How moderation works

AI-first review with human judgment where it counts.

Every listing, edit, review, and ad is scanned the moment it's submitted. Clean content publishes in seconds; only uncertain or high-risk cases reach a human.

Confidence routing

Each decision carries a confidence score:

≥ 0.90 — auto-approve or auto-reject.
0.70–0.89 — a softer action: publish with a warning, force an NSFW tag, or request changes.
< 0.70 — escalate to a human moderator with full evidence.

Child-safety and credible-threat categories always go to a human, regardless of confidence.

What we check

Scams, phishing, malware and token grabbers, unmarked adult content, hate and harassment, impersonation, redirect-only / no-content servers, keyword stuffing, misleading tags, and bump/review manipulation.

Actions

Approve · approve with warning · force NSFW · request changes · shadow-limit · reject · suspend · escalate.

Explainable & appealable

Every decision is logged with the text it saw, the policy it cited, its confidence, and the final action. Owners can appeal from their dashboard — an AI summary is prepared, then a human reviews the full context.