Regulating algorithmic management: A blueprint

Abstract

The promise—and perils—of algorithmic management are increasingly recognised in the literature. How should regulators respond to the automation of the full range of traditional employer functions, from hiring workers through to firing them? This article identifies two key regulatory gaps—an exacerbation of privacy harms and information asymmetries, and a loss of human agency—and sets out a series of policy options designed to address these novel harms. Redlines (prohibitions), purpose limitations, and individual as well as collective information rights are designed to protect against harmfully invasive data practices; provisions for human involvement ‘in the loop’ (banning fully automated terminations), ‘after the loop’ (a right to meaningful review), ‘before the loop’ (information and consultation rights) and ‘above the loop’ (impact assessments) aim to restore human agency in the deployment and governance of algorithmic management systems.