I recently ran into a frustrating issue while trying to use DALL·E and GPT-4 Turbo (4o). Apparently, the models are so heavily restricted that they can’t generate something as simple and innocent as a cartoon dragon or goofy hydra. Both attempts were blocked, citing violations of OpenAI’s content policy.
These creatures have been staples of mythology and storytelling for centuries, and I made sure to keep the prompts playful and non-violent. Despite that, the system flagged them, leaving me puzzled. A dragon and a hydra, especially when drawn cartoonishly, should be far from problematic content. The response I got suggests that OpenAI’s moderation system might be dialed up a bit to far—effectively preventing even creative, harmless outputs.
This seems to be a growing frustration across the community. Other users have reported similarly unexpected content blocks when requesting innocent imagery
. While I understand the need for safety and content moderation, I believe OpenAI has gone too far in this case. It’s disappointing to encounter these limitations in a platform meant to foster creativity.
Has anyone else run into this issue? If so, how are you working around it? Is there a way to petition OpenAI to relax the moderation a bit for content like this? Dragons and hydras are iconic in so many stories—blocking their creation seems counterproductive.
Would love to hear your thoughts or suggestions!