Video KYC verifies a customer over a live video session, sometimes with a human agent, where they show their identity document and their face in real time. Live prompts and human judgment raise assurance against static spoofs, which is why some regulators accept it as a remote stand-in for meeting in person.
What is video KYC, in plain English?
Video KYC is identity verification done over a live video call. The applicant joins a session, holds up their ID, shows their face, and responds to prompts, either from a trained agent or a guided automated flow. The idea is to recreate, remotely, the assurance of a face-to-face check where you can see the person and their document together in real time.
Its strength comes from being interactive and unscripted. Live prompts, follow-up questions, and a human reading the applicant's reactions are harder to beat than a single uploaded photo or a pre-recorded clip. In several markets it is an accepted way to satisfy KYC obligations without an in-branch visit, which made it central to remote onboarding.
In the fraud stack, video KYC sits inside identity verification, often for higher-risk or higher-value onboarding. But the same live channel that raises assurance has become a prime target for deepfakes, virtual cameras, and coached applicants, so a convincing feed can no longer be trusted on its own.
How a session runs
- Join — Start the live session. The applicant connects on video with an agent or a guided automated flow.
- Show — Present the document. They hold up the ID so it can be read and checked against the live face.
- Prompt — Respond to live prompts. Unscripted questions and movements test that a real person is present now.
- Decide — Score and record. The session is judged, backed by device and liveness signals, and archived for audit.
What it looks like in practice
In practice
An applicant joins a video KYC session and presents a convincing ID. The face on screen matches the document and answers the agent's questions fluently. It looks clean, but the agent notices small tells: the mouth movements lag the audio slightly, the edges of the face shimmer when the head turns, and the lighting on the face does not match the room.
Behind the session, device integrity checks flag a virtual camera feeding the stream. The agent escalates, and the case turns out to be a live-video deepfake driven through injected video, paired with a real stolen ID. The session is failed, the device signature is added to a block list, and the reviewers are briefed on the specific deepfake tells to speed up the next catch.
Why it matters to operators
Video KYC buys assurance that a static selfie cannot: an interactive, unscripted session is far harder to fake than an uploaded photo, and in regulated markets it can satisfy KYC without a branch visit. For higher-risk onboarding, that combination of assurance and remote convenience is valuable.
The danger is complacency. A live feed feels trustworthy, yet it can be a deepfake driven through a virtual camera, or a genuine-looking applicant reading a script under coercion. Train reviewers on deepfake and injection tells, and back every session with device-integrity and liveness signals, because the human eye alone is no longer a reliable defense against a well-made synthetic feed.
What to watch for
- Audio-video mismatch. Lip movements that lag the speech, or unnatural blinking, can signal a deepfake driving the feed.
- Virtual cameras. A session sourced from a camera driver or emulator rather than a physical sensor is a strong injection tell.
- Coached applicants. Reading from a script, glancing off-screen, or unnatural pauses can mean someone is being directed.
- Edge artifacts. Shimmer around the hairline or face edges during movement points to a synthetic overlay.
- Document and face seams. An ID that never quite shares the same lighting or focus as the face may be composited.
Quick questions
Is video KYC more secure than a selfie check?
It offers higher assurance because it is live and interactive, which is harder to fake than a single uploaded image. But it is not immune: deepfakes and injected video can defeat it, so it still needs device and liveness signals behind the session.
Can a deepfake beat video KYC?
Yes, a live-video deepfake fed through a virtual camera can pass a session that relies on the human eye alone. That is why reviewers should be trained on deepfake tells and every session backed by device-integrity and liveness checks.
Does video KYC always need a human agent?
No. Some flows are agent-led while others are automated and guided, with a human reviewing only escalations. The assurance comes from the live, interactive nature of the session rather than the presence of an agent specifically.
What is a coached applicant?
Someone real who joins the session but is being directed by a fraudster, often a scam victim or a paid face, reading a script to open an account that will be handed over. Watch for off-screen glances, unnatural pauses, and scripted answers.
Why record the session?
For audit and for later review if fraud surfaces. A recorded session lets you re-examine tells you missed live, supports regulatory evidence of the check, and helps train reviewers on real attack patterns.
Go deeper
- NIST Digital Identity Guidelines (SP 800-63) ↗ — The US standard for identity proofing and authentication assurance levels.
- FATF ↗ — The global standard-setter for AML, counter-terrorist-financing, and counter-proliferation. Recommendations, guidance, and jurisdiction lists.