MemoryBrain

MemoryBrain › Guides

Why does ChatGPT stop saving memories after using Gmail, Drive or other apps?

OpenAI says some sources or workflows prevent new memories from being saved. That does not confirm the cause of every reported failure after app use. Check the saved-memory list before and after; ChatGPT saying “saved” does not prove the fact persisted.

By Carlos Moya. Updated November 8, 2026.

Check whether it was actually saved

  1. Review memory before using the app

    Open Settings › Personalization › Memory and review Saved memories or the control available to your account, following the official help. Note your interface. If you only see a summary, do not treat it as an exhaustive record list.

  2. Run a control without an app using fictional data

    In a new chat, ask it to save a distinctive test preference, such as “For the fictional Lighthouse project, I prefer numbered lists.” Check whether it appears in the controls. If this control fails, you cannot attribute what follows to app use.

  3. Use an app once and check again

    In the same chat, make an innocuous request using the app. Then ask it to save a different fictional preference. Return to the controls and compare what appears. Do not use private data to reproduce the issue.

  4. Query from another chat and record the result

    Ask about the test preferences in a new chat without revealing their content. Compare the answer with the list where available. Record the date, model, app, client and whether the fact appears. Remove the test memories afterward.

  5. If it persists, give support evidence

    Include the order of steps, conversation links you choose to share and whether each fact appeared in Saved Memories. These are the checks support requests in the thread. Avoid repeated writes without reviewing their results.

What is known

OpenAI’s Google app data controls FAQ says eligible information from these apps may personalize responses when Memory is enabled. Saving a new memory depends on the source, workflow and applicable restrictions. Some sources or workflows prevent new memories from being saved.

The FAQ also separates the controls: disconnecting an app does not automatically delete existing conversations or saved memories. Personalizing an answer using app information and creating a new Saved Memories entry are different outcomes.

What was reported and what remains unknown

In thread 1395159, the author reports saving a test fact before using an app but not another afterward, within the same chat. They report cases involving Google services and Notion, with behavior varying between tests. These are user observations, not a confirmed cause.

In its October 26, 2026 reply, OpenAI Support points to the FAQ but says that restriction does not confirm the explanation for this case, especially with non-Google apps. It requests conversation links and confirmation of whether each test appeared in Saved Memories.

The cited sources do not establish which internal condition triggers each failure, which accounts are affected or a general fix. They also do not establish that every connector always disables memory or that earlier memories have been deleted.

How to interpret the test

Opening another chat can provide a comparison. It is not presented here as a guaranteed repair.

Native memory and saving externally

MemoryBrain saves notes with sources in personal memory for ChatGPT and Claude over MCP. Writing there and saving a native ChatGPT memory are separate operations. It is not a fix for this issue: if you save externally, also check that you can retrieve the note and its source.

More: what each assistant remembers and how to check a correction without duplicates.

Frequently asked questions

Do Gmail or Drive always disable memory?

The cited sources do not establish that. OpenAI says some sources or workflows prevent new memories from being saved, and support does not confirm that this explains every case in the thread.

Does this mean my old memories have been deleted?

That cannot be concluded from the thread. The described case concerns new writes after app use. Review existing memories separately.

Does disconnecting the app delete what ChatGPT already remembered?

Not automatically. OpenAI’s FAQ separates disconnecting an app, deleting conversations and removing saved memories. Review each control according to what you want to remove.

Try it

See how a memory that keeps the source of everything works, no account needed.

Sources