ChatGPT Memory: What It Remembers and How to Control It

ChatGPT Memory lets the assistant recall relevant details about you across separate conversations. To see what it remembers, open Settings > Personalization > Memory and look for the memory summary or saved-memory list. To turn it off, toggle Memory off in the same location. Turning Memory off does not delete your past chats.

Memory is not a recording of everything you have ever said. ChatGPT decides which available information is relevant to a response, and the set of remembered details can change as your context changes. The details in this guide follow OpenAI’s Memory documentation, checked November 2, 2026.

What ChatGPT Memory actually is

Memory is a personalization feature. When it is on, ChatGPT can use relevant information from your past chats, saved memories, custom instructions, saved files, and connected apps to shape its responses. The goal is to reduce repetition: if you told ChatGPT your preferred writing tone in one conversation, it can apply that in the next one without you repeating yourself.

Memory does not retain every detail from every conversation. OpenAI’s documentation is specific about this:

Memory does not retain every detail from every conversation. ChatGPT decides which available information is relevant to a response, and Memory can change as your context changes.

Source: OpenAI, Memory in ChatGPT, checked November 2, 2026.

This is not a transcript of your history. It is a selective summary that changes over time. If you asked ChatGPT to remember your dog’s name in March but never mentioned the dog again, that detail may or may not still be in Memory months later.

The sources Memory can draw from depend on your plan, region, platform, and workspace settings. They include past chats, saved memories, custom instructions, saved files available through Library or Space, and content from connected apps such as Gmail.

How to see what ChatGPT remembers about you

There are two ways to check. The first is the settings menu, and the second is asking directly.

Through settings

1. Open Settings.
2. Select Personalization.
3. Select Memory.

Depending on your plan and interface, you may see several controls: Memory, Reference saved memories, Reference chat history, Memory summary, Manage, or Saved memories. Not every option appears on every account.

Through conversation

You can also ask ChatGPT directly: “What do you remember about me?” The response gives you a conversational summary, but the settings view is more complete because it shows individual saved memories you can edit or delete.

When the memory summary is available, you can:

• Enter a correction in the text field.
• Highlight text and provide a correction.
• Select Don’t mention this again when that option appears.
• Select Delete and turn off memory from the more options menu (three dots).

Don’t mention this again reduces future references to a piece of information. It does not delete the original source. If you want the information gone, you need to delete it from its source, not just tell ChatGPT to stop mentioning it.

Memory summary vs saved memories

These are two different things, and confusing them leads to frustration.

Memory summary is a high-level view of information ChatGPT may use. It does not necessarily include every detail or source. Think of it as a brief profile, not a complete record.

The memory summary is a high-level view of information that ChatGPT may use to personalize responses. It does not necessarily include every detail or source that Memory can reference.

Source: OpenAI, Memory in ChatGPT, checked November 2, 2026.

Saved memories are individual details that you explicitly asked ChatGPT to remember or that ChatGPT saved as useful context. They are stored separately from chat history. Deleting the original chat does not automatically delete a saved memory created from it.

OpenAI introduced two Memory experiences that you may be able to switch between:

Changing the Memory experience does not delete your chat history. If you switch from improved Memory to legacy saved memories, your chats remain where they were.

How to delete what ChatGPT remembers

Deleting remembered information requires identifying where it is stored. A single piece of information can exist in multiple places.

Delete a saved memory

Open Settings > Personalization > Memory > Manage or Saved memories, find the item, and delete it. Deleting a saved memory prevents it from being used in future personalization. It does not remove mentions of that information from past conversations.

Delete the chat where you shared the information

If the information was mentioned in a chat but not saved as a separate memory, deleting that chat removes the source. Follow the ChatGPT history deletion guide for the steps.

Delete files from Library or Space

If the information came from a saved file, delete the file from Library or Space separately. Deleting a chat does not delete files saved separately.

