safety architecture

Why AI apps should not hard-block every flagged moderation result

AI apps should treat moderation flags as signals, not automatic shutdowns, because hard-blocking every flag overblocks legitimate content.