Multi-layered defence
Remote identity verification has to confirm several things at once: that the evidence is genuine, that the person is real, that they own the identity, that the capture was not manipulated, and that the overall behaviour is not suspicious. No single check covers all of these — so the IDV Platform applies a series of independent, reinforcing controls. Fraud that slips past one layer is caught by the next.
Why one control is not enough
Each control has a blind spot that another layer covers:
| Control used alone | What it can miss |
|---|---|
| Document authenticity | A genuine but stolen document |
| Face comparison | A high-quality spoof or manipulated portrait |
| Liveness detection | Injected or replayed camera input |
| NFC chip verification | A genuine document used by the wrong person |
| Data / rule checks | A synthetic identity assembled from valid data |
Layering the controls means an attacker has to defeat all of them in a single session — which is far harder than defeating any one.
The protection layers
Each layer maps to capabilities documented elsewhere in this section; follow the links for detail.
Layer 1 — Channel & session protection
Establishes that the evidence comes from a trustworthy device and app, in real time, bound to a single session. Detects injected or replayed media (virtual cameras, pre-recorded streams), substituted or reused evidence, and continuation of an expired session. → Injection attack detection
Layer 2 — Document authenticity
Determines whether the presented document is genuine, consistent, and unmodified: screen/photocopy presentation-attack detection, data extraction and cross-checks (MRZ, barcode, visual zone), and NFC chip authenticity with chip-vs-print comparison. → Document verification
Layer 3 — Biometric binding
Confirms the person in the session is the holder of the document: 1:1 face comparison against the document (and chip) portrait, plus a duplicate-identity search and watchlist/blocklist checks. → Face comparison · Duplicity check · Watchlists
Layer 4 — Liveness & presentation-attack detection
Confirms the biometric comes from a real, present person rather than an artefact — printed photos, photos on a screen, replayed video, masks or 3D artefacts, or manipulated portraits — including AI-generated imagery. → Passive liveness · Deepfake detection
Layer 5 — Data & risk intelligence
Catches fraud that is not visible from the document or selfie alone: validity and rule checks (expiry, issuing country, document type, estimated age vs date of birth, company-specific acceptance rules) and risk signals such as repeated attempts, unusual device reuse, and abnormal failure velocity. → IDV Configuration
Layer 6 — Decision orchestration
Combines the outputs of every layer into one explainable Accept / Review / Reject decision through configurable Trust Factors, and applies per-workflow policies so verification strength matches the operation's risk. It also resolves duplicity conflicts (unique / merge / reject) rather than leaving them ambiguous. → Trust Factors & decisioning · Workflows
Layer 7 — Continuous improvement
The models and rules are kept current against evolving fraud through representative datasets, manual-review feedback used as ground truth, controlled rollout with false-accept / false-reject analysis, and post-deployment monitoring.
Risk-based verification
Not every operation needs the same rigour. Because workflows and Trust Factor profiles are configurable, a Company can match verification strength to the risk of the action — a lighter journey for low-risk operations, stronger checks for high-risk ones.
| Operation | Typical risk | Verification emphasis |
|---|---|---|
| Account creation / onboarding | Medium | Full stack — document + face + liveness + injection detection, with duplicity check |
| Returning-user sign-in / step-up | Medium–High | 1:1 face comparison to the record + liveness + injection detection |
| Personal-data update | High | 1:1 face comparison to the record + liveness + injection detection |
| High-value action (e.g. transfer, loan) | High | 1:1 face comparison + liveness + injection detection |
See also
- Security & Anti-fraud
- Trust Factors & Decisioning
- How It Works — the end-to-end verification pipeline
- Deduplication