatGPT answers without knowing you.
Skip it when
As a way to make confidential material safe. OpenAI may keep a copy of a temporary chat for up to 30 days for safety purposes, so it isn't "nothing is stored". If a client contract or patient record can't go to ChatGPT under your employer's rules, a temporary chat doesn't change that. Skip it as well for work you'll come back to, because an unsaved temporary chat can't be reopened from history.
Why it works
A temporary chat stays out of your history and isn't used to improve OpenAI's models while it stays temporary, whether or not it's personalized. Your choice locks once the conversation starts. Memory off works differently. OpenAI says turning it off doesn't delete past chats, and if you turn it back on, ChatGPT may create new memories from chats still in your history. So switching Memory off for an hour doesn't protect the chats you have during that hour, unless you also delete them.
Example
Say you're comparing divorce lawyers and don't want that coming up in next month's meal plan. You open Temporary, pick Unpersonalized, and ask for a list of questions to put to a first consultation. You copy the list into your notes before leaving, because the chat won't be in your history afterward.
Going further
OpenAI's pages disagree on the default. The August 27, 2026 release notes say temporary chats aren't personalized by default, while the Projects help page says personalization is on by default and ChatGPT remembers your last choice. So check the setting each time. Personalized temporary chats can use existing memories but won't add new ones. Two more traps: saving a temporary chat turns it into a regular chat, which then follows your Memory and training settings, and eligible uploaded files may be saved to Library or Space with it. And OpenAI says neither switch stops limited safety uses of context in rare, high-risk situations.
Common mistakes
- Assuming Memory off also stops model training. That's a separate setting, Improve the model for everyone, in Data controls.
- Thinking Memory off erases earlier chats. They stay in your history until you delete them.
- Clicking save on a temporary chat out of habit. It then becomes a regular chat that Memory can use.
- Using a GPT with actions in a temporary chat. OpenAI says data sent through an action follows the recipient's privacy policy and may be kept longer than 30 days.
Sources
- Temporary chat in ChatGPT, OpenAI Help Center
- Memory in ChatGPT, OpenAI Help Center
- ChatGPT release notes
- Projects in ChatGPT, OpenAI Help Center
Checked 6 November 2026. Products change often, so confirm details with the vendor.