ChatGPT Memory can personalize replies using saved memories and, when enabled, relevant information from past chats and other available sources. To review or remove that context, open Settings > Personalization > Memory if the controls are available on your account. Memory, chat deletion, Temporary Chat, and model-training preferences are separate controls, so changing one does not automatically change the others.
How ChatGPT Memory works
Memory is context ChatGPT may use to tailor future responses—for example, a preference, project detail, or constraint. OpenAI describes its purpose as helping future conversations start with shared context rather than from scratch in its June 4, 2026 announcement. Memory is not a complete, searchable transcript: OpenAI’s Help Center says, “Memory does not retain every detail from every conversation.” What it uses can change as context changes.
As an Amazon Associate I earn from qualifying purchases.
Depending on plan, region, platform, and workspace, available context may include saved memories, relevant information from past chats, custom instructions, files in Library, and connected-app content such as Gmail. A memory summary is a high-level view, not necessarily an inventory of every source or factor used for a response.
Saved memories
Saved memories are details you ask ChatGPT to remember or that it may save as useful context when the feature is available. They are separate from the chat where you shared them. Deleting that chat alone does not necessarily remove the saved memory.
#1 Best Overall
Reference chat history
When enabled, this lets ChatGPT use relevant information from previous conversations. It does not mean every detail in every chat is retained or will be recalled. OpenAI says there is no separate storage limit for information it can reference through chat history; that is not a promise of perfect recall.
Memory changes and rollout
On June 4, 2026, OpenAI announced an improved memory architecture built on its background synthesis process, “dreaming,” initially rolling out to Plus and Pro users in the United States, with other countries and Free and Go users expected to follow over the coming weeks. That announcement describes a dated rollout, not a guarantee that every account now has the same controls. Check your own settings for current availability.
How to review, correct, or manage Memory
The usual route is Settings > Personalization > Memory. Names and screens vary; you may see controls such as Reference saved memories, Reference chat history, Memory summary, Manage, or Saved memories.
- Ask ChatGPT what it remembers to get a conversational overview.
- Correct outdated context directly, or edit the memory summary or an individual saved memory if those controls are available.
- Ask ChatGPT not to mention a detail again when you want to change how it personalizes responses. That request does not delete the underlying chat, file, email, or other source.
- If a Memory control is missing, feature availability may differ for your plan, region, platform, or workspace. Check the account settings and ask your workspace administrator if you use a managed account.
OpenAI notes that sources may appear with personalized responses, but they might not reveal every factor that influenced the answer. A summary or source display should therefore not be treated as a complete audit of all available context.
How to delete ChatGPT memories and their sources
Removing remembered context may require action in more than one place. Use the available Memory controls to remove saved items, and separately remove any original or connected material you no longer want available.
- Go to Settings > Personalization > Memory, then open the available memory summary or management controls and delete the saved item.
- Delete the chat where you shared or discussed the information if you also want that conversation removed. Deleting a chat by itself may not delete a separate saved memory.
- If relevant context came from a Library file or connected app, remove the file from Library or disconnect that app if you no longer want it available.
- Check other places the information may remain, including regular or archived chats and other sources.
OpenAI says Memory updates and deletion can take time to propagate. Logs of deleted saved memories may be retained for up to 30 days for safety and debugging; this is not a blanket guarantee about deletion of every copy in every system.
What turning Memory off does—and does not do
Turning off Memory does not delete past chats or, by itself, delete saved memories. OpenAI’s Help Center states, “Turning off Memory does not delete your past chats.” If you turn off Reference chat history, information remembered from past chats is scheduled for deletion from OpenAI systems within 30 days, while the original chats remain unless you delete them separately. Saved memories and their source material need separate attention. In some interface configurations, turning off Reference saved memories also turns off Reference chat history.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Memory, Temporary Chat, and model training compared
| Option or control | What it does | What it does not do |
|---|---|---|
| Saved memories | Carries selected details into future personalization; available controls may let you review or delete them. | Deleting the original chat alone does not necessarily remove a separately saved memory. |
| Reference chat history | Allows ChatGPT to use relevant information from previous chats when enabled. | Does not retain every detail. Turning it off does not delete the original chats or separately saved memories. |
| Temporary Chat | Does not use or create Memory, does not appear in history, and is not used to improve models while temporary. | Does not delete existing memories or previous chats. |
| Improve the model for everyone | Controls whether conversations on personal accounts are used to improve models. | Does not control whether Memory is referenced. |
Using Temporary Chat
Choose Temporary Chat when you want a conversation that does not use or update Memory for personalization. It is a separate option, not a way to erase earlier chats or saved memories. OpenAI’s privacy information describes Temporary Chat in this context.
Changing model-training preferences
For personal accounts, OpenAI’s Data Controls guidance says you can turn off Improve the model for everyone under Settings > Data Controls. Your conversations remain in history but are not used to train ChatGPT. OpenAI says content from Business, Enterprise, Edu, and ChatGPT for Healthcare workspaces is not used for training by default. This training preference is separate from Memory; see the Data Controls FAQ for details.
OpenAI also says turning off Memory or personalization does not disable limited safety and security uses of context in rare, high-risk situations. For current controls and account-specific availability, consult OpenAI’s Memory in ChatGPT help page.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
