Privatizing Sentencing: A Delegation Framework for Recidivism Risk Assessment

Abstract

This paper explores the use of privately developed risk assessment algorithms in criminal sentencing, arguing that these tools are developed in a way that hinders the enforcement of constitutional protections and gives private algorithm developers undue influence in sentencing determinations. Using the private delegation doctrine, which limits Congress's ability to delegate to private actors, the paper aims to strengthen the state statutory frameworks that govern the use of these tools to restore accountability to the sentencing process.

Keywords: risk assessment, sentencing, recidivism, machine learning, private delegation, delegation, State v. Loomis