Disconnect a connected app

If the information came from a connected app like Gmail, disconnecting the app prevents future access. It does not delete past conversations that already used app data.

To remove information saved as a memory, delete the saved memory and the chat where you first shared it. Also remove the information from any other sources where it appears.

Source: OpenAI, Memory in ChatGPT, checked November 2, 2026.

OpenAI may retain logs of deleted saved memories for up to 30 days for safety and debugging purposes. Deleting a memory from your view is not the same as immediate erasure from all systems.

How to turn off Memory entirely

Turning off Memory stops ChatGPT from creating new remembered information from future chats. It does not delete existing memories or past chats.

1. Open Settings.
2. Select Personalization.
3. Select Memory.
4. Toggle Memory off.

Turning off Memory does not delete your past chats. If you turn Memory on later, ChatGPT may create new remembered information from chats that remain in your history.

Source: OpenAI, Memory in ChatGPT, checked November 2, 2026.

If you want a clean break, use Delete and turn off memory from the memory summary’s more options menu. This deletes the remembered information shown in the summary and turns Memory off in one step. It does not delete past chats.

What turning off Memory does not do

Turning off Memory or personalization does not disable limited safety and security uses of context in rare, high-risk situations. OpenAI describes this as separate from personalization controls. You cannot switch off safety features through the Memory toggle.

For a conversation that should not use any existing memories, start a Temporary Chat and choose Unpersonalized. Temporary chats do not create or update memories while they remain temporary.

Reference chat history explained

If your settings include Reference chat history, turning it on lets ChatGPT use relevant information from past conversations to personalize future responses. This is different from a saved memory.

A saved memory is an explicit item. Reference chat history is broader: ChatGPT can pull relevant context from past conversations without you having asked it to remember anything specific.

If you turn Reference chat history off, information remembered from past chats is scheduled for deletion from OpenAI systems within 30 days. The original chats remain in your history unless you delete them separately.

Turning off Reference chat history does not delete your saved memories. The two controls are separate, though turning off Reference saved memories also turns off Reference chat history.

How Memory interacts with other ChatGPT features

Memory does not operate in isolation. It overlaps with several other features, and the boundaries matter.

Memory and Temporary Chat

Before starting a temporary chat, you choose whether it uses existing memories, custom instructions, and plugins. Select Unpersonalized if you do not want them used. Temporary chats do not create or update memories, even with personalization on.

Saving a temporary chat converts it to a regular chat. If Memory is on, ChatGPT may then use the saved conversation to personalize future responses.

Memory and archived chats

Archiving a chat hides it from your sidebar. It does not delete the conversation or any memory created from it. An archived chat remains saved under the same retention rules. If you want to remove a memory, deleting the archived chat is a separate step from archiving it.

See how to find archived chats for archive management steps.

Memory and locked chats

When Memory is enabled, locked chats can use relevant information from regular Memory. New messages sent while a chat is locked do not personalize regular chats. Locking a conversation does not remove information that ChatGPT remembered before it was locked.

After permanent unlock, the full conversation may be used for regular Memory if your settings and workspace policies allow it.

Memory and connected apps

ChatGPT may use content from connected apps like Gmail when those features are available and enabled. Disconnecting an app prevents future access to that app. It does not delete past conversations that already used app data.

Sources: a new way to check personalization

When ChatGPT uses relevant personal context, Sources may appear below the response. Sources can help you review what contributed to a response: a past chat, saved memory, custom instruction, file, or connected email.

Depending on the source and available controls, you may be able to:

• Open or review the source.
• Correct or delete a saved memory.
• Delete a referenced chat.
• Mark a source as relevant or not relevant.
• Open a referenced file or email.

Sources may not show every factor that shaped a response. Memory Sources are not included when you share a conversation through a shared link. If you are reviewing what ChatGPT knows, use Settings rather than relying on Sources alone. For a fuller list of what link viewers can and cannot see, see ChatGPT Shared Links.

