Fraud Detection
If they uploaded a document, we read every byte.
A forged pay stub is designed to fool your eyes. ProofUp reads the raw PDF data instead, where every forgery leaves a trace. Patterns discovered by AI become permanent rules, so every customer benefits from every catch across the network.
Signals
Every document gets every check.
For applicants who upload instead of connecting directly, ProofUp runs all of these checks on every file.
Hidden text layers
A PDF can show one number to your eyes and store a different one in its data. We compare both. Any mismatch flags the document.
Annotation artifacts
Payroll software generates a document once and doesn't touch it again. Post-creation edits mean something changed after the fact.
Known generator sites
We maintain a running list of fake pay stub generator sites. If a document came from one, it flags on intake.
Metadata fingerprinting
ADP, Gusto, Wells Fargo: every real payroll and banking system leaves a specific fingerprint in the document. A consumer PDF editor leaves a different one.
Document misclassification
Applicants sometimes submit an offer letter where a pay stub is required, or a year-end summary instead of a current statement. We catch the substitution automatically.
Year-to-date math
YTD earnings only go up. When they don't, or when the numbers don't add up across pay periods, it flags automatically.
Browser-generated PDFs
Real payroll systems generate structured PDFs with consistent internal signatures. A "print to PDF" from a browser looks completely different at the data level.
AI → Rules
New forgery patterns become permanent rules
When our AI identifies a new forgery pattern, our team converts it into deterministic detection logic. Your portfolio benefits from every catch made anywhere in the network.
See Real Catches Live