Veriff has launched Presence Verification to confirm whether a real, present, and freely participating person is behind a digital session.
The capability extends Veriff’s identity verification offering beyond conventional liveness detection, which checks whether a face shown to a camera is genuine and not a photo, mask, or replayed video. According to the company, that check is still necessary but no longer enough to show if a session can be trusted. A genuine face may belong to someone reading from a script or acting on instructions from a fraudster.
Why liveness checks no longer suffice
The launch comes as more than half of all web traffic is estimated to be non-human. Generative AI has also reduced the effort needed to produce convincing deepfakes and carry out injection attacks. Veriff says businesses now need to separate people from bots running scripted attacks and to protect vulnerable users from coached verification attempts. They also need to confirm that an authorised, accountable individual is directing any AI agents acting on their behalf. For marketplaces, fintechs, and other high-trust platforms, the company frames proof of human presence at sign-up and at sensitive points in the user’s journey as a trust and safety requirement.
Commenting on the launch, Rob Brazier, Chief Product Officer at Veriff, noted that any process can now be automated and any person can be deepfaked or coached. Brazier added that Presence Verification looks beyond the face to signals from the device, the network, full video, and audio.
How the product works
Presence Verification combines more than 1,000 risk signals into a single automated decision. Its liveness and deepfake detection layer identifies screen replays, printouts, 3D masks, mirror selfies, and altered or synthetic media injected into a session. A coercion detection layer analyses full-session video and audio for signs of third-party coaching, off-camera influence, or forced compliance, and works in 98 languages. The Veriff Network crosslinks data from sessions worldwide to flag known fraud patterns, repeat offenders, bot farms, and emulated or simulated devices. Then, a context-aware decision engine merges these inputs into one explainable result.
Integration runs through Veriff’s iOS and Android SDKs. Customers start a session via REST API, end users record a short selfie video in the customer’s app, and Veriff returns a decision via webhook. The company reports coercion detection accuracy above 98% and funnel conversion above 90%. Geo Jolly, Group Product Manager at Veriff, cited an 84% improvement in detection rates in social engineering fraud sessions.
Payment fraud and agentic commerce among use cases
In financial services and buy now, pay later (BNPL), Veriff positions the tool as a way to confirm that the genuine account holder is present before funds move. It can also detect coaching linked to mule accounts or scam-driven loans and transfers. The UK shows the regulatory pressure involved. UK Finance data puts authorised push payment (APP) fraud losses at approximately EUR 689 million (GBP 576,4 million) in 2025.
Veriff also links the product to agentic commerce, as AI agents transacting on behalf of consumers expand. Presence Verification is designed to confirm that the person behind the agent is present at key moments, such as account setup or checkout. On marketplaces, the tool can verify sellers at onboarding and add a step-up check before payout changes or high-value listings.
Presence Verification is part of a broader set of autumn 2026 releases, including Unstructured Docs, Know Your Business, Database Verification, and a separate update to Identity Verification.
