Whose [Crime] is it Anyway?

Abstract The concurrent increase in states’ cyber-capabilities and in the potential (mis)uses of artificial intelligence (AI) to further harmful state objectives raise the question: is ICL prepared to grapple with AI-powered cyberattacks conducted by and against states? Or does AI’s conduct run the risk of falling into a legal loophole, shielding from liability the human actors who commissioned its use? Addressing these questions, this article analyses how AI impacts ICL’s fundamental elements of international crimes — namely, the requirements of actus reus and mens rea. Specifically examining the literature’s more recently explored notion of the crime of ‘cyber-aggression’, this article scrutinizes the efficacy of the current text and interpretation of the Rome Statute in according international criminal responsibility where AI causes an act of cyber-aggression. It examines the spectrum on which potential acts of cyber-aggression may be fairly ‘reduced to human activity’, and thus where criminal responsibility may flow to the AI’s human instructor, versus where the act committed was perhaps too removed from human intervention to justify criminal responsibility under the applicable culpable mental state. In this article, the author argues the elements for criminal responsibility of an act of aggression, in their current construction and interpretation, are ill-prepared when faced with the murky implications of cyber-aggression conducted via state-backed AI systems. In order to engage in such a discussion, this article proceeds by providing an overview of the crime of aggression and introducing the emerging concept of ‘cyber-aggression’. Then, the article explores the challenges AI-powered acts of aggression present for the conventional elements of international criminal responsibility. Finally, the article details solutions posed by the literature and proposes an additional approach, before offering concluding remarks.