Data Report · 8 min read
Document Fraud Statistics 2026
Live data from the Turing Verify Wall of Forgeries · Updated April 2026
2.4M+
Documents analyzed in Q1 2026
86,400
Confirmed forgeries flagged
3.6%
Average fraud rate across pools
47%
AI-generated forgeries (share)
+241%
YoY growth in diploma-mill credentials
22 seconds
Median detection time
These numbers are aggregated from the Turing Verify detection pipeline over Q1 2026. The live, always-current version of the dashboard is published at our Wall of Forgeries. We make our stats publicly quotable so researchers, journalists, and policy teams can cite real numbers rather than guesses.
Most-forged document types
Diplomas remain the single most-forged document type, accounting for roughly 41% of all forgeries we catch. Academic transcripts are second at 22%, employment letters third at 14%, professional certificates fourth at 11%, and the remaining 12% is split across passports, driver's licenses, and miscellaneous credentials. The diploma share has been stable for three years; the share of employment letters is growing the fastest, because remote-first hiring has made salary verification a more common friction point.
Highest-risk issuing countries
Fraud rates vary dramatically by issuer country. The countries with the highest measured fraud rates in our pipeline in Q1 2026 were, in order: Nigeria, India, Pakistan, the United States (a surprise that reflects the enormous absolute volume of US credentials checked), the Philippines, and Ukraine. Note that these are detection rates — they reflect what our engine catches, not a moral judgment of the countries themselves. A high detection rate often correlates with a large population of institutions and a fast-growing remote-hiring economy, not with institutional integrity.
Rising fraud techniques
The single fastest-growing fraud technique in 2026 is AI-generated template substitution — using a diffusion model to regenerate a real school's template with fabricated student data. This technique now accounts for nearly half of all forgeries we flag, up from 18% a year ago. Second is metadata stripping (removing PDF producer information to defeat naive scanners). Third is registry-ID fabrication — inventing a plausible but invalid credential number in jurisdictions where registries are not publicly queryable.
Estimated cost of document fraud
Credible estimates put the annual global cost of credential and document fraud between USD 600 billion and USD 1 trillion, including wage misallocation, insurance fraud, and identity crime downstream of a fake document. We don't publish our own estimate because the number is genuinely uncertain — but even the low end of that range makes document fraud one of the largest categories of economic crime on Earth.
Methodology
Every number in this report comes from the Turing Verify production detection pipeline. Confirmed forgeries are documents that failed at least three independent forensic signals and, where possible, were cross-referenced against a registry. We deliberately exclude low-confidence flags to keep the baseline conservative. The live dashboard updates every few minutes; this article is refreshed monthly.
If you're a journalist or researcher and want to quote these numbers, please do — and link back to verify.turingcerts.com/wall. For the full press kit with quotable stats and logo assets, see our /press page.
Check a document against the engine
Free, no signup required. Same engine that produces these numbers.
Run a check →