Content Ratings and Safety

Last updated: September 5, 2026

1. The rating tiers

Every reader-visible story resolves to one of four content-rating tiers, ordered by increasing maturity: Everyone, Teen, Mature, Adult. This is the one canonical rating vocabulary used everywhere on the platform. A story uses its own `rating_tier` when set, otherwise it inherits its world's current `content_rating`; an unresolved rating defaults to Teen, not Everyone. Retired Story rating objects and Story `content_rating` fields are not interpreted.

2. How a rating is set

Creators set the content rating for their own stories and worlds when authoring them. Mystrix's automated systems and human moderation review (see Moderation, Takedown, and Appeals) can catch a rating that turns out to be wrong, but there is no independent, pre-publication ratings board reviewing every story before it is set live — the initial rating is the creator's own declaration.

3. How ratings affect what you see

Your account has a safety-filter preference, on by default, that limits story browsing and recommendations to content the filter judges appropriate; turning it off allows Mature and Adult-rated content to appear in what you are shown. This is a self-managed preference you control, not an age-verified gate — see Minors and Adult-Content Eligibility for exactly what that distinction means and does not mean.

4. Automated safety scanning

During classifier-backed processing of images and videos you upload, a reduced image derivative or, for video, one extracted poster frame is sent to DeepSeek for automated safety verification (Privacy Policy §4, Subprocessors and Data Retention §2). DeepSeek uses those pixels only to return safety scores; it does not generate or edit images. This upload-verification input is not sent to OpenAI or Google Gemini. A separate explicit Creator image request may send its bounded scene brief and creator-selected references to OpenAI or Google Gemini; the Creator-selected Z.ai image model receives the bounded text brief without generation references, and its independent quality check may inspect the selected references. The included house image path is separate from these Creator-funded connections. Generated candidates remain private and must pass the applicable quality, rights, and safety review before a finished frame can be applied. Reviewed house-owned assets retain their existing technical, rights, and review evidence. For uploaded images, safe verification approves, unsafe results reject, and unavailable verification stays pending for bounded automatic retry. Exhausted retries stay unavailable with their concrete failure; images never enter human moderation review. Generated candidates likewise need a passing independent quality verdict before application. Video moderation retains its authorized review workflow. Automated verification is a moderation aid rather than a guarantee that every problematic item is caught, and it does not replace the ability to report content you encounter.

5. Reporting content

If you encounter content you believe is unsafe, mis-rated, or otherwise violates our community standards, you can report it from within the product. See Moderation, Takedown, and Appeals for what happens to a report after you submit it.

6. Contact

Questions about a specific rating or the safety-filter setting should go through our support system.