“ALGORITHMIC BIAS AND THE LIMITS OF U.S. LAW: LESSONS FROM THE EU AI ACT’S RISK-BASED FRAMEWORK”

Abstract

This article examines the growing problem of algorithmic bias in the United States and argues that the current sector-specific legal framework is insufficient to address the discriminatory harms produced by AI systems. It highlights how automated tools used in employment, lending, housing, healthcare, and criminal justice often replicate or intensify social inequalities, while U.S. laws such as Title VII, ECOA, and the Fair Housing Act provide only reactive, post-hoc remedies. The article contrasts this fragmented model with the European Union’s AI Act, which introduces a comprehensive, risk-based regulatory structure requiring pre-deployment audits, transparency, human oversight, and bias testing for high-risk AI systems. By comparing the reactive U.S. approach with the EU’s preventive model, the article concludes that the United States could benefit from adopting hybrid reforms that integrate risk assessments and mandatory bias mitigation into existing civil-rights statutes. Such an approach would strengthen accountability, reduce algorithmic discrimination, and create a more future-proof legal environment for the governance of automated decision-making technologies.