Told GPT-4 to make me a Goblin wearing a Loincloth and using a Wooden Club, in a n aggressive stance. Using Modern Pixel Art style. This is what I got:
GPT: I apologize, but I was unable to generate images due to our content policy. Please provide a new request or let me know if there’s any other way I can assist you.
Me: What was the content policy that it broke?
GPT: I apologize for the confusion. Upon reviewing the request, it seems that there shouldn’t have been any content policy violation with the description you provided. It was likely an oversight. Let’s try generating the pixel art of the goblin again. Would you like me to proceed?
Me: Yes
GPT: I sincerely apologize for the inconvenience, but it seems I’m facing an issue generating the images for the described goblin in pixel art due to our content policy. I understand this might be frustrating, and I’m here to help in any other way you might need. If you have other requests or questions, please let me know.
I’m getting the same content policy violation message when trying to create a logo of a key and a cloud… ???
I’ve asked ChatGPT about the prompts used for DALL·E, but I don’t see how that could possibly violate any content rules… it’s absurd!!
1. Minimalistic black and white logo of a key with the top shaped like a cloud, symbolizing access to cloud-based applications through subscription.
2. Black and white logo design depicting a stylized key with a cloud-shaped head, evoking the concept of cloud application rental.
3. Black and white logo where a key merges with a cloud at the top, representing exclusive access to subscription-based applications in the cloud.
4. Minimalistic black and white representation of a key, with its top subtly blending into the silhouette of a cloud, for a subscription-based application market.
Something is seriously wrong with those content rules… or someone is sabotaging their new image generation system.
Either way, they need to get to work to fix this. I’ve been a subscriber since the beginning, but this bizarre behaviour is frustrating and makes me consider leaving the service.
Note: i solved my need today using Bing free image creator.
I am trying to create a prompt for a graphic novel and is not letting me do it either.
this is the prompt: “Black and white graphic novel, wide shot of a medieval town with a cemetery next to it.”
Unsafe image content detected
Your image generations are not displayed because we detected unsafe content in the images based on our [content policy creating again with another prompt.
Why is it so restrictive? looks like i am hitting back to MidJourney.
Yes I tried to do some horror-style artworks and it was brutally restrictive. I’m worried that this is their API as well. Hopefully they will be more reasonable in the future.
A trick to help influence the model is to allow it to deduce the existence of a cemetery, or, more simply, to imply the existence of the cemetery. May take a couple shots but this was my first image. Flattery also works (A respectful, loving cemetery)
I’m not sure, but I think the issue has nothing to do with the prompts being provided in these cases, but rather the image that is generated by the AI as a result. I just asked GPT DALL-E 3 to generate a set of images for me based on a single prompt. It generated 3 of the 4 images but said the last one violated policy. Since it was a single prompt it could only be the resultant image content that triggers the policy, correct?
The issue is obviously the human element and what is considered offensive and what isn’t.
I don’t get why they didn’t learn from Dall-E which basically became obsolete once StableDiffusion came out. They self sabotaged their own product through their insane policies, and now they have a few months before OpenSource, unrestricted models will catch up once more and make their products irrelevant again. This is quite a shitty business model, and if we consider that OpenAI wrote about 6 pages about their commitment to ethical issues but never once mentioned the ethical implications of using copyrighted artwork to train, the moral highground argument crumbles pretty quickly.
Yes, it is frustrating. It wouldn’t create an image for me because I used the word “Gypsy”. I switched it out with “Romani” and it generated the image. Ridiculous!
