Sign inStart creating

How moderation decides

Automated pre-screen, then human review with a two-reviewer rule on borderline cases and a full audit log.

Every decision is recorded with its reviewer, reason code and rationale, in an immutable audit log.

Automatic blocks are narrow and specific: CSAM (which also triggers a documented NCMEC reporting procedure), unverified real-person likeness, and content matching a rights fingerprint.

Strikes accumulate against repeat violations and can lead to suspension. A strike issued by a case that is later overturned on appeal is cleared automatically.

Viewer reports feed a re-review queue rather than acting directly — a report is a signal, not a verdict.

An honest note on the classifiers: where a safety provider is not configured, those checks report that they did not run. They are never reported as passed. The CSAM gate in particular throws rather than returning a clean result, which correctly blocks public launch until a provider is contracted.

Did this answer your question?

Related

All help articlesAsk the community