Intelligenza artificiale, human oversight e responsabilità penale: prove d’impatto a livello europeo

Artificial intelligence (AI) and criminal law are now an inseparable pair in the scholarly discourse. It is indeed undisputed that AI serves as a strength test for traditional notions of substantive criminal law, such as mens rea and actus reus. Hence, the question that criminal legal scholars are doomed to ask themselves – in the aftermath of yet another advance in technology – seems to be a recurring one: if something goes wrong with these complex systems, should criminal law care? If yes, how? This paper, after a brief digression on the meaning of accountability, responsibility, and liability, will focus on the concept of human oversight and its relationship with the criminal notion of negligence. Particular attention will be paid to the European actor, which is emerging as a key normative player in this field, specifically to the contents of the European Parliament resolution of October 2021 “Artificial Intelligence in Criminal Law” and the proposed European regulation “AI Act”