I get the above message anytime I try to create a customGPT using my DM Tool Kit api dice roller for d20 game systems.
I see others published public regularly. My use case is to make different specialized versions, but as it stands only the one was allowed to finally be made public after giving me the same message as above for a while.
I’m at a loss to understand how I’m actually violating anything, and feel like its just not liking some of the words in the instructions without providing any clarity on which ones. Or it’s simple the external apis which are the trigger. Either way, some way to get detailed feedback would be nice. Like a button there to take us into the gpt builder chat dialogue so it could provide detailed conversational feedback on the specific issue and how to resolve it would be nice.
OpenAI has put in place a content moderator, checking for brands and suitable content.
The message seems to indicate nonconformity is detected.
Follow the links. Comply.
Actions also have unclear rules about the domains under which they can operate and your need for adding those domains or subdomains as verified and under your control in your builder profile.
Yeah, so like i intended to imply if i didn’t state it explicitly. I have followed the links, as far as I can tell by the definition of the words used in them, I’m not in violation actually. So I’m asking for more clarity. I asked GPT-4 to review the guidelines and my gpts instructions and it also couldn’t find a clear violation that it might be referring to. Hence my confusion.
I want to comply with the requirements, how?
I would minimize the title and description. See if “my new bot” will publish to “everybody” or “with a link”. If not, save the profile image you want locally and make a bot generated one also. Then build up the words and update until the cause is identified.
If still no joy, go after the actions it employs.
The message is more critical, saying that you can only publish to “only me”, while the domain verification and domain name profile enabled is required for going from “anyone with a link” to “everyone”.
If you’ve gotten "report"s pressed on it, then the GPT ID may have been flagged as unshareable.
That’s all useful information, and much appreciated. Yeah it seems procedural cloning with alterations until a discernable trigger(s) can be found is the only way forward at this point.