Supported document type
Detect tampered and deepfake Receipt
A receipt is evidence that a purchase happened, at a price, on a date, and that is the whole of its value to an expense auditor or a claims handler. The fraud follows the evidence: a total nudged upwards, a date moved into a claimable period, or a slip for an item that was never bought at all. Generator sites now produce convincing ones from a merchant name and a figure. TamperCheck reads the pixels rather than the print.
API slug: receipt
Last updated:
What a genuine receipt contains
A till receipt is written by a point of sale terminal, and its fields arrive in a fixed order: merchant name and branch address at the head, then the store or terminal number, the transaction ID, and the date and time to the minute. The body lists each item with its price, followed by a subtotal, tax broken out by rate, and the amount tendered. The payment method closes it, usually a card type with the last four digits and an authorisation code.
The transaction ID is what ties the rest together. It is issued in sequence by one terminal on one day, so it carries an implicit relationship with the timestamp beside it and with the slips printed either side of it. Tax lines constrain the total the same way they do on an invoice, and the amount tendered has to reconcile with the change given. A total raised by twenty units leaves the tax figure describing a purchase that no longer exists.
How receipt forgeries are made
Total and tax edits
The amount is raised on a photograph of a genuine receipt, usually by a character or two, and the tax line is left as it was. The claim then describes a purchase whose tax no longer matches its rate, and the edited digits sit on a different noise floor from the ones printed beside them.
Date shifting
The purchase date is moved so a slip falls inside an open expense period or after the start of an insurance policy. It is the smallest possible edit and among the most valuable, because the document is otherwise entirely authentic and the merchant will confirm the sale took place.
Merchant substitution
The trading name and logo at the head are replaced so a personal purchase reads as a reimbursable one. The swapped logo is almost always a resampled web image, and it carries sharper edges and a different compression history than the thermal text underneath it.
Reprinting and screen capture
Thermal printing has a characteristic grain: uneven dot density, slight fade along the roll, and paper texture under the ink. A receipt edited on a screen and photographed back loses that grain and reads flat, which separates it from an original even when every field is correct.
Generator output
Receipt sites and image models now build a complete till slip from a merchant name, a date, and a figure. The result is internally consistent, so arithmetic and format checks pass, and it is identified instead by generation signatures and by the absence of any real print or capture history in the file.
Reuse across claims
One receipt is submitted to more than one employer, insurer, or period, sometimes with a single digit changed to disguise the reuse. The copies share the lighting, crop, and sensor noise of a single original photograph, and that fingerprint survives recompression and resizing.
What TamperCheck checks on a receipt
TamperCheck runs 200+ forensic checks across three layers and returns a risk score from 0 to 100 with plain-English findings tied to specific regions of the document. These three carry the most weight on this document class.
Arithmetic reconciliation
Stated totals are recomputed from their components, and running or cumulative figures are checked for continuity across periods. A single edited value breaks the chain even when the page still looks right.
Letterhead, logo, and issuer plausibility
Issuer branding is checked for resolution and compression consistency against the body text it sits on. A logo lifted from a website carries the signature of its source, not of the document.
Field-level forensics
Every text region is analysed in its own right rather than as part of one whole-page image. This catches the smoothed and feathered edits that whole-image Error Level Analysis misses entirely.
And many more checks
The three above are the layers that carry the most weight on a receipt. Every upload runs the full suite of 200+ checks regardless of document class, spanning file structure and metadata, pixel-level forensics, font and text rendering, optical and print characteristics, provenance signals, AI-generation signatures, and many more checks.
Who verifies a receipt, and why
Finance teams auditing expense claims, and insurers validating loss evidence. In each case the document is being used to unlock money, access, or a legal status, which is exactly what makes it worth forging.
How to verify a receipt in 4 steps
Check the fields against each other
Read total amount, purchase date, and merchant name together rather than one at a time. Forgers typically change one value and leave the rest of the document describing the original.
Inspect the file metadata
Open the document properties and look at the producer, creation date, and modification date. A document produced by a consumer PDF editor, or created long after the date printed on its face, is worth a closer look.
Ask for a second document
Request a corroborating document from the same issuer or an adjacent period. Forgery effort concentrates on one file, so inconsistencies surface as soon as there are two to compare.
Run a forensic check
Manual review catches obvious edits but not field-level pixel manipulation or synthetic generation. TamperCheck runs 200+ forensic checks on a receipt and returns a risk score from 0 to 100 within seconds, at $0.50 per document.
Frequently asked questions
How do companies detect fake receipts?
Companies detect fake receipts by comparing the claimed purchase against the physical evidence of how the slip was produced. Thermal print has a texture a screen capture does not, an edited total sits on a different noise floor from the digits beside it, and a swapped logo carries its own compression history. Automated forensic checks make this practical across a full month of expense claims.
Can AI generate a realistic fake receipt?
Yes, and convincingly enough that visual review is no longer a useful filter. Generator sites and image models produce till slips with plausible merchant details, sequential transaction IDs, and arithmetic that ties out. What they do not reproduce is the history of a real capture: the printer's grain, the camera's sensor noise, and the compression trail of a photograph taken in a shop.
How do I check if a receipt is real?
Read the transaction ID, the timestamp, and the tax breakdown against each other before you look at the total. Tax should be the stated percentage of the subtotal, the time should be plausible for the merchant's trading hours, and the payment method should match the card on the claim. Then run a forensic check, which covers print texture and per-field edit traces.
Which field is edited most often on a receipt?
The total amount, with the purchase date close behind it. Both are single-token changes to an otherwise genuine document, which is what makes them attractive to the claimant and hard to see for the reviewer. TamperCheck analyses receipts through one REST endpoint at $0.50 per document and names the field that moved rather than returning a score alone.
TamperCheck analyses receipt uploads with a hybrid forensic and AI pipeline tuned for this document class. Upload endpoints accept PDF and common image formats; class is inferred automatically. See the API & MCP documentation for authentication, async jobs, and webhooks.