AI Immutability and Discrimination: The Legal Status of Algorithmic Groups Beyond Protected Classes

Abstract

Artificial Intelligence (AI) systems increasingly structure social, economic, and legal relations through automated decision-making that often reproduces and amplifies preexisting inequalities. Conventional anti-discrimination frameworks in U.S. constitutional and statutory law are built around human categories—race, gender, religion, or national origin anchored in the idea of immutability. Yet, algorithmic classifications generate new, non-human groupings defined by data correlations, predictive inferences, and proxy variables that operate beyond traditional protected classes. This paper examines the constitutional and doctrinal challenges posed by these “algorithmic groups,” arguing that their emergent immutability—rooted not in biology but in code and statistical fixity—necessitates an evolution of equal protection jurisprudence. Drawing on legal theory, administrative law, and computational fairness research, the paper explores the limits of existing due process and equal protection doctrines and proposes a normative framework for recognizing algorithmic discrimination as a constitutional concern. The analysis situates algorithmic immutability within broader debates on due process, accountability, and the rule of law, offering pathways for reforming anti-discrimination and administrative adjudication mechanisms in an algorithmic state.