The moderation done that is over the top is highly keyword-based.
Here, the fourth pointing to a direct protest about DALL-E 3 being degraded by a poor hyper-reactive filter, with no action.
They fixed the error message concern, I suppose.
For me, the exact snippet you say was part of more: no problem, except the elements described are completely non-followed. On point with expectation now - the quality of the model has been degraded, just like DALL-E 2, at the same price-per-image. $0.04 or $0.06 per image is less desirable that getting $0.25 out of you for gpt-image-1…
You could use a language AI on it, replacing any interpretation of a copyrighted character name or situation, and any depiction of conflict. That would damage the whole point of this prompt, though. Read a nametag.
The gpt-image-1 version:
