How Our Data Works

Our sources, how we classify reports, when we publish a page, and the limits of what the data can tell you.

1. Sources

The live registry of feeds is on the Sources page.

2. Classification

Each report is processed by an AI classifier that assigns a scam type (one of 15: phishing, romance scam, employment scam, crypto scam, investment fraud, bank impersonation, government impersonation, AI/deepfake, pig butchering, QR phishing, rental scam, advance fee, crypto recovery, tech support, grandparent scam), the victim's country when it is stated, the platform where contact happened, and a short neutral summary of the script.

Reports where the type cannot be determined with confidence stay unclassified and are excluded from statistics. As of September 2026, 8,300+ reports carry a confident type; the rest are stored but not counted.

3. Publication thresholds

4. Known limitations

5. Update cadence

News: daily. Weekly digest: Mondays. Guides and statistics: rebuilt daily from the database. Each page's lastmod in our sitemap changes only when its content actually changes.

Last updated 2026-10-03