VU cloud services

Detect AI-powered identity spoofing attempts

Manipulated photos, deepfakes and synthetic identities can enable fraud attempts in facial verification processes. Antispoofing Facial is a service you can consume via API to analyze selfies and determine whether they belong to a real person present in front of the camera or to an image used to fool the system.

To generate your API Key, you need a paid VU plan.

The same selfie with Antispoofing Facial's analysis overlaid. "spoofing": true, "fake": 0.95, "confidence": 0.99

Selfie of a person, unanalysed

Security backed by certifications and independent testing

iBeta PAD Level 3 compliant

ISO/IEC 30107-3 · Liveness

ISO/IEC 27001:2022

Information security management

ISO/IEC 27017:2015

Security for cloud services

The problem

If it doesn't show up in your data, your data may not know how to look for it

Deepfake fraud does not always look like fraud. A constructed identity can behave like a legitimate user, meet its obligations and never trigger an alert. The industry calls it measurement blindness: the absence of a signal in your data does not mean the risk is not there.

The Deepfake Summit Event Intelligence Report estimates that between 15% and 20% of some portfolios could hold dormant synthetic identities with clean credit histories. This is an estimate, not a measurement. Read the report.
Our solution

What this API answers, plainly.

For every image it determines whether there is a real person present in front of the camera or a spoofing attempt. It does not determine whether the identity behind it is synthetic: that is a different control.

Case 01

You already have another provider

You are not going to migrate everything to try one specific capability. You consume only this API, alongside the provider you already have.

Case 02

You want to audit what you already have

You run the API in batch over your image archive, without touching the flow in production. You send selfies at 640 × 480 or above: below that resolution it loses accuracy.

Case 03

You are building your own product

You need the detection piece, not a whole platform. You integrate it and keep control of the rest.

Key features

One question, one clear answer

VU's other APIs return data for your policy to interpret. This API returns a verdict directly. That is why you can easily test it against cases you have already identified.

  • Edges of another screen inside the frame
  • Photocopies and printouts
  • Watermarks
Portrait analysed by the API, with the facial landmarks marked
  • Faces generated by artificial intelligence
  • Screenshots
  • Colour alterations

It works on an image, not on a camera stream

  • Liveness detection

    It takes a selfie that you send, without controlling the capture. It does not depend on how or when the image was taken. Liveness detection is a camera-driven flow.

  • Integration

    It does not replace Identity Link when you need a full verification. It is a piece that integrates on its own or with VU's other 3 solutions.

  • Complementary controls

    A deepfake is not stopped by a single control. This API analyses the image. An injection enters the flow directly, with no camera involved, and calls for device controls. They are different threats and they call for different controls. How that defence is put together.

The integration

Two endpoints and one decision rule.

No SDK, no dependencies. Call the API from your backend: the key is never exposed in a mobile app or in the browser.

ElementHow it works
GET /versionChecks connectivity and credentials before integrating
POST /inferenceSends the image and returns the verdict
AuthenticationApi-Key header. Each service has its own key, per environment
Base URLVU provides it along with the key, depending on the environment
Inputimage in base64, info as a business reference and logAnonymity
Outputid, code, spoofing, message, confidence and benchmark
RequestPOST /inference
// Header: Api-Key: <your service key>
{
    "image": "<JPEG in base64, no prefix>",
    "info": "your business reference",
    "logAnonymity": true
}
  • "spoofing": false The image comes from a real person present. You approve.
  • "spoofing": true There is a spoofing attempt. You do not approve.
The decision ruleYour backend
if (status === 200 && body.spoofing === false) {
    approve();
} else {
    doNotApprove();
}
Image requirements

The size of the selfie decides how much accuracy you get.

Aspect
Recommended resolution
Value
640 × 480. This is the resolution the service is calibrated with*
Aspect
Format
Value
JPEG, quality 95 or higher
Aspect
Encoding
Value
base 64 with no data: prefix and no line breaks

*The minimum accepted resolution is 224 × 224; anything lower will return an error.

Send 640 × 480 even though the minimum is lowerBelow the calibration resolution the API answers just the same, with less accuracy and no warning. Capture quality is an integration decision: settle it at the start.
How you start

Self-service: connect this API with no middlemen

With a paid plan, you generate the key from VU's panel and call the service from your backend.

  1. 02

    You confirm accesswith the method the panel offers you

  2. 03

    You open Cloud Servicesin the side menu

  3. 04

    You choose Antispoofing Facialamong the available services

  4. 05

    You generate the API keywith a paid plan, for the environment you're going to use

  5. 06

    You copy the key and call GET /versionto confirm that everything responds

Generate your API Key

To generate your API Key, you need a paid VU plan.

Frequently Asked Questions

No. You can consume it independently, with its own key, over whichever flow you choose. It coexists with the provider you already have installed.
Yes. It works on a still image, so it serves both live and in batch over what you have already captured. Accuracy depends on the image's resolution.
Liveness detection drives the capture with the camera and verifies presence at that moment. This API analyses an image that you send to it.
In your backend, in the Api-Key header. Never in a mobile app or in the browser. One key per service and per environment.
Below 224 × 224 it returns 400. Between that minimum and 640 × 480 it responds, but with less accuracy. Send 640 × 480 whenever you can.

You already have the images. The answer is missing.

Generate your key and test a selfie from a case you already know.

Generate your API Key

To generate your API Key, you need a paid VU plan.