API image creation moderation: improve details to include the reasons furnished to ChatGPT

[_j]

Reason for denial surfaced: ChatGPT

“similarity to third-party content” is shown and thus available

Reason for denial obfuscated: API

API returns only “other” from similar request:

moderation_details: {"moderation_stage": "output", "categories": ["other"]}

“Other” can also arise from obscure input reasons such as celebrity depictions - where OpenAI and its model knows the reason.

High-quality error reporting including moderation detail is requested.

[LarisaHaster]

I totally agree with this.

[_j]

The API telling us the image moderation reason, just as categories on the moderations endpoint produces scoring in discrete categories, would be feedback that could avoid misunderstanding, allowing redirection able to produce an acceptable product.

[Robertcarlos]

@_j

Reason for denial surfaced: ChatGPT

“similarity to third-party content” is shown and thus available

Reason for denial obfuscated: API

API returns only “other” from similar request:

moderation_details: {"moderation_stage": "output", "categories": ["other"]}

“Other” can also arise from obscure input reasons such as celebrity depictions - where OpenAI and its model knows the reason.

High-quality error reporting including moderation detail is requested.

I agree that clearer moderation feedback would be very helpful. Knowing the specific category or reason behind an “other” result would make it much easier for developers to understand and fix legitimate issues.