Cryptographic Proof-of-Origin: Detecting Invoice Fraud & PDF Tampering with HMAC-SHA256 Visual Stream Fingerprinting

PDF invoice fraud — where a bad actor intercepts a legitimate PDF invoice and edits the IBAN bank account number or total amount in a PDF editor before forwarding it to accounting — costs enterprises billions of dollars annually. Traditional PKI digital signatures require expensive hardware tokens (HSMs), complex certificate authorities, and often break when processed by automated web pipelines.

How Tamper Detection Works Under the Hood

1. Deterministic Content Stream Hashing: Tailprint iterates over every page in the PDF document, decompresses the raw postscript drawing operators (FlateDecode streams), and hashes them alongside the page geometry and annotation count.
2. Canonical Manifest Signing: The doc_id, issued_at timestamp, issuer, content_fingerprint, data_hash, and cortex_hash are canonicalized and signed using HMAC-SHA256.
3. Instant Verification (POST /v1/pdf/verify): When a PDF is submitted for verification, Tailprint recomputes the visual page stream fingerprint from the uploaded file and compares it against the signed manifest and proof ledger in constant time (crypto.timingSafeEqual).

Code Example: Issuing & Verifying a Signed PDF

// 1. Render a cryptographically signed PDF with a visual verification seal
const renderRes = await tp.renderPdf({
  template_id: 'certificate-classic',
  data: {
    recipient_name: 'Dr. Elena Rostova',
    certificate_title: 'ISO-27001 Lead Auditor Certification',
    issue_date: '2026-11-13',
    issuer_name: 'European Compliance Institute'
  },
  sign: true,
  verify_seal: true,
});

// 2. Later: Verify authenticity and detect any visual or data tampering
const report = await tp.verifyPdf(renderRes);
console.log(report.valid);            // true
console.log(report.tampered);         // false
console.log(report.signature_valid);  // true
console.log(report.content_intact);   // true