What is a face liveness check?
A face liveness check assesses whether the face in a submitted capture belongs to a person who was actually present. The alternative is a printed photograph, a face displayed on a screen, a replayed video or a mask held up to the camera.
The question is narrow on purpose. Liveness does not establish who the person is. It assesses whether what the camera captured behaves like a live human being, which is the assumption every other remote identity check quietly depends on.
Face liveness and face match answer different questions
Face match compares two faces and reports how similar they are. Face liveness looks at one capture and assesses whether a live subject produced it.
A borrowed photograph can pass a match check while failing a liveness check, and a live capture of the wrong person does the opposite. Most remote journeys use both: liveness to establish that a real person is present, and a match to establish which person.
How the face liveness API works
The check runs inside your own capture flow. Your application collects the selfie or short capture on the customer's device, submits it, and receives an assessment it can act on without a person reviewing every session.
The check may be passive or prompted, depending on the selected service contract. A passive check assesses the capture without asking the customer to do anything, which keeps the journey shorter. A prompted check asks the customer to perform a simple action during capture, and gives your flow something explicit to evaluate.
The capture your app collects and sends
The check starts from a capture your own application collects: a selfie image or a short capture, in the formats and capture modes your service contract supports. What comes back is an assessment your workflow can act on, and the detail returned with it may vary by capture, data source and record.
Poor lighting, a compressed upload, a partly obscured face, an unstable connection or a low-end camera all reduce what can be assessed. A capture that could not be assessed is different from one that failed, and a flow that treats them alike turns away customers it should have retried.
Where a liveness check earns its place
- Lending and NBFCs confirm a live applicant before a remote onboarding or disbursal step, where nobody sees the borrower in person.
- Insurance adds a live capture to remote proposal journeys and claim intimation, where a supplied photograph has always been the easy thing to send.
- Fintech and payment service providers use liveness as a step-up before a high-risk account action, such as a device change or a change to payout details.
- Staffing, gig platforms, logistics and fleet check at shift start or vehicle handover that the worker on the job is the worker who was onboarded. It is the practical answer to shared accounts.
- Marketplaces, telecom and travel re-verify a seller after a change of banking details, and support remote activation and unattended self-service check-in.
How to add a liveness check to a customer journey
The design decisions matter more than the integration:
- Place the check where the risk is: onboarding, a step-up action, or periodic re-verification.
- Give a clear retry path with guidance. A poor capture should not read as a rejection.
A liveness check is one control and does not act as a perimeter. Treat the result as a signal that raises or lowers confidence, and combine it with your other session controls and the review path you already run.
Consent and storage for face captures
A face capture is sensitive personal data and belongs in a consent-led flow. Tell the customer what you capture, what you use it for and how long you keep it. Offer a path for anyone who cannot complete a camera step.
Keep captures out of general-purpose storage and logs, restrict who can view them, and remove them when the purpose is served or your retention obligations end.
Availability and supported capture modes depend on the selected service contract. How you use the result stays your organization's responsibility under applicable data protection rules.