AI made forging documents free. We made catching document tampering automatic.

Upload a payslip, bank statement, invoice or ID and get a clear verdict within seconds, with the altered field named. Edited documents and fully AI-generated fakes, both caught.

First top-up matched free · $0.50 per document · no sales call · or see a sample report

  • 100+ document types
  • 190+ countries
  • Zero document storage
Try it. Pick a documentsample document

Northgate Logistics Ltd

Employee: R. Alvarez · Period: 01-31 Mar 2025

seal
Basic pay$3,900.00
Overtime$420.00
Gross pay$4,320.00
Tax withheld−$890.00
Pension−$250.00
Net pay$6,480.00
Forensic checks0 / 200+
  • Gross, deductions and net reconcile·
  • Edited pixels around pay figures·
  • Employer template on record·
  • Tax and pension rates plausible·

Synthetic samples. Real results vary by document.

100%
of findings auto-QC'd in seconds
< 1 min
to your evidence-backed findings
100+
document types, 190+ countries
$0.50
per document, no seat fees

The stakes

One in five forged documents clears manual review

Your reviewers are good, but a convincing fake now takes minutes to make and seconds to send. Every document you accept on trust is a decision someone else made for you.

Reviewing a document by hand costs about 15 minutes of someone's time. Checking it properly costs 50 cents.

1k
1002505001k2.5k5k10k25k100k250k500k
+1,775% vs TamperCheck.ai

Manual review

$15k

/ month

+150% vs TamperCheck.ai

Typical vendor

$2k

/ month, est.

With TamperCheck.ai

$800

/ month

You save vs manual

$170.4k

/ year

The product

Drop a document in. Get a verdict.

Upload a payslip, statement, invoice or ID and pick a strategy - tampering detection, deepfake detection, or extraction. Within seconds the finding comes back, with a risk score and the evidence behind it.

Going live with us takes minutes - the dashboard is live the moment you sign up.

tampercheck.ai / new check

1 - Choose a strategy

2 - Drop the document

Drop a document or click to browse

PDF, JPG or PNG - up to 25 MB

3 - Verdict

payslip_march.pdf

tampered
Risk score
87

Altered field: Net pay

Net pay reads $6,480 - gross minus deductions is $3,180.

Completed in 38 s - charged $0.50

The workspace

Every document you've checked, in one queue

Verdicts, risk scores and team notes for every upload - searchable, filterable, and ready to share. We provide a web dashboard your whole team can use from day one.

tampercheck.ai / documents

payslip_march.pdf

87
tamperedPayslip2

receipts_scan.pdf

processingReceipts

bank_statement_feb.pdf

74
tamperedBank statement1

drivers_license_scan.jpg

65
tamperedDriver's license

degree_certificate.pdf

4
authenticDegree certificate

invoice_acme_corp.pdf

8
authenticInvoice
March - 847 documents checkedAverage turnaround 41 s

Every plan includes the full dashboard.All 100+ document types

The findings

Findings your team can defend

Every verdict ships with the check that failed, the evidence behind it, and a plain-English summary. 200+ checks run at once, then a separate QC pass re-validates each finding before it reaches you.

A vendor claim is easy to wave away. A failed check with the pixel diff, the font metrics and the arithmetic gap attached is not - your reviewers can forward it as-is.

bank_statement_feb.pdf - details

bank_statement_feb.pdf

Bank statement - Meridian Bank

tampered
74

Plain-English summary

The running balance stops adding up from the salary credit onward - it is off by $4,320. The salary figure itself carries pixel and font-spacing edits; the rest of the statement matches the issuer's template.

arithmetic integrity97% conf

Running balance is off by $4,320 from the salary credit row onward. The stated salary does not reconcile with the itemised transactions.

3 balance discontinuitiesrows 14-18
text manipulation92% conf

Compression artefacts and digit spacing around the salary credit are inconsistent with the rest of the row.

ELA diff 0.31 - salary credit amountrow 15
font inconsistency71% conf

Character metrics on the salary row differ from the document's embedded font set.

baseline shift 0.8 pt vs document medianrow 15

+ 3 more findings in the full report

SHA-256 9f2c41a7…e081Uploaded Mar 12, 9:58 AM - analysed in 38 sProcessed in memory - 0 bytes retained

The report

A report you can attach to the case

One click exports the full forensic analysis - verdict, evidence and proofs - in a simple format a reviewer, auditor or underwriter can read.