Privacy and model training

For personal ChatGPT accounts, OpenAI may use chats and remembered information to improve models when Improve the model for everyone is on. You can turn that setting off in Data controls.

By default, OpenAI does not use content from ChatGPT Business, Enterprise, Edu, or ChatGPT for Healthcare workspaces to train its models.

Sensitive information can appear in Memory if you share it with ChatGPT. Reviewing Memory settings and Sources regularly is the practical defense. For conversations that should not use or create memories, choose a non-personalized temporary chat.

Memory in workplace and healthcare accounts

Improved Memory can use context from past chats to keep its memory current. In ChatGPT for Healthcare and ChatGPT Enterprise with Regulated Workspace, improved memory is disabled by default.

Workspace owners and admins can make improved memory available to the default workspace role or eligible custom roles. If a memory setting is unavailable in your workspace, contact your workspace admin.

This feature is not covered under your BAA. PHI should not be entered when using this feature.

Source: OpenAI, Memory in ChatGPT, checked November 2, 2026.

This means that if you work in a healthcare setting with a Business Associate Agreement, the improved memory feature is not part of the covered services. Do not enter protected health information expecting that Memory will handle it compliantly.

Members who choose to use improved memory in regulated workspaces can use Project-only memory to limit context to a specific project. Chats within that project can reference each other, but cannot reference conversations outside the project, and outside chats cannot reference conversations in it.

A five-minute memory check exercise

Practice finding and managing what ChatGPT remembers without sharing anything sensitive.

1. Open Settings > Personalization > Memory.
2. Read the memory summary or saved-memory list.
3. Pick one item that is outdated or unimportant.
4. Delete that single item using the available control.
5. Start a new chat and ask: “What do you remember about me?”
6. Compare the response to what you saw in settings.

The response from the chat may differ from the settings view. The settings list is more complete. If the deleted item still appears in the response, it may be referenced from a chat rather than a saved memory. Check whether the original chat still exists.

Finish by deciding whether to leave Memory on or turn it off. There is no universally correct answer. Memory is useful for reducing repetition in low-risk tasks. It is less suitable when the information in your conversations is sensitive or changes frequently.

Questions beginners often ask

Does ChatGPT remember everything I say?

No. ChatGPT decides which information is relevant and retains that. It does not keep a word-for-word transcript of every conversation in Memory. Past chats remain in your history, but Memory is a selective summary, not a full recording.

How do I delete all ChatGPT memories?

Open Settings > Personalization > Memory and look for Delete and turn off memory in the more options menu (three dots). This deletes the remembered information shown in the summary and turns Memory off. It does not delete past chats or separately saved files.

Is Memory the same as custom instructions?

No. Custom instructions are direct guidance you provide about what ChatGPT should know and how it should respond. Memory is information that develops across conversations and other sources. Custom instructions are always active when enabled. Memory is selective and changes over time.

Can someone else see my ChatGPT memories?

Memory Sources are not included when you share a conversation through a shared link. However, workspace admins may have access to supported memory settings in enterprise accounts. For personal accounts, someone with access to your signed-in device can view Memory settings.

Does turning off Memory delete my chats?

No. Turning off Memory stops new remembered information from being created. Your existing chats remain in your history. If you turn Memory back on later, ChatGPT may create new memories from chats that are still in your history.

When to ask a person

If you entered sensitive personal, medical, financial, or workplace information and want to ensure it is removed, review Memory settings, Sources, connected apps, and saved files. For workplace accounts, contact your administrator before making changes to business conversations. For health information in a regulated workspace, follow your organization’s compliance process rather than relying on Memory controls alone.

Keep learning in small steps

Compare Memory with Custom Instructions and Temporary Chat to understand which control fits each situation. The AI safety and privacy guide covers broader upload decisions. You can also take the AI level test to find a practical privacy habit to work on next.

A final reminder

Memory is a convenience feature, not a vault. Review it regularly, and share only what you would be comfortable seeing in a summary.