Can programmed interfaces be punished? Criminal liability for decisions taken by artificial intelligence

Abstract

The anthropocentric view is the basis of traditional Criminal Law, which considers only the human being as subject of rights and holder of legal dignity. However, machine learning, a technology used in mathematics and computer science, raises the philosophical question of whether a robot can commit crimes and be held responsible for them. As more and more of people's lives become virtual, the development of algorithms to analyze information becomes increasingly complex and is used to read behavioral patterns. Nonetheless, The more data the algorithm receives, the more varied its response can be. The figurative criminal responsibility of artificial intelligence can be an interesting boundary for the causal nexus, preventing the programmer from being punished for acts decided by electronic neural connections. The lack of regulation can lead to serious consequences, suggesting then the adoption of compliance programs in the field of computer science.