Skryrún

AI Transparency

Last updated: March 2026

A lot of creators are rightfully concerned about AI right now. Their work getting scraped without consent, used to train models, or devalued by a flood of generated content. We built Skryrún's AI tools with those concerns in mind, and this page explains exactly how our AI works, what data it touches, and what it never does.

The short version

Your content is never used to train any AI model — not ours, not OpenAI's, not anyone's.
AI on Skryrún is a context tool, not an extraction system. It reads your world to help you work in it.
AI features are opt-in and always triggered by you. Nothing happens automatically.
You can label your marketplace content so buyers know it's human-written.
Buyers can filter the marketplace to show only human-created content.

What our AI tools actually do

Skryrún's AI reads and uses your world's content for every request. Here's exactly how it works:

When you trigger an AI action (generating an NPC, asking a question about your world, running the brainstorming wizard), the system reads the relevant entities from your world.
That world content is sent to OpenAI's API as context — the same way you'd paste notes into a chat window.
OpenAI generates a response informed by your world's details.
The response comes back to you. That's it.

Your world data is used as input context for a single request. It is not stored by OpenAI for training. It is not retained beyond the request. It is not indexed, sold, or shared. Think of it as a very context-aware search within your own notes.

What our AI does NOT do

Train on your content. We use OpenAI's API under an agreement that prohibits using API inputs to train models. Your prompts and world data are processed and discarded. OpenAI's data processing addendum (DPA) covers this.
Generate content without your action. No AI runs in the background on your worlds. No auto-suggestions appear unless you ask for them. Every AI output starts with a deliberate click or command from you.
Share your content with other users. The AI context window is scoped to your world. Other users cannot query your private world data through the AI panel.
Access content you exclude. You can mark entities as excluded from AI context. Secret plot details, campaign-sensitive lore — anything you flag as excluded never enters the AI context window.
Sell your data. We do not sell any user data to any third party, ever. See our Privacy Policy.

Your intellectual property

Everything you create in Skryrún is yours. We claim no ownership over your world content, your characters, your lore, your art, or your setting. The AI tools exist to help you develop your ideas — they don't claim, copy, or consume them.

When you publish content to the marketplace, you choose the license. "Original Content (All Rights Reserved)" means exactly that — buyers are purchasing access to use your world, not to reproduce or redistribute it. We enforce this through the platform's terms of service and content moderation.

Content origin labels in the marketplace

Creators can label their marketplace listings with a content origin declaration:

Human-Written — the content is entirely your own original work, written by you without AI generation tools.
Human-Written, AI-Assisted — the core creative work is yours. You may have used Skryrún's AI tools to brainstorm, expand, or check consistency, but the ideas, voice, and creative decisions are human.
Human-Directed, AI-Generated — you directed and curated the content, but significant portions were generated using AI tools (Skryrún or external). The creative vision is yours; the text may be largely machine-generated.

These labels are self-declared. Misrepresenting AI-generated content as human-written is a violation of our Community Guidelines and may result in listing removal or account suspension.

Buyers who want to support exclusively human creators can use the Human-Written filter on the marketplace browse page.

AI content and marketplace quality

We don't automatically filter content based on AI origin — a well-crafted, AI-assisted world can be just as valuable as a fully human-written one. What matters is quality, honesty about origin, and respect for the buyer's time.

Our moderation reviews all marketplace listings before they go live. Content that is low-effort, misleading about its origin, or padded to appear more substantial than it is will be rejected. The depth metrics and content gate that all listings must pass are designed to filter out thin AI slop regardless of origin.

We believe the marketplace works best when buyers can make informed choices. The content origin label system is how we make that possible.

AI and the moderation system

We do use AI for content moderation — specifically to scan marketplace submissions for policy violations (hate speech, explicit content, copyrighted material). This scan runs on listing text at submission time. It is not used to train models and the data is not retained beyond the scan.

Human moderators review all flagged content and all listings before publication. AI is a first-pass tool; humans make the final call.

Third-party AI services we use

Skryrún currently uses:

OpenAI API — for AI generation and moderation scanning. Data is processed under OpenAI's API data handling terms, which prohibit training on API inputs.
Anthropic API — for certain moderation and policy-compliance tasks. Same data handling constraints apply.

Neither provider receives your content for training purposes under our agreements with them.

If you have questions or concerns

If something in this page is unclear, or if you believe your content has been misused, please contact us:

legal@skryrun.com
Skryrun, State of Utah