We cross-check creator content and audience signals across the last 30 days - semantic analysis of video text, hashtag/keyphrase matches, engagement patterns, subscriber growth and view distributions. Fraud flags include abnormal subscriber spikes, low reach vs. subscriber size, bot-like engagement, and audience-market mismatches; results produce an engagement score and audience-quality diagnostics.