How stories are made.

Every story on FolioPress is AI-assisted and human-edited. Here is exactly what that means, and why we believe being upfront about it produces better fiction than pretending otherwise.

The short version

The drafts are produced by large language models working from carefully tuned prompts; the editorial judgement, prompt design, story selection, image curation, and final word on every piece is human. No story is published without being read end-to-end by a human editor.

If you would rather read fiction with no AI involvement at all, we respect that, and you should look elsewhere — there are wonderful publications that work that way. We are not trying to compete with them. We are trying to do something different: publish a steady stream of unusual, well-edited stories across genres that would not otherwise exist.

What the AI does

  • Drafts the prose. A language model writes the first draft of each story or anthology chapter from a detailed prompt that specifies tone, pacing, character expectations, structural rules, and quality guardrails.
  • Generates the cover and inline images. An image model produces every illustration. There is no stock photography on this site — every image is unique to its story.
  • Suggests metadata. Genre, tone, vocabulary, mood tags, content warnings, social captions, and discovery hints are proposed by the language model and then reviewed.

What the human does

  • Designs the prompts. The prompts that produce these stories took months to refine and are still being tuned. They enforce setting diversity, character distinctness, opening-line quality, and dozens of other constraints. The prompt is the editorial voice.
  • Reads every story end-to-end before publication. Stories that read like AI slop, repeat themselves, fall apart in the third act, or rely on cliché get rejected. Many drafts never reach the site.
  • Curates and discards. For every story you read here, others were generated and thrown out. The selection itself is a creative act.
  • Edits the prose where needed. Awkward phrasings, weak openings, factual slips, and continuity bugs are fixed by hand before publication.
  • Owns every editorial decision. Genre tags, content warnings, what gets featured on the homepage, what gets paywalled, what gets pulled — all human calls.

Why we are upfront about this

Hiding AI involvement is a short trip to a long crisis of trust. It also tends to produce worse work, because writers who pretend to be writing alone do not develop the editorial muscles a hybrid workflow requires. Saying “AI-assisted, human-edited” on every story is both honest and a quality forcing function — if a story is not good enough to put your name and editorial judgement behind, it should not ship.

What we will not do

  • We will not claim a story is human-written when it is not.
  • We will not invent fake authors or generate fictional bylines for stories.
  • We will not train models on reader data, comments, or accounts. Reader behaviour is used only to operate the site (recommendations, reading progress, account features).
  • We will not publish anything that has not been read by a human first.

If you spot a problem

Hallucinations, factual errors, plagiarism concerns, broken continuity, problematic tropes, or anything that feels off — please tell us. The fastest fixes here come from readers who notice things the editor missed. The contact form is on the Contact page.

Last updated: April 2026