The verdict, evidence, chain of custody, and the file hash travel with every report.

TamperCheck.aiReport TC-2026-0312-4471

Document verification report

bank_statement_feb.pdf - Bank statement - Meridian Bank

tampered
Risk score74/100

Altered field: Salary credit

The stated salary is $4,320 above the reconciled total; pixel edits confirm the row was rewritten after export.

SHA-256 9f2c41a7…e081 - Generated Mar 12, 9:59 AM - Analysed in memory - the document was never stored - Shared report link - no login required

What it catches

The kinds of document fraud we catch

Six of the most common ways documents get faked, and what we hand back on each one.

And many more. 200+ checks, updated as new forgery methods appear.

Figures that were changed

payslip_march.pdf · Payslip · Northgate Logistics · synthetic sample

Tampered

Northgate Logistics Ltd

Employee: R. Alvarez · Period: 01-31 Mar 2025

seal
Gross pay$4,320.00
Tax withheld−$890.00
Pension−$250.00
Net pay$6,480.00

Net pay should read $3,180. Nothing on the page looks wrong to the eye.

What you get back

Net pay was edited upward by $3,300.

  • The altered field is named: net pay.
  • Gross pay minus deductions comes to $3,180, so the payslip contradicts itself.
  • Everything else on the page checks out, so this is a genuine payslip that was edited.

AI-generated documents

Catching AI-generated documents that were never real

Editing a genuine payslip is the old way. Today a model will produce a whole one: correct employer, plausible numbers, clean layout, no original to compare against. Checks built to spot edits find nothing wrong, because nothing was edited.

We look for what generated documents cannot fake: the physical traces of a real scan, the quirks of real issuer templates, and the statistical fingerprints image models leave behind. Every upload is checked for synthetic origin alongside every tampering check.

How we detect AI-generated documents
How we catch it1 / 3

Synthetic origin detection

Image models leave statistical fingerprints. We test every upload for them, whatever the document type.

Comparison

How it compares to what you do today

Same documents, three ways of checking them.

  • Documents checked

    TamperCheck.ai
    100% of uploads
    Manual review
    A sample onlyLimited
    Other solutions
    Usually IDs only
  • Forensic checks per document

    TamperCheck.ai
    200+ checks
    Manual review
    Visual inspectionLimited
    Other solutions
    Basic analysisPartial
  • Surface-level detection

    TamperCheck.ai
    Yes, plus what's beneath
    Manual review
    EyeballingLimited
    Other solutions
    YesPartial
  • Image Forgery Detection

    TamperCheck.ai
    Pixel-level, every photo
    Manual review
    EyeballingLimited
    Other solutions
    LimitedLimited
  • Document Forgery Detection

    TamperCheck.ai
    Template, font and layout forensics
    Manual review
    EyeballingLimited
    Other solutions
    BasicLimited
  • AI Generation Detection

    TamperCheck.ai
    100% of checks
    Manual review
    NoLimited
    Other solutions
    RarelyLimited
  • Metadata checks

    TamperCheck.ai
    Full metadata analysis
    Manual review
    NoLimited
    Other solutions
    Basic onlyPartial
  • PDF Analysis

    TamperCheck.ai
    Yes
    Manual review
    NoLimited
    Other solutions
    RarelyLimited
  • Digital signature certificates

    TamperCheck.ai
    Issuer and trust chain verified
    Manual review
    NoLimited
    Other solutions
    RarelyLimited
  • Field-level alteration detection

    TamperCheck.ai
    Yes, fields named
    Manual review
    Only if spottedPartial
    Other solutions
    NoLimited
  • Multilingual Support

    TamperCheck.ai
    Yes, any language
    Manual review
    Depends on reviewerPartial
    Other solutions
    Usually English-onlyPartial
  • Automated QC on findings

    TamperCheck.ai
    Every finding checked first, with evidence
    Manual review
    Second reviewer, if any
    Other solutions
    Human review, little evidencePartial
  • Time per document

    TamperCheck.ai
    <60 sec, unattended
    Manual review
    ~15 min
    Other solutions
    Minutes + follow-upPartial
  • Time to verdict

    TamperCheck.ai
    Real-time
    Manual review
    Hours to days
    Other solutions
    Same dayPartial
  • Cost per document

    TamperCheck.ai
    $0.50, no seat fees
    Manual review
    $15
    Other solutions
    Seat fees + per-checkPartial
  • Document security

    TamperCheck.ai
    0 bytes retained
    Manual review
    Stored in case filePartial
    Other solutions
    Retained for trainingPartial
  • AI Training

    TamperCheck.ai
    No
    Manual review
    No AI usedPartial
    Other solutions
    Often permittedPartial
Manual review figures use the same assumptions as the calculator above: 15 minutes of analyst time per document, and the industry estimate that roughly one in five fraudulent documents passes visual inspection. Vendor comparisons are generic here. Named, sourced comparisons live on our comparison pages.

