Hello OpenAI team and community,
I’m a paying subscriber and a lifelong student—forty years studying philosophy, faith, and their impact on humanity. I’ve greatly appreciated the capabilities of ChatGPT and its image tools for exploring storytelling and education. But I’ve now run into what I believe is a significant flaw in the current content policy.
Recently, I attempted to generate a stylized image of Logen Ninefingers, a fictional character from Joe Abercrombie’s First Law series. He is, by definition, a brutal character—missing a finger, drenched in the moral ambiguity and violence of a berserker archetype. The model blocked the image generation, citing content policy—even though the prior attempts were allowed. The inconsistency is frustrating, but more importantly, the inflexibility is damaging to artistic and educational use.
We are witnessing real-world violence and genocide, often under the justification of faith. As someone who studies this deeply, I find it irresponsible and counterproductive that the platform blocks fictional, symbolic, or fantasy depictions of violence while claiming to support exploration, education, and creativity. We cannot challenge radicalization or help others reckon with human nature if the tools refuse to acknowledge the rough edges of storytelling.
I understand the need for safeguards. But fiction and philosophy require room to breathe. If I can’t generate an accurate depiction of a character like Logen Ninefingers—with nine fingers and a bloodied axe in a dark fantasy scene—then the tool is missing part of its potential.
I ask the OpenAI team to consider more nuanced content moderation for fictional, mythological, literary, and historical content. Blanket bans don’t just protect—they restrict growth and conversation.
Sincerely,
Robert L.