The Ontological Shift in Surveillance: Revisiting the “Surveillant Assemblage” in the Age of Facial Recognition

Abstract

It has been twenty-five years since Haggerty and Ericson’s (2000) seminal article on the “surveillant assemblage” was published and since then reshaped surveillance studies, offering a conceptual framework that moved beyond Orwellian and Foucauldian metaphors towards the rhizomatic, hybrid qualities of emerging surveillance systems. Their theory remains profoundly influential. Yet the surveillance landscape has since undergone an ontological disruption, here exemplified by the proliferation of live facial recognition (LFR) in policing. These systems do not merely reassemble individuals into “data doubles” but operate through algorithmic approximation, assessing probabilistic resemblance rather than identity. The aim of our dialogue paper is to revisit the analytical core of the surveillant assemblage in light of this shift. Earlier regimes of surveillance were anchored in representation, constructing digital proxies that rendered individuals visible across institutional domains. By contrast, contemporary AI-driven infrastructures increasingly act through opaque probabilistic models that collapse recognition, suspicion, and intervention into automated inference. This shift is not only technical, but ontological, epistemological and political, reshaping how knowledge, risk, and personhood are enacted. By exposing the shift from representation to approximation, and from data doubles to biometric resemblance, we advance critical debates on the surveillant assemblage. If surveillance now approximates rather than recognises, predicts rather than represents, then what is at stake is nothing less than democratic accountability itself. Must we not, then, fundamentally rethink the very foundations of surveillance in the age of algorithmic governance?

Keywords:

• surveillant assemblage,
• facial recognition,
• data doubles,
• ontology,
• probabilities,
• algorithmic governance