Hi there, I am trying to upload a custom image to use for my first GPT that will be submitted to the GPT Store. I have tried various suggestions to no avail including changing file type, importing from different places, changing browsers, limiting files size, etc. I am getting an error code and it is not saving the version with my new image even after it appears to be updated. This is the error code “Error saving draft”. Thanks in advance!
Create a new GPT to try out images in a clean slate. Just ask the GPT creator “exit the interview process. Just create a GPT that acts like a monkey character named Bobo”. Then see if that will accept your image upload.
You can ask for a DALL-E AI image to be generated for your GPT profile, and see if that gives you success. Or if you still have problems updating even with no images, that may originate elsewhere in the existing GPT.
Then, for the actual image you upload, I would emulate the profile image format exactly: 512x512 pixels, a PNG file in 24 bit full color. See if that gives you success in saving.
If you have tried to submit prohibited terms like trademarks, company names, copyrighted character names too many times and gotten denials, the GPT updating may simply be blocked and the whole GPT needs to be discarded.
In that case it is better just to start again with a new one, slowly updating the “configure” text boxes and ensuring good updates and publishing.
BTW: If you don’t want your personal given name shared also in the store, you’ll need a website under your control that you can verify instead, in “settings”.
An AI-created image may be fine, they are small, and nobody really cares. Good luck!
Thank you. I started a new GPT as you suggested and am running into the same error with the GPT logo issue.
You don’t say if ANY image fails, or if just YOUR image fails.
Computer vision is used to inspect the images when an attempt to share is made. As an example of what won’t be published and continues to revert to “only me”:
That inspection could be what you are experiencing if you persist with one image the vision AI icon moderator won’t allow through.
Any image i have uploaded fails. The GPT generated image does not.
Note my comment about “the whole GPT has to be discarded”.
If a denied image has been submitted, this may permanently prohibit the use of uploaded images in a particular GPT. The same is seen with the textual content being flagged (such as using an unapproved domain name).
OpenAI may want to stop continued attempts at variations of denied images, users probing to discover what will go through, while the whole time each try is using AI computation to inspect.
Another factor: the AI is just bad. Denials have been recently produced for previously working image use cases, and even inputs like “I got this error in my code” have produced content policy violations. Bad decisions may be made based on personal identification criteria or extrapolation on possible misuse (like if you were to upload a picture of an attractive woman for store share).
The best technique would be to copy all the text boxes to a notepad editor. Then start anew with “Create a GPT”. Complete all the text fields, upload the image, and give a single press of “Create” with the selection to share to the GPT store. See if it doesn’t go through. Review “everyone” status by looking at “My GPTs”, not by further button presses.
I doubt a desired picture alone will be crucial to the success of a GPT, though. Best of luck.
1 Like
Thank you. I was eventually able to upload the custom image. I think the issue was that my Builder profile wasn’t complete and/or my usage limit for the day had been exceeded.
1 Like