Insurance · Where the file can be faked

First notice of loss is where the file can be faked.

  • Photos
  • Estimates
  • Invoices
See it on your files
A property-damage photograph clipped to a repair estimate in a claims file, with a flagged line.
WARNrisk 76 · estimate total altered

01Built around your team

Citadel flags what to review. Your adjusters keep the judgment.

  • Photos
  • Estimates
  • Invoices

Clean files stay fast. Suspicious ones go to SIU with the evidence, before the claim pays.

02The synthetic claim

The defining fraud is a whole file built to agree with itself.

Photo, estimate, and intake, on one claim

One claim · reconciled across the fileWARN

Damage photo

Loss shown, vehicle and angle

Repair estimate

Line items, parts, labor, total

Intake packet

Date, location, narrative of loss

Contradiction found. The estimate total does not match the damage in the photo. Routed to SIU before the payout.

Exhibit A

Reconciliation across photo, estimate, and intake, on a single claim file.

03Where the file can be changed

Photo, estimate, invoice, packet. Each one can be faked.

WARN
Damage photo

Damage added or erased

Paid on a loss that never happened

BLOCK
Repair estimate

Totals and line items altered

A bigger payout than the damage supports

BLOCK
Medical / vendor invoice

Fabricated charges

Padded spend on every file

WARN
Intake packet

Hidden instructions planted in the file

Automation steered to pay without review

04How it works

Screen the file at intake. A verdict before the claim pays.

01Receive

Any document, photo, or page of text. One scan before your workflow acts.

02Verify
  • Document math
  • Pixel & layout tampering
  • Cross-document consistency
  • Hidden instructions

Citadel inspects the pixels, the layout, and the text.

03Route
ALLOWcontinue the workflow
REVIEWpause and follow the next step
WARNreview suspicious evidence
BLOCKstop the workflow

A clear next step before funding or payout.

05The stakes

About one in ten property claims is fraudulent.

of property claims are fraudulent
~10%

Deloitte 2025

yearly property-claim fraud losses
$122B

Deloitte 2025

human accuracy on AI-generated images
~50%

56-study meta-analysis

YoY rise in digital document forgery
244%

Entrust 2025

Sources: Deloitte Center for Financial Services 2025 · Entrust Identity Fraud Report 2025 · Diel et al., deepfake detection meta-analysis, 2024.

On your files

See what your intake is already missing.

Bring a sample from live intake. We'll walk the edits and fakes your reviewers and automation miss, before they reach a decision.

See it on your filesHow a team starts