Use casesPayslip and document fraud
USE CASE · LENDING

Payslip fraud: catching the document that only has to be plausible.

Someone opens a real payslip in a free PDF editor, changes one number, and saves. Nothing looks wrong, because nothing is wrong with how it looks.

Use case

Catching the document that only has to be plausible.

The problem

Income verification in a lending file usually rests on an uploaded payslip. A payslip does not have to be forged. It only has to be plausible. A classical document check that only asks whether a document looks right passes a document like this without noticing anything, because nothing about how it looks was touched.

The application

Aletheia reads the parts of a document that a forger does not think to fix, asking whether the document makes sense rather than whether it looks right: whether the arithmetic holds, whether the claimed salary matches money that actually arrived, whether the pay date matches the employer's cycle. Three layers work independently and are correlated: file and pixel forensics, content logic and plausibility, and behavioural consistency against real account data. In a proof of concept, Aletheia recognised every case the bank had found and surfaced additional suspicious cases the existing process had missed.

Why it matters beyond lending

A manipulated document is real and then edited, so forensics catches it. An original fake was never real: it was generated cleanly in one pass and leaves forensics nothing to find. A tool with one layer catches one of these and calls the other clean. The same three layers apply wherever a document is asked to support a decision, not only a payslip in a lending file.