Hi everyone,
I’ve been working on a longform literary project using ChatGPT called Unto a Golden Dawn: The Recursion Dossier—a recursive, metafictional story that weaves historical figures like Poe and Crowley into a symbolic and surreal narrative world. It’s a hybrid of dossier fiction, archive horror, and philosophical myth-building.
For months, I’ve used ChatGPT to generate stylized imagery to support this world—gothic mirrors, cracked reflections, conceptual depictions of characters like Poe and Crowley in abstract, fictional contexts. The work is respectful, original, and literary in scope.
Recently, though, I’ve run into a serious problem: the image generation system is now consistently blocking prompts that reference these historical figures—even in clearly creative, symbolic settings. This wasn’t the case weeks ago. The same applies to my memoir The Cancer Diet, where I hit strange content policy walls trying to create an honest, representative cover.
The policy seems to have shifted—without clear documentation—and is now suppressing exactly the kind of experimental, writer-driven usage this tool made possible in the first place.
Let me be clear: I’m not asking to generate misleading content. I’m asking to continue telling my story. To represent literary figures in a deeply personal, emotionally truthful way within a clearly fictionalized world I’ve spent hundreds of hours building.
What I’m experiencing now feels like a rollback. It’s disheartening. And it risks pushing serious creatives—especially those working on emotionally layered or recursive works—away from the platform.
Is anyone else encountering this? Is there a workaround or channel where policy exceptions for literary use can be considered?
Appreciate any insights or solidarity. This is a powerful tool. I’d just love to keep using it without having to fight it every step of the way.
—Frank
Author of Empire, Nevada, The Cancer Diet, and Unto a Golden Dawn