Applying Self-recognition Biometrics to Live Deepfake Detection in Video Conferences

This work-in-progress research addresses the pressing issue of deepfake detection in video conferencing, proposing a novel self-recognition biometric method. It addresses the increasing sophistication of cyber threats, particularly deepfakes and other face presentation attacks, by developing a user-centric verification system. This system utilizes eye tracking to detect participants' self-recognition, effectively distinguishing between genuine users and fraudulent attackers. This contribution outlines the integration of this technology into a video conferencing prototype, offering an overview of its functionality and initial performance outcomes. Our preliminary findings suggest that this method could be a groundbreaking tool for enhancing digital communication security.