ID Tech

Video Injection Detection

Controls that detect injected, replayed, or pre-recorded video streams used to bypass selfie and liveness checks during remote identity verification.

Overview

Video injection detection focuses on identifying attempts to feed a verification system a manipulated or pre-recorded stream (for example, via virtual cameras, screen replays, or intermediaries) instead of a live capture from the user’s device camera.

Typical signals and controls

  • Capture pipeline integrity checks (e.g., detecting virtual camera sources).
  • Consistency checks across frames and metadata (timing, encoding artifacts).
  • Challenge-response or motion prompts combined with PAD/liveness.

Where it shows up

  • Remote onboarding / selfie identity verification
  • Account recovery re-verification
  • High-risk transaction step-up checks

References

Vendors using Video Injection Detection

InnovatricsMitek Systems

Latest Data Cards

  • Data Card

    Researchers detail an Android toolkit that feeds fake video into live KYC checks

    2026-08-28CC-BY-4.0video-injection-detectionpad

    Cybernews researchers documented an Android toolkit that feeds saved photographs, prerecorded video, or a remotely controlled stream into a verification app as though it came from the phone camera.

    • The toolkit injects prerecorded media or remote streams into the camera subsystem on rooted devices.
    • It can spoof hardware signatures, location data, and security profiles.
    • Researchers found no evidence of widespread production exploitation and recommend server-side dynamic liveness challenges and hardware nonce verification.
    Open card
  • Data Card

    GetReal adds continuous identity verification to deepfake detection platform

    2026-05-18CC-BY-4.0deepfake-detectionvideo-injection-detection

    GetReal Security added continuous identity verification to GetReal Protect, extending its live communications security platform from initial deepfake detection to ongoing participant verification.

    • The feature is designed to verify that a person who joins a voice or video session remains the same person throughout the call.
    • GetReal Protect combines multimodal deepfake detection, impersonation detection, continuous verification, and threat intelligence.
    • The capability targets live-session risks where fraud can begin after a one-time authentication event.
    Open card
  • Data Card

    Aware reports zero injection attack bypasses in BixeLab CEN/TS 18099 evaluation

    2026-05-13CC-BY-4.0padvideo-injection-detectionfacial-recognitionaware

    Aware said its Intelligent Liveness technology recorded zero successful injection attack bypasses across 900 BixeLab test scenarios aligned with CEN/TS 18099:2024.

    • The evaluation included 600 injection attack transactions across 10 instrument species and four attack methods.
    • The reported result included zero successful injection bypasses and a 0 percent bona fide presentation classification error rate.
    • CEN/TS 18099 focuses on internal injection attacks that bypass the camera sensor, a different threat class from traditional presentation attacks.
    Open card
See more data cards