Punishing the Unpunishable: A Liability Framework for Artificial Intelligence Systems

Abstract

Artificial Intelligence (AI) systems are increasingly taking over the day-to-day activities of human beings as a part of the recent technological revolution that has been set into motion ever since we, as a species, started harnessing the potential these systems have to offer. Even though legal research on AI is not a new phenomenon, due to the increasing “legal injuries” arising out of commercialization of AI, the need for legal regime/framework for the legal accountability of these artificial entities has become a very pertinent issue that needs to be addressed seriously. This research paper shall investigate the possibility of attaching civil as well as criminal liability to AI systems by analysing whether mens rea can be attributed to AI entities and, if so, what could be the legal framework/model(s) for such proposed culpability. The paper acknowledges the limitations of the law in general and criminal law in particular when it comes to holding AI systems criminally responsible. The paper also discusses the legal framework/legal liability model(s) that could be employed for extending the culpability to AI entities and understanding what forms of “punishments” or sanctions would make sense for these entities.