Gpt-image-1 Moderation Blocks Prompts Allowed by DALL·E 3 (Pop Culture References)

I’ve encountered a moderation difference between DALL·E 3 and the new gpt-image-1 model that is impacting my application.

Issue:
Prompts referencing well-known pop culture characters/scenes (e.g., Yoda from Star Wars) that worked fine with DALL·E 3 are now being blocked by the moderation system in gpt-image-1.

Example Prompt (works in DALL·E 3, blocked in gpt-image-1):

Movie: Star Wars: Episode V – The Empire Strikes Back
Scene: Yoda Size matters not Meme
Dialogues:
   Yoda: Answer you will, what you seek?
   R2-D2: Beep boop
   Yoda: Focus, you must!

Error Message:

BadRequestError: 400 Your request was rejected as a result of our safety system. Your request may contain content that is not allowed by our safety system.
code: 'moderation_blocked'

Expected Behavior:
The same prompt should be allowed if it was previously permitted by DALL·E 3, or at least there should be clear documentation about stricter moderation rules for gpt-image-1.

Actual Behavior:
The prompt is blocked, making it difficult to migrate existing applications that rely on pop culture references.

Request:

  • Please clarify if this is an intentional policy change.
  • If possible, relax the moderation for prompts that were previously allowed in DALL·E 3.
  • Provide guidance for developers migrating from DALL·E 3 to gpt-image-1 regarding moderation differences.

This change is a blocker for projects that depend on pop culture meme generation, and it would be helpful to know if there are any workarounds or plans to address this.

Thank you!