The Opacity of Seeing: Algorithmic Welfare Surveillance and the Politics of Visibility

Abstract

Algorithmic governance is increasingly embedded in the practices of public authorities, often promoted as a means to enhance efficiency, optimise resources, and prevent benefit fraud. This paper examines how legitimacy is articulated, negotiated, and contested in relation to AI-based risk profiling in welfare governance, and what tensions become visible around the exercise of algorithmic classification and control. Empirically, the paper analyses a recent Swedish case in which the Social Insurance Agency deployed an AI-based risk-profiling system, prompting a supervisory investigation by the Swedish Authority for Privacy Protection. Drawing on formal correspondence exchanged during the supervisory process, the analysis approaches legitimacy through the interrelated conditions of accountability, transparency, and explainability. Five tensions expose asymmetries of visibility, responsibility, and scrutiny in algorithmic welfare surveillance. The paper contributes to surveillance studies and critical data studies by showing how technical, legal, and institutional conditions intersect to produce layered opacity, expanding the state’s capacity to classify and scrutinise individuals while the processes through which risk is produced become increasingly difficult to see.

This document is currently not available here.

Share

COinS