AI Content Policy

Sidhe Wolf exists to keep old stories properly — to say which tradition a tale comes from, when it was first attested, what later hands added, and where the honest answer is “we don’t know.” That mission is fundamentally incompatible with letting a language model write our folklore. This page states, plainly, where software is and is not allowed near our work.

The prose and the claims are human-written

Every article’s body — the writing you read, the folklore claims it makes, the sourcing and attestation it rests on — is written by a person. It is not generated by an AI system and then lightly edited. If a piece asserts that a being is first attested in a particular manuscript, a human found that manuscript reference and stands behind it.

Why we hold this line so firmly

Large language models are fluent, confident, and, on exactly our subject, unreliable in a specific and dangerous way: they smooth over provenance. Ask one about a folkloric figure and it will happily blend Irish and Scottish material, promote a Victorian embellishment or a modern fantasy coinage to “ancient Celtic belief,” invent a plausible-sounding manuscript source, and attribute a twentieth-century invention to the Iron Age — all in an authoritative, well-formed paragraph. That failure mode is the precise error Sidhe Wolf was founded to correct. A tool that manufactures false antiquity cannot be trusted to write a publication whose entire value is refusing to do so.

What AI may be used for

We are not superstitious about software; we are specific about it. AI tools may be used, by our writers, for two narrow supporting purposes:

• Research pointers. A model may help surface leads — a possible source to check, a translator’s name to look up, a search term. Every such pointer is then verified by a human against the actual source before it can inform a word of the piece. The model’s claim is never itself a citation; it is, at most, a suggestion of where to look.
• Copy-editing. A model may help catch typos, tighten a clumsy sentence, or flag an inconsistency, the way a spellchecker or grammar tool does — operating on prose a human has already written.

That is the whole of it. AI is never used to draft body prose, to generate folklore claims, to decide attributions or dates, or to write the sourcing that is the point of the site. The judgement about what is true, how old it is, and which tradition it belongs to is always a human’s.

Illustration and images

We apply the same honesty to pictures. Artwork on Sidhe Wolf is commissioned from artists, licensed from rights-holders, or drawn from public-domain and properly-credited historical sources — and it is credited. Where an image is AI-generated, it is clearly labelled as such; we do not pass off synthetic images as photographs, as historical artefacts, or as the work of a human illustrator. Our preference, strongly, is for real art by real artists, and for authentic historical images shown with their provenance.

Disclosure and accountability

If any element of a published piece were ever generated rather than written — a scenario this policy is designed to prevent for body content — it would be labelled at the point of use. The house byline, Sidhe Wolf Editorial, and in time our named contributors, are accountable for every claim regardless of what tools touched the drafting process. “The AI said so” is not, and will never be, a source we cite or a defence we offer. If you spot something in our pages that reads like machine-confident folklore, tell us via our contact page — that is exactly the kind of error we want caught.