Prohibited artificial intelligence systems under artificial intelligence act

While artificial intelligence has a long history, it can be mentioned that it has gained popularity today as AI technologies have penetrated many areas of our lives. Over the years, developments have occurred that have either increased or decreased interest in AI research area. While research funding was available during periods of increased interest in the field, when the promised results were not achieved, state aid was cut, leading to a halt in research. The development of new AI systems and their successes have led to a resurgence of interest. Consequently, periods when interest in the field waned due to failures have been characterized as AI winter, while periods when research regained momentum have been characterized as AI summer. Artificial intelligence, at the center of the digitalization process, is more than just a technical concept. Within this scope, it can generate content, make predictions, and make decisions. It utilizes data and algorithms to achieve these results. It processes numerous data sets, which are impossible to process with traditional data processing methods, through algorithms and produces output data in line with predefined and implicit objectives. Algorithms process numerous datasets that cannot be processed using traditional data processing methods, yielding output data based on predefined criteria or objectives. For artificial intelligence systems that generate data through algorithms to produce high-quality outputs, the accuracy and free of any sort of prejudice are of paramount importance. This study aims to explain the Artificial Intelligence Act, the first binding regulation in the world regarding artificial intelligence, and the European Union's perspective on the risks posed by artificial intelligence. The Act defines systems with high, limited, and minimum risks in general; systems with unacceptable risks are presented with examples. The thesis consists of three sections: the first covers the history of artificial intelligence, closely related concepts, and prominent documents of international organizations; the second addresses the goals the European Union aims to achieve with the Artificial Intelligence Act, the preparatory and legislative process of the Act, and the risks posed by artificial intelligence systems in the context of human rights, as well as the risk classes included in the Act; and the third section explains systems in the unacceptable risk class in detail with examples. The conditions and exceptions regarding the prohibition of artificial intelligence systems classified as unacceptable risk in the Act are clarified, and aspects deemed deficient from a legislative technique are also presented.