Liveness detection explained, how a face check spots photos, screens and replays

The spoofs a face check has to stop

• A printed photo held up to the camera.
• A photo or video on a screen, a phone or tablet shown to the webcam.
• A replay, a recording of the real player passing an earlier check.

Photos of most people are one search away. A check that only compares a still selfie against a stored face invites all three.

Passive versus active liveness

Passive liveness looks at a single frame or a short clip for signs of a fake, such as print texture, screen glare and moiré patterns. It is invisible to the player, but on its own it is a bet that the model has seen every trick.

Active liveness asks the player to do something. The strength of it comes from randomness. If the challenge is the same every time, a recording of a past attempt passes. If the order of head turns and the moment of the blink change every check, a recording can't follow along.

How Netrik does it

Every Netrik check combines both. The player gets a random sequence of head turns and a blink, tracked live in the browser, while an AI anti-spoof model looks at the frames for printed photos and screens. Only then is the face matched against the registered player. Your backend receives the liveness and anti-spoof scores with the verdict, by signed webhook; the player only sees pass or fail, so a fraudster learns nothing about what tripped the check.

The whole thing takes about ten seconds on any phone or laptop browser, which matters as much as the security. A check players abandon protects nothing.

What liveness doesn't do

Liveness proves a live person is present. It doesn't prove who that person is in the legal sense, which is what document checks are for. Netrik is built for everyday fraud prevention, not as certified identity proofing. For the full picture of where it fits, see facial KYC for iGaming.