Factlabel is a verification system that audits AI-generated content against source data to detect fabricated figures, false attribution, and misleading framing. It uses a three-tier approach combining deterministic code checks and Claude-powered analysis to flag problematic claims, enforce publication gates, and optionally rewrite text until it passes verification.