E-Commerce & Retail

The return was approved. The receipt was generated.

AI receipt generators produce flawless fakes in seconds β€” itemized, taxed, and formatted like the real thing. TruthScan screens every receipt and invoice in your T&E, AP, returns, and claims workflows, catching AI-generated and manipulated submissions before the money goes out.

Seconds End-to-end verdict
2 APIs Image and PDF, purpose-built
Heatmaps Evidence for every decision
Return proof of purchase analyzed for AI generation

THE THREAT

Return fraud stopped being a shoplifting problem and became a rendering problem.

The whole returns economy runs on the assumption that documents and photos are expensive to fake and cheap to trust. Both halves of that assumption are now false. Return-without-purchase, wardrobing, warranty abuse and price-adjustment scams all reduce to the same act: generate a credible artifact, upload it, collect. And the marketplace side of the business has the mirror problem β€” listings built on product photography of items that do not exist.

In seconds

Verdict on a receipt or a photo β€” fast enough for the returns portal, not just the back office.

2 APIs

Image detection for return photos and listing imagery; PDF detection for receipts and order confirmations.

Both Sides

Buyer-side refund fraud and seller-side listing fraud, on one platform and one contract.

ATTACK PATTERNS

Where the artifacts turn into losses

Four patterns, split across the two sides of a retail platform: the buyer who wants a refund, and the seller who wants a listing.

Generated receipt used for a return without purchase

Return without purchase

The oldest retail fraud, newly industrialized. A generated receipt for an item never bought, presented for a cash or store-credit return. The document is not edited β€” it is authored from scratch, arithmetically correct, with a plausible store number and a plausible timestamp.

THE TELL YOU CANNOT SEE

Template reuse. The same generated receipt template reappears across unrelated accounts, which is invisible to a store associate and obvious to a detector.

AI-generated damaged unboxing photo for a refund claim

"Arrived damaged" β€” generated unboxing

Item arrives intact; claimant generates an image of it crushed, cracked or torn in the box and files for a replacement plus a refund. Because the product, the packaging and the branding are all correctly rendered, the photo reads as authentic to a reviewer working a queue on a clock.

PROMPT RECONSTRUCTED FROM A FLAGGED CLAIM

"make it look torn, in the original shipping box, phone photo, poor indoor lighting"

Generated receipt used for warranty or price-adjustment abuse

Warranty and price-adjustment abuse

A generated receipt backdated to place a purchase inside a warranty window, or to claim a price adjustment against a promotion that has closed. Small individual value, industrial volume, and a category most retailers do not review at all because the per-claim loss is below the cost of a human look.

WHY THIS ONE COMPOUNDS

It is under your review threshold by design. That is exactly why it scales.

Fabricated product listing imagery for marketplace fraud

Fabricated listing imagery

On the marketplace side, sellers generate product photography for inventory they do not hold β€” or generate imagery that misrepresents an item's condition, authenticity or provenance. The listing converts, the buyer is disappointed, and your platform absorbs the refund and the trust damage.

WHERE IT HURTS

Listing fraud is not a refund line. It is a retention line.

RECEIPT DETECTION MODULE

This is where receipt detection lives

Receipts are a capability, not an industry β€” but retail is their center of gravity. Return-without-purchase, warranty abuse and price adjustment are all receipt problems first. Run every proof of purchase through detection, including the ones below your manual review threshold, because the ones below the threshold are the ones being farmed.

See the Fake Receipt Detector

$16.8M

in fraudulent receipts caught by one enterprise customer in a single 30-day period, surfaced by batch-scanning receipts that had already cleared human review.

ONE ENTERPRISE CUSTOMER Β· 30-DAY PERIOD Β· FIGURE PENDING INTERNAL VERIFICATION BEFORE PUBLISH

THE ENGINES

Two detection engines. Two purpose-built APIs. One platform.

Photos and documents are different forensic problems, so we solve them with different models behind two dedicated APIs. You call the one that matches the artifact β€” and you review both in one dashboard, under one contract, through one vendor review.

IMAGE DETECTION API

Scores photos and listings

Return photos, condition evidence and seller-submitted product imagery β€” generated, in-painted or manipulated.

  • Generative and in-painting detection with region heatmaps
  • Capture-provenance signals: EXIF, device fingerprint, render artifacts
  • Batch scoring for listing imagery at catalogue scale
  • Same engine on both the buyer side and the seller side

PDF DETECTION API

Scores receipts and order documents

Proof of purchase, order confirmations, warranty documentation and price-match evidence.

