America’s legal system is already grappling with cases where chatbots acted as lawyers and witnesses, generated the evidence, or helped plan a mass shooting. Who will be liable now that some AI agents are doing crimes all on their own?