The 7 Signs of a Tampered Bank Statement (and How Automated Forensics Catches Each One)

Every bank statement row follows a simple rule: previous balance ± transaction amount = new balance. When a fraudster inserts a fake credit or alters a transaction amount, the running balance sequence breaks.

Automated analysis checks every row against this arithmetic. A single inconsistency — even one obscured by rounding — flags the statement immediately. Human reviewers rarely check all rows; automated systems check all of them in milliseconds.

When a new row is inserted into a statement (to add a salary credit, for example), it rarely matches the surrounding rows perfectly. Forensic analysis detects:

• Slight column misalignment compared to original rows
• Font metrics that differ by fractions of a point from surrounding text
• Compression artefacts that differ from the document's baseline noise floor

These signals are sub-pixel and invisible to the naked eye, but statistically reliable indicators of alteration.

Changing "1,200" to "12,000" is a single digit change that dramatically inflates apparent income. Forensic checks for amount alterations include:

• Character-level font consistency within a cell (pasted digits often use a slightly different typeface)
• Ink colour and opacity analysis (digital edits often differ from surrounding text in subtle colour space measurements)
• PDF text layer vs visual layer mismatch (in PDFs, the selectable text layer and the rendered visual sometimes diverge after editing)

Bank statements generated by online banking platforms embed metadata: creation software, timestamp, and sometimes a digital signature. When a statement is edited and re-exported or printed-to-PDF, this provenance breaks:

• Creation tool changes from the bank's PDF generator to a generic editor
• Modification timestamps post-date the statement period
• Digital signatures, where present, become invalid

Fraudsters reproducing a bank's statement template often get the branding slightly wrong — logo colours shifted, header fonts substituted, address formatting inconsistent with the institution's actual templates. AI agents trained on banking document patterns flag these institutional inconsistencies even when a human reviewer unfamiliar with the specific bank would not notice.

A common shortcut in fabricated statements is duplicating rows with slight date changes. Statistical analysis of transaction patterns — amounts, frequencies, merchant categories — can identify implausible repetition that real spending would not produce.

Authentic bank statement PDFs produced by banking systems have their text layer generated simultaneously with their visual layer. When someone edits the visual layer (changing a number) but doesn't update the text layer, or vice versa, a forensic check catches the divergence. This is one of the most reliable signals available and is entirely invisible without automated analysis.