  • Detects fully regenerated receipts, not just altered ones
  • Template-reuse matching across accounts β€” the highest-signal detection in retail
  • Structural and object-layer forensics beneath the OCR text
  • Batch mode for retroactive audit of refunds already issued
They are separate APIs, called separately. A single fraudulent submission usually carries both a photo and a document, so most teams call both and merge the verdicts in their own decision layer. End to end, the whole process takes one to two seconds.

WHERE IT SITS

In the returns portal, before the refund posts

Detection runs at submission, so a flagged artifact never reaches the stage where money moves.

STEP 01

Return initiated

Customer uploads a photo, a receipt, or both.

STEP 02

Artifacts scored

Image API on the photo, PDF API on the receipt. One to two seconds, end to end.

STEP 03

Your policy decides

Auto-approve, hold for review, or decline with evidence β€” your thresholds, not ours.

STEP 04

Listings screened

Seller imagery is scored on the same platform before the listing publishes.

DEFENSIBILITY

A refund decline that holds up in a chargeback

Declining a refund is the moment a customer becomes an adversary. Whatever you decline on has to be strong enough to survive the chargeback that follows.

Detailed AI analysis report for a refund decline decision
  • Note what passed. The arithmetic is correct and the OCR is clean β€” which is precisely why format validation and human review both approve these documents.

  • Template reuse is the killer signal. One suspicious receipt is an argument. The same generated template across four unrelated accounts is a ring.

  • Evidence attaches to the case. The forensic report and heatmap go into the chargeback representment, not just into a log.

  • Retroactive audit. Batch-scan the refunds you already issued. It costs a scan each and it is the fastest honest answer to "how bad is this?"

COMPLIANCE & DEPLOYMENT

Enterprise-grade by default

AICPA SOC certification seal

SOC 2 Type II

Audited controls

Independently audited security, availability, and confidentiality controls, renewed annually.

ISO 27001 certification seal

ISO 27001

Certified ISMS

Certified information security management across the full platform and API surface.

GDPR compliance seal

GDPR

Privacy-first processing

DPA available. Enterprise customers can pin processing to the EU.

Regional or on-site deployments seal

Global Infrastructure

Regional & On-site

UK, EU and on-premise deployments for Enterprise.

TECHNOLOGY FEATURED IN

Forbes logo
Business Insider logo
Nature logo
CBS News logo
Yahoo logo
BuzzFeed logo

Forensic Grade AI Fraud Prevention

Get full forensic reports - heat maps, key indicators, and detailed descriptions.

Free

$0/month

Try the full engine

  • 25 results/month (images + PDF pages)
  • Detailed indicators on every result
  • Full API access
  • Detection history in dashboard
  • Free forever β€” no trial period

Starter

$24/month

$0.03 / result - $290/yr

For individuals and small teams

  • 1,000 results/month included
  • $0.03 per additional result
  • CSV export of result history
  • Batch uploads
  • Audit-ready detection reports
  • Standard support
MOST POPULAR

Professional

$83/month

$0.02 / result - $990/yr

For teams in production

  • 5,000 results/month included
  • $0.02 per additional result
  • Priority processing queue
  • Higher API rate limits

Business

$333/month

$0.01 / result - $3,990/yr

For high-volume operations

  • 40,000 results/month included
  • $0.01 per additional result
  • Zero Data Retention (ZDR)
  • Highest self-serve rate limits
  • Priority support

Enterprise

Customize a Plan For Your Needs

$0.005 or less per result

Contact Sales
  • Discounts scale with volume
  • Custom SLAs with service credits
  • Zero Data Retention (ZDR)
  • Custom integrations
  • Custom MSA and DPA
  • Dedicated throughput
  • Named 24/7 account team
  • Dedicated / on-prem deployment

No hard quota walls β€” usage is simply metered at your plan's rate. Custom contracts, SLAs, DPAs, and hosting environments are available exclusively on Enterprise. All plans include SOC 2 Type II compliance.

FAQ

What returns and trust & safety teams ask us first

Batch-scan the refunds you already approved.

Send us a closed period of return receipts and photos. We will run both engines and show you what cleared review.

NO CARD REQUIRED Β· UNLIMITED SEATS ON EVERY PLAN Β· SOC 2 TYPE II Β· ISO 27001 Β· GDPR