We flag suspicious audience patterns by combining engagement scoring with audience audits. Signals include abnormal subscriber growth, low reach vs. subscriber count, repetitive commenter accounts, inflated ER vs. view ratios, and inconsistent post recency. Semantic analysis of comments and account clustering highlights botnets; anomalies are surfaced for rapid review.