Fake Invoice Checker

Vendor invoice fraud detection covers the highest-loss invoice attacks: silent bank-detail swaps on legitimate vendor invoices, vendor impersonation with spoofed logos and templates, and recycled or round-tripped invoices submitted across periods.

TrueDoc compares structural fingerprints across submissions in your queue, flags metadata that disagrees with claimed vendor and date, and surfaces inconsistencies between the rendered invoice and the underlying PDF objects.

AI-generated invoice detection evaluates layout regularities, font fingerprints, and generative-AI signatures common to invoices produced by current image and PDF generators.

Fully synthetic invoices from non-existent vendors are now one of the most common ways busy AP queues are attacked. TrueDoc's AI-generated invoice detection catches them alongside hand-edited tampering in a single trust score.

Wire-fraud variants where bank details on a legitimate vendor invoice are silently swapped.

Fully synthetic invoices from non-existent vendors targeting busy AP queues.

Round-tripping where the same invoice is resubmitted with minor edits across months.

AI-generated invoices that perfectly match a known template but reference no real PO.

Math that doesn't reconcile. Line items, quantities, unit prices, subtotal, tax, and total should add up exactly. Even a $0.02 mismatch on a clean line-item invoice is a red flag for hand-editing.

Last-minute bank-detail changes. A request to update bank details on a known vendor's invoice — especially via email and especially under time pressure — is the single highest-risk signal in AP fraud. Verify via a known, out-of-band channel before paying.

Vendor identity drift. Subtle changes in legal name, address, tax ID, or domain (e.g. acme-payments.com vs acme.com) often indicate vendor impersonation.

Invoice number anomalies. Sequence breaks vs the vendor's prior invoices, duplicated numbers across periods, or formats that don't match the vendor's normal scheme.

Missing or mismatched PO reference. An invoice with no PO when the vendor normally issues against POs — or a PO number that doesn't match any open order — should never auto-approve.

Font, alignment, and metadata inconsistencies. Mixed fonts, misaligned columns, or PDF metadata (Producer, ModDate, incremental updates) that disagrees with the claimed vendor and date.

AI-generation signatures. Fully synthetic invoices from non-existent vendors are increasingly common — TrueDoc's AI-generated invoice detection catches them alongside hand-edited tampering in a single trust score.