Technological developments have had the impact of creating Artificial Intelligence to create creative works in the form of images/paintings just by using commands in the form of text. The image creation process carried out by Artificial Intelligence relies heavily on creative works to be used as input data obtained from the database. The input data is then abstracted to produce new images/creative works. This research focuses on the process of creating Artificial Intelligence which produces new creative works in the form of images/paintings as well as forms of accountability for violations committed by Artificial Intelligence. This research was conducted using a juridical-normative legal methodology with a conceptual and statutory approach. Abstracting of copyrighted works that do not belong to him which is carried out without the permission of the copyright holder by Artificial Intelligence is used so that Generative Adversarial Networks (GAN) can recognize the object referred to by the user's command. Copyright violations committed by Artificial Intelligence occur as a result of the use of other people's input data without permission and this data is abstracted to produce new works. The form of liability resulting from copyright infringement committed by Artificial Intelligence is divided into 2, namely criminal in the aspect of violation of economic rights and civil compensation in the aspect of violation of moral rights.