Last updated: 2026-05-30
How ScamCheck collects scam reports, scores severity and trends, and decides what to publish.
We ingest public scam reports through a lightweight, rate-limited submission pipeline with spam and abuse filtering.
Personal data is redacted on intake; we never store raw contact details.
Reports are converted to embeddings and clustered by semantic similarity, so thousands of near-identical scams collapse into one canonical pattern.
Severity is an explainable 0–100 score (category risk, indicators, spread, velocity).
Trending uses a recency-weighted score; "viral" and "active" flags reflect momentum and the last-reported window.
Pages are auto-generated for genuinely trending patterns, deduplicated to one per pattern, and only published if they pass quality and budget gates.