Abstract

The growing use of facial recognition technology (FRT) by law enforcement in South Africa presents urgent legal, ethical, and governance challenges. While FRT is often promoted as a tool to enhance safety and crime prevention, its deployment reveals deep tensions between technological efficiency and constitutional rights to privacy, dignity, and equality. This article examines the operation of contemporary FRT systems – rooted in machine learning and high-dimensional data analysis – highlighting their opacity and potential error and bias. It further explores the phenomenon of function creep, where surveillance technologies expand beyond their original intent, often without public awareness or regulatory oversight. Drawing on South African case examples, including Vumacam’s CCTV network and biometric data collection under the Independent Communications Authority of South Africa’s (ICASA) proposals, the article situates these developments within comparative frameworks such as the European Union’s Artificial Intelligence Act. It argues that South Africa urgently requires a comprehensive, human rights-based regulatory framework to prescribe the specific circumstances and safeguards for the law enforcement use of FRT. Such a framework would ensure proportionality, transparency, and accountability, thereby preventing technological advancements from outpacing legal protections.