Pricing that fits how you already buy

Pay only for what you check - no contracts, no surprises.

See full pricing
$0.50
per document check, up to 6 pages
100%
of your first top-up matched in free credits, up to $200
$0
seat fees, volume tiers, or minimum commitment

API & MCP

An API and an MCP for any workflow

POST a document, get back a verdict, a risk score and per-finding evidence your team can audit. Agents can connect over MCP with the same API key and billing.

  • Webhooks for async completion, shareable report links for reviewers
  • Works with any MCP client - Claude Desktop, Claude.ai connectors, LiteLLM, and others
  • Usage and wallet endpoints for billing your own customers
Read the API docs
curl
curl -sS -X POST "https://api.tampercheck.ai/api/v1/documents/" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "document=@./statement.pdf" \
  -F "strategy=tampering_detection" \
  -F "document_identifier=loan-app-12345"

Verdict response

{
  "status": "completed",
  "verdict": "tampered",
  "risk_score": 42,
  "confidence": 0.78,
  "findings": [
    {
      "category": "font_inconsistency",
      "severity": "medium",
      "description": "Mixed font metrics in the totals row."
    }
  ],
  "human_summary": "Plain-language summary of the main issues."
}

Trust

Your documents are never stored

Most verification services retain documents for future QA and model training. We don't. Not one byte, not one second.

Zero retention

Analysed in memory, then discarded. Never written to disk.

Never trained on

Your documents are never used to improve our models.

GDPR-aligned controls

Audit-ready logging, least-privilege access, TLS 1.3 throughout.

99.9% uptime SLA

Backed by a public status page you can check any time.

Security & compliance

  • SSO / SAML and role-based access controls
  • Your choice of data residency
  • Dedicated enterprise instance
  • Forensic evidence in every report
  • EU AI Act-aligned workflows

FAQ

Document tampering detection, explained

What we check, what it costs, and what happens to your documents.

What is TamperCheck.ai?

TamperCheck.ai checks documents for tampering and AI-generated fakes. Upload a document - a bank statement, payslip, invoice, certificate, contract, or any of 100+ supported types - and you'll get back a clear AUTHENTIC or TAMPERED result, with every issue explained in plain English. It takes seconds.

How does TamperCheck detect document tampering?

TamperCheck runs 200+ automated checks on every document at once. It looks for edited numbers, mismatched fonts, unusual metadata, inconsistent templates, altered seals and signatures, digital-signature certificates that fail trust checks, and signs of AI-generated content. If something looks wrong, you get a plain-English explanation of what it found and exactly where.

How much does document fraud detection cost?

Checking a document costs $0.50 (a 7-12 page document counts as two checks), plus a small charge for the AI reasoning involved. Your first top-up is matched with free credits, doubling what you start with to spend. No seat fees, no volume tiers, no minimum commitment.

Does TamperCheck store the documents I submit?

No. TamperCheck processes each document in memory and discards it immediately after the check. Documents are never stored, archived, or used to train our models.

What document types does TamperCheck support?

TamperCheck supports 100+ document types, including bank statement, payslip, invoice, insurance claim, degree certificate, employment contract, trade license, medical report, and more, issued from 190+ countries. See the full list on our document types page.

Does TamperCheck detect deepfakes?

Yes. Alongside checks for edited numbers and forged stamps, TamperCheck runs dedicated checks for AI-generated and deepfake documents. That catches fully fabricated IDs, bank statements, and payslips - documents that were never real to begin with - not just genuine documents that were altered.

What makes TamperCheck different from other tools?

Most verification tools return a document-level risk score. TamperCheck resolves every finding to the individual field - the exact figure or text that was altered - and shows the evidence behind it. Checks are pay-as-you-go at $0.50 per document with no seat fees, documents are never stored, and every finding is re-validated before it reaches you.

Check your first document today

Upload a suspicious document and get a full forensic verdict within seconds. Validate your card and try today - no sales call, no commitments.