As an increasing number of organisations begin to use facial recognition technologies (FRTs), concerns have mounted over the potential risks the technology may pose to privacy and other civil liberties. At all levels of government, policy-makers have begun to propose new rules and regulations to govern the use of FRTs and manage any risks. It is important that these efforts be grounded in fact about how the technology works, since much public discussion so far has been muddled by exaggerations about the technology’s performance, misunderstandings about the details of its operation, and conflation of different types of systems.

This paper provides an examination of how the technology works and how to understand questions about its performance and operation.