Biometric
For crypto exchanges

In crypto, a mistake at signup can't be undone.

Let the wrong person in and the funds are gone in minutes. Turn a real one away and they leave for good. Biometric.Vision covers both sides: verification in seconds, accuracy proven by NIST.

500 checks a month free— no card
Verification · crypto_signupPASS
Photo in the document
Customer's selfie
98%face-to-document similarity
Document readpassport · RUS
Liveness confirmedlive person
Face matches98%
AML screeningclear
Approve the customerAll checks passed · 1.4 s
Someone else's passport and deepfakes are rejected automatically
The cost of one mistake

Sound familiar?

3 a.m., a withdrawal alert. The account was taken over with a stolen password, the transaction confirmed. There are no chargebacks in crypto — the money is gone for good. In the morning the user posts in the public chat that the exchange "drained" their wallet.

In an industry where reputation is the whole asset, one case like this costs more than a month of marketing. And a passport check at signup does not save you from it — you need something that confirms the owner is the one behind the screen.

Wallet
03:00
Withdrawal confirmed− 4.20 BTCTo external address bc1q••••7ha9 · irreversible
New device · password login
On-chain transaction — it cannot be reversed.
Three outcomes

What you get

01

Onboarding in seconds

The customer completes eKYC from their phone — document and selfie, with on-screen capture hints. They pass on the first try and don't leave for a competitor.

Auto frame capture and shooting hintsDocument and selfie in one flow
PASSPORT
SurnameMORGAN
NameALEX
Date of birth12.06.1994
P<UTOMORGAN<<ALEX<<<<<<<<<<<<<1203567890UTO9406129F3001018<<06
Keep the document in frame
Done
DocumentSelfieDone
02

Fraud doesn't get through

Someone else's passport, a deepfake, a hijacked account — all filtered out automatically, with no manual review.

Liveness against photos, video and masksDetection of AI generation and stream injection
PASSPORT
SurnameMORGAN
NameALEX
Date of birth12.06.1994
P<UTOMORGAN<<ALEX<<<<<<<<<<<<<1203567890UTO9406129F3001018<<06
SPOOF · STRANGER'S PASSPORTDEEPFAKEACCOUNT TAKEOVER
Someone else's passport
Deepfake
Account takeover
03

One person — one account

The face is checked against your entire database. Multi-accounts, sybil attacks and bonus hunting are blocked at signup.

1:N search across your whole customer baseA second account on the same face will not pass
New registrationchecked against the whole customer base
Database · 1,240,508
Duplicate found · 97% similarity
Accuracy = trust

NIST-verified accuracy — not just our word

Our face-matching algorithm ranks 37th out of 570+ worldwide in the independent NIST rating, the U.S. standards institute. In practice the system almost never mistakes one person for another — neither letting a fraudster through nor rejecting an honest customer.

37 / 570+
in the world on the NIST FRVT ranking
28 / 10M
errors per 10 million comparisons
#1
in Central Asia for face comparison
Against AI fraud

A deepfake costs pennies today. Your defense has to cost more than the fraud

Generating an AI face, a synthetic identity or injecting a fake camera stream is now within anyone's reach — and it is aimed squarely at exchange onboarding. Liveness that told a live person from a photo "by eye" five years ago does not catch it.

Biometric.Vision confirms there is a live person in front of the camera — not something generated an hour ago.

Blocked at signup
Photos and videoblocked
Silicone and 3D masksblocked
AI generation and deepfakesblocked
Video-stream injectionblocked
Segments

Built for your model

Centralized exchange (CEX)

Strict eKYC at signup and a repeat face check before a large withdrawal.

On/off-ramp services

Fast verification at the fiat-crypto boundary, where fraud risk is higher — without losing conversion.

Wallets and custody

Owner confirmation when accessing funds and changing security settings.

How it works

Four steps

01

Capture the document

The customer photographs the document — the system hints how to get it right the first time.

PASSPORT
SurnameMORGAN
NameALEX
Date of birth12.06.1994
P<UTOMORGAN<<ALEX<<<<<<<<<<<<<1203567890UTO9406129F3001018<<06
Hold steady
Done
02

Recognition

The data is read automatically and the document type is detected on its own.

PASSPORT
SurnameMORGAN
NameALEX
Date of birth12.06.1994
P<UTOMORGAN<<ALEX<<<<<<<<<<<<<1203567890UTO9406129F3001018<<06
SurnameDate of birthNumberMRZ
03

Selfie and match

The customer takes a selfie — liveness and the face-to-document match are checked.

98%
04

Verdict

You get a verdict via the API and in the dashboard: approve, reject or send to review.

Document is genuine
Liveness confirmed
Face matches · 98%
Approve the customer
Integration

Integrate in a day, not a quarter

One REST API, a Mobile SDK and a web widget — it drops into your current user flow without reworking the interface. A developer wires it up in a few hours, or delegates the integration to an AI agent (Cursor, Claude Code).

verify.js
const result = await biometric.verify({
checks: ['document', 'liveness', 'face_match', 'aml'],
scenario: 'crypto_signup'
});
// → { decision: 'approve', face_match: 0.98, aml: 'clear' }
Trust

Crypto exchanges choose Biometric.Vision

ISO 27001 № 760281NIST FRVT ranking#1 in Central Asia
FAQ

Questions & answers

To comply with AML/CFT, keep fraudsters out and protect user funds. In most jurisdictions verification is mandatory for access to trading and withdrawals.

500 checks a month free

Scale your exchange without compromise

Create an account and get API access in 60 seconds. Test it on your own registration flow — no card.