Researchers Create 'Master Faces' to Bypass Facial Recognition

Source: Radhamely De Leon

This article describes how researchers generated nine "master key" fake faces that are able to impersonate almost half of the faces in a dataset powering three leading facial recognition systems. While the researchers want to use the master faces they generated to help protect facial recognition systems from such attacks, they also predicted that if their master faces were used in deepfakes, they would be able to bypass liveness detection (a test to determine if a biometric sample is real or fake).

Keywords: Computer Science , Deepfakes , Facial Recognition , Tech Ethics