Code of Justice: Comparing US and EU Regulation of AI Use in Criminal Sentencing

Abstract

The United States' legal framework concerning predictive policing and recidivism risk assessment tools is underdeveloped, prompting recent federal actions aimed at safeguarding fairness and impartiality in the criminal justice system. This paper assesses the current regulatory landscape in the United States regarding AI usage in sentencing decisions, comparing proposed and recently passed U.S. legislation to the European Union's Artificial Intelligence Act. Academic research highlights concern about biased data training AI models and their potential impact on sentencing fairness. Additionally, legal challenges to algorithm-based risk assessment underscore the need for transparency and accountability in AI's role in sentencing decisions. Through a comparative analysis of U.S. and EU legislative efforts, this study explores key themes such as explainability, accountability, and safeguards against discriminatory practices in AI-driven criminal justice systems.