AI Policy

This page is a commitment, not a feature list. RuleBindery's target market — independent TTRPG publishers — faces real, concrete penalties for AI-generated content: ineligibility from industry awards, marketplace bans, and public bans from major publishers and platforms. We built RuleBindery for that market, and this policy is why.

What RuleBindery never does

• RuleBindery does not generate publishable content. No prose, no statblocks, no tables, no art. Not in the current version, not on the roadmap.
• RuleBindery does not train on your content. Ever. There is no opt-out toggle, because there is no opt-in — your project content is never used to train any model, by us or by anyone we work with.

What limited assistance is permitted

Some machine assistance is genuinely useful and doesn't cross the line above, because it's deterministic and inspectable rather than generative. Specifically:

• Reference validation (catching a broken link to a deleted or missing entity)
• Terminology-drift detection (flagging near-duplicate names that are probably the same thing, spelled two ways)
• Duplicate-name detection
• Orphan detection (entities defined but never referenced anywhere)

These are static analysis — the same category of tool as a spell-checker or a linter. None of them write, rewrite, or suggest new content on your behalf.

If this ever changes

Any future feature that could reasonably be described as generative will:

• Require explicit, per-project opt-in
• Be disabled by default
• Stamp the export metadata of anything produced with it, so you can honestly declare your creation method to marketplaces and award bodies

Our public position

Human-authored by design. That's the whole policy, stated plainly.