Skip to main content
Tool · Credit · Fraud triage

Statement Authenticity Checker

Drop a bank-statement PDF. The tool reads its metadata and tells you whether it looks consistent with a bank-generated document, or whether it has been through a consumer PDF editor.

Your file never leaves your browser. The PDF is parsed locally in JavaScript. Nothing is uploaded to Terra Insight or any third party.

What this checks

PDF metadata

Title, Author, Subject, Creator, Producer, CreationDate, and ModDate fields in the PDF trailer — parsed directly in your browser.

Consumer-editor signatures

Flags Producer strings from common PDF editors (iLovePDF, Smallpdf, PDF24, Sejda, Foxit, Nitro, Adobe Acrobat, Word, Chrome save-as-PDF, and similar).

CreationDate vs ModDate drift

If a document was modified after it was created, the two timestamps diverge. That alone is not proof of fraud, but it is a signal worth a second look.

Known bank-generation signatures

Finacle, Flexcube, BaNCS, T24, Intellect, Oracle BI Publisher, Crystal Reports, iText, Adobe LiveCycle, Acrobat Distiller — Producer strings that match these are the expected shape of a bank-rendered statement.

What this does not check

Balance-chain integrity

Whether every running balance in the statement follows from the prior balance plus the transaction. A tampered statement can have clean metadata and a broken balance chain.

Digit-distribution forensics

Whether the distribution of digits across transaction amounts conforms to expected patterns. TransactIQ runs this at depth; this checker does not.

Holiday-calendar sanity

Transactions posted on settlement holidays, unusual cut-off behaviour, or value-date anomalies specific to Indian bank processing calendars.

Round-trip and circular-flow detection

Inflows that immediately exit to the same counterparty, shell-company patterns, or layering signals that only a full transaction-graph analysis can surface.

OCR-layer tampering

Scanned statements where the visible characters disagree with the underlying text layer. A metadata-only check cannot see this.

A fraud verdict

This tool produces a signal, not a verdict. "Edited" metadata can be benign (a treasury team adding a watermark). "Clean" metadata can hide sophisticated tampering.

For a full forensic report across all 10 tampering and AML signal families, see TransactIQ →

Common questions

Is my PDF uploaded anywhere?

No. The tool runs entirely in your browser. The file is read as a local ArrayBuffer, parsed with JavaScript on your device, and never transmitted to Terra Insight or any third party. You can verify this in your browser's developer tools — the Network tab will show zero outbound requests when you run the check.

What does an "Edited" verdict actually mean?

It means the PDF metadata shows evidence of modification after creation — either the ModDate is materially later than the CreationDate, or the Producer string matches a known consumer PDF editor. Editing is not the same as fraud. A bank may re-watermark a statement before emailing it. A borrower may compress the file. The signal tells you to look closer, not to reject the application.

Will this catch all statement fraud?

No — and anyone claiming otherwise is misleading you. This is a metadata-only check. It catches the easy cases: statements re-saved through editors, statements generated from Word or Chrome, statements with inconsistent timestamps. It does not catch sophisticated tampering that reconstructs metadata, nor tampering that leaves the PDF wrapper intact while editing the underlying text. For that you need a multi-signal forensic pipeline.

Which banks are supported?

All of them. The tool reads PDF metadata, which is a standard field set independent of the issuing bank. It recognises the Producer strings used by the major statement-generation stacks in use across Indian public-sector banks, private banks, cooperatives, and NBFCs. If the Producer is unrecognised the verdict is Unknown, not Edited — absence of a known signature is not evidence of tampering.

Is this check enough for credit decisioning?

No. This is a triage tool. For underwriting decisions at any meaningful loan volume, you need balance-chain verification, digit-distribution analysis, OCR-layer consistency checks, holiday-calendar flags, round-trip detection, and counterparty-graph analysis — across the full population of statements, not one at a time. That is what TransactIQ is for.

Run a full forensic analysis with TransactIQ

Metadata is one signal of many. TransactIQ runs the full pipeline — balance-chain, digit-distribution, holiday-calendar, round-trip, OCR-layer consistency, and counterparty-graph analysis across every statement in your underwriting queue.

Talk to the team