Bank statement fraud detection that catches what reviewers can't
Edited balances, inserted salary credits, and fully AI-generated statements are the backbone of income and affordability fraud. TamperCheck runs 130+ forensic checks on every statement - digital PDF, scan, or phone photo - and returns a plain-English verdict in about a minute.
The problem
Why bank statement fraud gets through
A fraudulent applicant rarely fabricates a whole statement. They take a genuine one and change three numbers: the opening balance, a deposit or two, and the closing balance. Done carefully in a free PDF editor, the result is 95% authentic - which is exactly why visual review fails.
The template-matching tools many teams rely on fail for the opposite reason: a tampered statement still matches the bank's template perfectly, because it started life as a real statement from that bank.
And the newest wave skips editing entirely: AI-generated statements built from a text prompt, with plausible transaction histories and correct branding, that have never touched a bank's systems at all.
What we detect
Forensic signals that expose fraud
Running balance arithmetic
Every transaction row's running balance is recomputed against the previous row. A single edited amount breaks the chain - even when the edit is visually perfect.
Font metrics and substitution
Edited figures are usually re-typed in a near-match font. TamperCheck compares character-level font metrics across the document and flags fields that don't match their neighbours.
Editing tool fingerprints
Banks produce statements with a small set of known PDF generators. Traces of Acrobat, LibreOffice, or online PDF editors in the file's metadata and object structure are a strong tampering signal.
Transaction row insertion artifacts
Inserted or deleted rows leave structural residue - inconsistent line spacing, misaligned columns, orphaned formatting objects - that forensic parsing exposes.
Cross-document consistency
When statements are submitted alongside payslips or employment letters, salary credits, employer names, and dates are checked against each other. Fabricated bundles rarely stay consistent.
AI-generation signatures
Fully synthetic statements carry noise patterns, compression artifacts, and layout regularities that differ from genuine bank output. TamperCheck's computer-vision layer detects these signatures.
The solution
How TamperCheck detects bank statement fraud
TamperCheck treats every statement as a forensic object, not a set of numbers to be read. Structural analysis, computer vision, and AI adjudication run in parallel, and the verdict comes back with specific findings tied to specific regions of the document.
How it works
- Upload via API or dashboard - single endpoint, single file, verdict in about a minute
- 130+ forensic checks run automatically on every statement
- Works on any bank - no template library to maintain, institutions covered globally
- Plain-English findings your credit or ops team can act on directly
- $0.50 per document - no subscriptions, no minimums
- Zero document storage - files are processed ephemerally and never retained
FAQ
Common questions
See it working on your documents
Start with $5 in free credits - no contract, no card required. Upload your first document and get a verdict in about a minute.