Intelligent and machine learning systems have infiltrated cyber-physical systems and smart cities withtechnologies such as IOT, image processing, robotics, speech recognition, self-driving, and predictive maintenance.To gain user trust, such systems must be transparent and explicable. Regulations are required to control crimesassociated with these technologies. Such regulations and legislations depend on the severity of the AI crimes subjectto these regulations, and on whether humans and/or intelligent systems are responsible for committing such crimes,and therefore can benefit from a classification tree of AI crimes. The aim of this paper to review prior work in ethicsfor AI, and classify AI crimes by producing a classification tree to assist in AI crime investigation and regulation.