Abstract

The article analyzes the guidelines issued by the European Commission on 6 February 2025, regarding the definition of artificial intelligence system pursuant to the AI Act. The authors compare the content of the AI system definition with the content of the aforementioned guidelines, while also referring to the existing doctrinal perspectives on the interpretation of this definition. The guidelines have been scrutinized from the standpoint of individual criteria constituting AI system definitions. The authors identify two key sections of the guidelines as particularly significant. The first one regards unequivocal determination that the premise of a system’s adaptability after its implementation is merely optional. The second one refers to the examples of systems provided by the Commission that do not fulfill the inference premise and hence are excluded from AI system definitions. These examples will substantially influence the practical application of the AI Act.