Google drive (and other) connector no longer available in customGPTs

Since a day or 2, I am no longer able to upload files via Google drive or any other connector. I can only upload files from my device. The connector options are gone?? They only show up in a normal chat and projects.

I also am not able upload a profile image or generate with dalle a profile image for the customGPT. It creates an error.

Anyone else having this problem?

1 Like

I experience the same. In fact, I’ve never been able to use ChatGPT Connectors like Slack or Google Drive within our Custom GPTs (something we really miss). I also haven’t been able to find any information explaining why those aren’t available there.

It’s unclear to me form the documentation as well - https://help.openai.com/en/articles/11487775-connectors-in-chatgpt

They’ve refunded and replied:

Regarding your report about connectors within Custom GPTs, after a deeper review, we were able to confirm the following:

  • Connectors and integrations (such as Google Drive access) are currently supported only for newly created Custom GPTs.

  • Existing Custom GPTs made prior to this update may no longer retain connector access due to a recent change in how integration permissions are managed at the account level.

  • To enable connectors in your Custom GPT, you can create a new Custom GPT and link your integrations again. Simply go to your ChatGPT home screen, select Explore → My GPTs, then click the Create (+) button. From there, click the company knowledge button — this will display the available connectors, including Google Drive and other integrations you can reattach.

I tried that out too and the issue is not resolved.

However,

bye bye chatgpt for me.

1 Like

It sounds like you got a bot answer. It is answering after a web search finds the ChatGPT enterprise feature “company knowledge” online.

You can assume the rest is mistruth and fabrication, a demonstration of why AI is not something to face customers, despite two years of promises. It is a step below third-world outsourced human script-readers as support that say “thank you mister Jay” and then put you on hold because they have no concept or agency either.


Can’t even create a profile image? Could be a content moderator on the prompt or vision on the upload.

Can’t update or publish? One denial with the “appeal” button, and that GPT ID is permanently useless.

I tried to see where I got failings, but the failing was immediate - immediately having me wait for 20+ minutes to have nothing about the GPT ID updated after reloading on a separate page with the GPT URL (and this is still hanging with no more progress bar to grow).

What is curious is the builder’s use of a model that is not specifically-chosen for the application, making one wait around for a long “pro” thinking time to show the results.

Try #2, after selecting GPT-5 “auto” in the preview pane (which implies it is just for the testing):

Neither of those ever recovered back to a completed AI builder response. At this point of “Updating GPT”, it has written a seen title and conversation starters, but no image is ever fulfilled by image gen, and again just hanging.

Simply: the profile image button for “use dalle” alone (yes, it still says dall-e) is non-functional.

Not that I would ever talk to the AI to do GPT stuff, but one must conclude GPT:“broken” - regardless of whether you are on a future “approved partners” list or have the faculties to actually use “actions” correctly instead of connecting consumer cloud products.