Solution
Your KYC provider checks the face. TamperCheck checks the document.
Most KYC stacks were built for a 2015 threat model: photo swaps, expired documents, mismatched dates. They handle face matching and watchlist screening. What they don't do is forensically verify that the submitted document is genuine. TamperCheck fills that gap - analysing every identity document for tampering, AI generation, and field-level edits.
- No document storage
- Digital, scanned or photographed
- ~1 minute per verdict
Last updated:
The problem
Why legacy KYC can't catch AI-generated documents
Most KYC providers were built on a 2015 threat model: photo swaps, expired documents, mismatched dates. They use template matching and OCR validation - which works fine until the attacker uses an AI tool to generate a fully synthetic document that doesn't match any template the provider has seen.
Deepfake detection for KYC is now a separate problem from face matching. The face is real (or convincingly synthetic), the document is plausible, and every visible field is internally consistent. The only signals are forensic: noise patterns, spectral artifacts, font rendering inconsistencies, and producer metadata.
AI KYC has to operate at two layers: identity verification (is this person who they claim to be?) and document forensics (is this document authentic?). Most providers do the first well. Almost none do the second. TamperCheck fills the gap.
What we detect
Forensic signals that expose fraud
Deepfake document detection
Catches fully AI-generated IDs, passports, and driving licences using spectral and noise pattern analysis trained on current-generation generative models.
Photo zone substitution analysis
Detects portrait swaps via boundary contrast, sharpness gradients, and compression mismatch in the portrait region.
MRZ integrity & checksum validation
Validates Machine Readable Zone characters, checksums, and consistency with visible passport fields.
Field-level edit detection
Detects edits to date of birth, expiry date, ID number, and name fields - the most commonly altered fields on identity documents.
Security feature plausibility
Analyses holograms, security overlays, and watermark patterns for plausibility relative to the issuing authority's known templates.
Cross-document consistency
When multiple documents are submitted (passport + payslip + utility bill), cross-checks for inconsistent metadata, producer fingerprints, or generation signatures.
How-to
How to detect forged documents in KYC
Template matching confirms an ID looks like the right kind of document. Detecting a forgery means proving the document was genuinely issued and never altered - which takes these five checks on every submission, not just flagged ones.
Validate the machine-readable data
Recompute the MRZ check digits and compare the MRZ, barcode and chip data with the printed fields. Forgers edit the visible text and forget the encoded copy.
Inspect the photo zone
A swapped portrait leaves blending edges, lighting that disagrees with the rest of the card, and resolution or noise breaks at the boundary.
Check fonts and layout against the issuer
Each issuer uses fixed typefaces, spacing and field positions. Re-typed names, dates and numbers drift from them at the character level.
Test for AI generation and face swaps
Fully generated IDs and face-swapped portraits carry generation artifacts that a template check never looks for. Run a dedicated AI-generation pass on every image.
Verify the supporting documents too
Proof of address, bank statements and payslips submitted after the ID check are forged far more often than the ID itself - and they never pass through the capture SDK.
The solution
The forensic document layer your KYC stack is missing
The document forensics layer in a KYC pipeline. Keep your existing provider for face matching, watchlist, and OCR. Add this as the forensic check on every submitted document.
How it works
- Drop-in forensic layer - works alongside Onfido, Jumio, Persona, Sumsub
- Deepfake and AI-fake detection - the layer your KYC provider doesn't have
- Works on any country's documents - forensic analysis, not template-based
- API-first - integrate in under 30 minutes
- per-document pricing - less than re-running KYC checks
- Zero document storage - critical for KYC compliance
- Findings your compliance team can act on - not just a flag
Why us
What you won't get elsewhere
Findings, not a score
We name the field that changed and where it sits on the page.
Others hand you a number to justify.
One endpoint, every type
100+ document types, 190+ countries, no per-type setup.
Elsewhere, each new form is a project.
Catches what was never real
A dedicated pass for AI-generated and deepfake documents.
Template matching only catches edits.
Nothing is retained
Analysed in memory, then discarded. Never trained on.
Most services keep documents for QA.
Live in minutes
Published pricing, no minimum. Test it before you talk to us.
Platforms want a contract first.
Sits alongside your stack
Keep Onfido or Jumio for face match. We add the forensics.
Not a rip-and-replace.
FAQ
Common questions
What is AI KYC?
Using AI to automate Know Your Customer verification - reading identity documents, matching faces, screening watchlists, and detecting forgery. Modern stacks combine OCR, face matching, liveness, and document forensics. TamperCheck provides the forensics layer.
How is AI KYC different from traditional KYC?
Traditional KYC relies on template matching, manual review, and basic OCR. AI KYC adds machine learning for face matching, liveness detection, document classification, and forgery detection. The biggest gap in most stacks is document forensics - that's where deepfake and AI-generated IDs slip through.
Can AI KYC catch deepfake documents?
Only with a forensic layer trained on generative AI signatures. Most providers focus on face deepfakes, not document deepfakes - different problems. TamperCheck specialises in document deepfake detection, using spectral analysis and noise patterns to catch synthetic IDs.
How do you detect forged documents in KYC?
Validate the MRZ and barcode data against the printed fields, inspect the photo zone for splicing, compare fonts and layout with the genuine issuer template, run an AI-generation and face-swap check, and apply the same forensics to supporting documents such as proof of address. TamperCheck runs all of these in one API call per document, alongside your existing KYC provider.
Do I need to replace my existing KYC provider?
No. TamperCheck slots in alongside any existing provider. Keep Onfido, Jumio, Persona, or Sumsub for face matching and screening; add TamperCheck as the forensic check on every document. One API call per document.
Is AI KYC compliant with GDPR and AML regulations?
TamperCheck processes all documents ephemerally - file bytes are analysed and immediately discarded, nothing stored. That's GDPR-friendly by design and supports zero-retention policies common in regulated industries.
How fast is AI KYC document verification?
TamperCheck returns a verdict fast enough to run synchronously in real-time onboarding flows. For batch reviews, the API supports async webhooks. No per-account rate ceiling on the standard plan.
Or skip the call entirely
We'll match your first top-up in credits, so you can run your own documents through it now.