Solution · Biometric Verification

Biometric Verification That Catches the Deepfakes

One Constellation verifies identity through document-to-selfie face match, active and passive liveness detection, NFC ePassport chip read, and deepfake detection — NIST PAD-aligned, accessible across devices, and built for the volume modern onboarding demands.

The Challenge

Photo-Capture KYC Was Always Going to Lose to AI Deepfakes

The shift to digital onboarding solved one problem and created another. Customers expect to verify their identity in minutes, on a phone, without paper. Regulators expect institutions to remain certain the customer is who they say they are. The instrument that bridges those two expectations is biometric verification — face capture, liveness checks, and binding the biometric to a verified identity document.

The threat landscape has moved fast. Static photos no longer pose much challenge to detection. But injection attacks using pre-recorded video, presentation attacks using printed faces, 3D masks, screen replays, and now AI-generated deepfake video are all in the wild — and the gap between fraudster capability and basic-tier liveness detection is now wide enough that some onboarding flows are effectively defenceless against motivated attackers.

One Constellation handles biometric verification as a layered defence: NIST-aligned active and passive liveness, document-to-selfie face match, NFC chip read for ePassports, deepfake-specific detection, and continuous model retraining against the evolving threat landscape. Built to verify legitimate customers in seconds and to surface the attacks that matter for human review.

How It Works

Multi-Modal Biometric Verification Built for Real Threats

Layered defence

Face match alone is not biometric verification. Active liveness (challenge-response), passive liveness (texture and depth analysis), document-to-selfie biometric binding, NFC chip read where available, and deepfake-specific detection all run together — with a layered scoring model that surfaces ambiguous cases for human review.

Accessibility & conversion

Customers complete verification on any modern device, including older smartphones. Active-liveness challenges are accessible for users with motor or visual impairments. Capture-quality guidance prevents the failed-retry loops that destroy onboarding conversion rates.

Biometric verification has to do two things that pull in opposite directions — let real customers through in seconds, and stop sophisticated fraud at the door. Anything that does only one is a tool for the other side.
— Operating principle, One Constellation biometric engine
Operational Impact

Biometric Verification, Measured

Performance figures from One Constellation biometric verification across production onboarding flows.

<3s
Avg. Verification Time
99%+
Genuine-User Pass Rate
NIST PAD
Aligned Liveness Standards
Deepfake
Specific Detection Layer
190+
ID Documents Supported
What's Included

Full Biometric Stack — Not Just Face Match

Biometric verification on One Constellation is a layered capability covering document verification, biometric match, liveness, NFC and deepfake-specific detection.

🪪

Document & Selfie Capture

Branded customer-facing capture flow with quality guidance, retry handling and accessibility for low-spec devices and impaired users.

Explore KYC →
🔬

Liveness & Match Engine

Active and passive liveness running in parallel with document-to-selfie biometric match — single decision returned in under three seconds.

Explore Onboarding →
📡

NFC Chip Read

Where the ID document is an ePassport or chipped national ID, the platform reads the chip directly — the strongest possible identity binding.

Explore Portal →
Capabilities

Built for the Threat Landscape of 2026, Not 2018

The features that distinguish modern biometric verification from photo-capture KYC.

🎭

Active & Passive Liveness

Active challenges (head movement, blink, smile) combined with passive analysis (texture, depth, micro-movement, illumination) — NIST PAD Level 2 aligned across both.

🤖

Deepfake Detection

Dedicated detection layer for AI-generated face content — synthetic media indicators, generative-model artefacts and video-injection patterns surfaced explicitly.

📡

NFC ePassport Read

Direct chip read from ePassports and chipped national IDs (Singapore IC, EU eIDAS, UK passport) — the highest-assurance biometric binding available.

📱

Cross-Device Capture

Works on iOS, Android and modern web browsers — including older devices, low-bandwidth connections and accessibility-mode users.

🌍

190+ ID Documents

Document verification covers passports, national IDs, driving licences and residence permits from 190+ jurisdictions — with security-feature checks beyond OCR.

📊

Reviewable Evidence

Every verification preserved with captured frames, scoring details, liveness result, deepfake indicators and document evidence — reviewable by analysts and exportable for audit.

Standards Alignment

Aligned to the Biometric Standards Regulators and Industry Bodies Recognise

Biometric verification on One Constellation is built against the published industry standards that regulators reference in their AML and digital-identity guidance.

🏅
NIST PAD
Presentation Attack Detection
📐
ISO 30107-3
Biometric PAD Standard
🌟
FATF Digital ID
Digital Identity Guidance
🇪🇺
eIDAS / EU DI Wallet
EU Digital Identity Framework
🇸🇬
MAS e-KYC Guidelines
Singapore Digital Onboarding
🇬🇧
JMLSG Digital ID
UK Digital Identity Guidance
🇺🇸
NIST 800-63A
Identity Assurance Levels
🔐
ISO 27001
Information Security Management
Part of the Platform

Biometric Verification, Native to the Onboarding Flow

Biometric verification doesn't sit in a separate vendor stack. It's wired into the KYC workflow, the customer record, the risk-scoring engine and the audit trail — one decision, one evidence chain.

🪪
Part Of
KYC · Customer Onboarding · IDV

When biometric verification completes, the result becomes part of the customer's KYC record — pass / fail, with the underlying evidence preserved. Failed verifications route to analyst review with the captured frames and scoring detail. Successful verifications proceed to risk scoring, sanctions and PEP screening, and ongoing monitoring.

The verification evidence remains in the customer record for the duration of the relationship and beyond — replayable for any subsequent investigation, exportable as evidence in any regulator's expected format.

Biometric Verification FAQ

What Onboarding & Compliance Teams Ask Us

What liveness standards does the platform meet?+
Active and passive liveness aligned to NIST PAD Level 2 and ISO/IEC 30107-3, with continuous model retraining against the current threat landscape. Liveness runs as a layered defence (active + passive in parallel) rather than a single-mode check.
How does the platform handle deepfakes and synthetic media?+
A dedicated detection layer analyses for AI-generated face content using synthetic-media indicators (generative-model artefacts, anomalous texture patterns, depth-map inconsistency, and video-injection indicators where a recorded or generated stream is substituted for a live camera feed). Detection capability is continuously updated against new generative-model families.
What devices and browsers are supported?+
Native iOS and Android SDKs, plus a web SDK that works on modern browsers (Safari, Chrome, Firefox, Edge). Capture quality remains usable on older smartphones and on lower-bandwidth connections — capture-quality guidance prevents failed-retry loops that destroy conversion rates.
How is accessibility handled for users with impairments?+
Active-liveness challenges have alternative modes for users with motor and visual impairments. Capture flows support screen-reader compatibility and high-contrast modes. The platform retains the user's ability to complete verification without forcing capture patterns that disability would prevent.
Which ID documents are supported?+
190+ jurisdictions covering passports, national IDs, driving licences and residence permits. Document verification goes beyond OCR — security-feature checks (holograms, MRZ validation, watermarks, font and layout analysis) are performed alongside the optical character read.
What about data residency for biometric data?+
Biometric data is processed in the customer's chosen residency region — Singapore by default, with EU, UK, US, UAE and other regional options available. Biometric templates can be stored for re-verification or deleted post-decision per the customer's policy and regulatory obligations.

Identity Verification That Holds Against Real Attacks.

See how One Constellation handles biometric verification — layered liveness, deepfake detection, NFC chip read and document verification across 190+ jurisdictions.

Scroll to Top