A privacy threat model for identity verification based on facial recognition

Abstract

The proliferation of different types of photographic and video cameras makes it relatively simple and non-intrusive to acquire facial fingerprints with sufficient quality to perform individuals’ identity verification. In most democratic societies, a debate has been occurring regarding using such techniques in different application domains. Discussions usually revolve around the tradeoffs between utility (security in access control, mobile phone unlocking, payment processing, etc.), usability or economic gain and risks to citizens’ rights and freedoms (privacy) or ethics. This paper identifies the common aspects of different solutions for identity verification based on facial recognition techniques within different application domains. It then performs a privacy threat modelling based on these common aspects to identify the most critical risk factors and a minimum set of safeguards to be considered for their management.