A Legal Framework for Determining The Criminal Liability And Punishment For Artificial Intelligence

Abstract

Today, Artificial Intelligence (AI) is being used increasingly in every field. It has immense positive effects but every coin has two sides. AI is being used in criminal activities as well. Many countries are using AI-powered lethal autonomous weapons. These weapons decide their target and make the decision to kill someone without human intervention. Same way, a self-driven autonomous car may meet with an accident and kill a pedestrian. Who will be liable in both of the above cases? The developer, user, owner, supervisor of AI, or AI itself?  For holding a person criminally liable two elements must be present namely, ‘actus reus’ – a wrongful action and ‘mens rea’- a guilty mind or intention for the crime.  Can AI have mens rea? Can AI be punished just like a natural person? The article answers such novel issues with plausible arguments. Currently, there are no laws for punishing AI anywhere in the world but this is a dangerous situation where many crimes will go unpunished and it will create an atmosphere of insecurity in the minds of people. To deal with this situation, the article describes three different models that any nation can adopt in its criminal justice system for holding or not holding an AI criminally liable. Even if AI is held criminally liable how to actually punish an AI is a challenging task. The article further explains how the punishment of the death penalty, imprisonment, fine, etc can be extended to an AI. It also explains how the object of punishment can be served by punishing AI by explaining various theories of punishment in the context of AI. A parallel is drawn in the article with corporate criminal liability. Corporates cannot think but still can be held criminally liable but AI can think so why can't AI be held liable for crimes? A critical analysis of punishing AI is done in the article by discussing the pros and cons of punishing AI. The author has done a cost-benefit analysis of punishing AI vis-a-vis other remedies for AI-generated crimes.