HOW IT WORKS
The fastest way to understand it is to take the check yourself.
THREE STEPS
The check loads in the respondent’s browser when they reach your survey. No app to install, no redirect maze, no friction.
The check runs on the device — face shape, small natural movements, the device and the session. The camera picture never leaves the device.
You receive a signed result — real person · not seen before · not a bot — before they answer a single question. Fraud is stopped at the door, not cleaned up later.
The image and video never leave this box.
WHY IT TAKES A FEW SECONDS
Most fraud tools feel instant because they ship your respondent's data to a server — IP address, a device fingerprint, a selfie, sometimes a government ID. VerifyHuman does the opposite.
The analysis runs on the respondent's device. Those few seconds are the face and behaviour checks running there. What leaves the device is a short list of face measurements (numbers) that cannot be turned back into an image of your face — never a photo, and never your name or identity.
WHERE YOUR FACE GOES
Your face never leaves your device. No image is uploaded or stored.
Your photo is sent to a server, where it is matched and may be retained.
SILENT AND CAMERA CHECKS
Same script, same API. A silent check needs no camera, and on the Server API nothing on the page at all. A camera check is the strongest proof: choose it for a study that needs it, or turn on the option that lets an unsure silent check on your page ask for one (off by default).
Network, device, behaviour and track record — no camera, and, with the Server API, nothing on the page at all. It can run for every respondent and lets honest people through without them noticing. It stops bots, click farms and AI agents on its own; a camera check adds proof that each person is unique.
The face is the one thing a fraudster can't swap. They can spin up a fresh device, a clean IP, and human-like behavior — and still fail here, because they can't become a different, live person on demand. Face geometry proves a real, present human and that they haven't already taken your study — without ever learning who they are.
No trade-off
Worried a camera costs you completes? It doesn't have to. Start silent — respondents see no puzzles or hoops — and use a camera check only where a study needs it, or let an unsure silent check ask for one. If a camera check can't finish, you choose what happens: use the camera result as it is, or fall back to the silent check. By default, if our service can't be reached, the respondent continues and the response is marked for review. You keep your response rate and your data quality.
DEMOGRAPHIC CHECK · BETA
In a camera check, VerifyHuman estimates an age range and gender from the shape of the face, in the browser, so you can compare it against what the respondent claimed. It catches the 45-year-old who says he's a 24-year-old woman to qualify for a quota. Like everything else: derived from face measurements, worked out on the device, never stored, never an image.
Beta — estimates, not identity. Something to compare, not a result.
Pay per check, rate drops with volume — and an honest look at how we compare.
