Artificially Intelligent Criminal Justice? A policy oriented critical plea.

This study aims to demonstrate that the use of AI is triggering a change in the criminal justice system and, more specifically, it examines what kinds of changes can be expected. There are at least two different paths. On the one hand, the use of AI could provoke a shift towards securitarian criminal law, where criminal law performs the primary task of social control and management of social interactions. This would result in a sort of hypertrophy of criminal law subordinated to other penal goals than rehabilitation: surveillance, control, fear reduction, management of risks, and incapacitation. On the other hand, conversely, the use of AI could generate a push towards social criminal law, more focused on the peculiarities of the considered offender. AI could be an instrument to improve the treatment, punishment, and ultimately, the rehabilitation of offenders by providing them with individualized rehabilitation treatment. To explore these scenarios, the research first provides an overview of the main current applications of AI systems in the criminal justice system, starting from those systems used in police investigations to the ones adopted in the prison context. Second, it analyzes the EU policy process that has recently led to the adoption of the EU AI Act. A first interpretation of the provisions concerning AI applications in law enforcement is developed, in this way highlighting how their applications have the potential to seriously jeopardize human rights, civic space, and the rule of law. Lastly, the study focuses, from a critical theoretical perspective, on the impact that AI applications allowed by the AI Act can have on the criminal justice system, and more broadly, on our society. Indeed, it underlines the risk that they can produce a drift toward digital authoritarianism. However, it also presents the potential for some AI systems to improve the fairness of sentencing and the effectiveness of prison rehabilitation.