In February 2025, the European Commission (the “Commission”) published two important instruments related to the interpretation of certain provisions of the AI Act, namely: (i) Guidelines on the definition of an artificial intelligence system (“Guidelines on AI System”), and (ii) Guidelines on prohibited artificial intelligence practices (“Guidelines on Prohibited AI”). Both Guidelines relate to the rules of the AI Act that have started applying since 2 February 2025.  

The Guidelines on AI System clarify the purpose and each of the seven elements of the definition of an AI system under the AI Act. These Guidelines should facilitate the analysis and decision whether an AI system falls within or remains out of the scope and rules of the AI Act.  

The Guidelines on Prohibited AI provide helpful insights into the Commission’s interpretation of the AI practice prohibitions under Article 5 of AI Act. Specifically, the Guidelines on Prohibited AI: 

• offer extensive and important interpretation regarding the substantive and personal scope of the AI Act clarifying the meaning of key terms such as “placing on the market,” “putting into service” and “use” of an AI system as well as “providers” and “deployers”; 

• provide an overview of the practices excluded from the AI Act’s scope, such as the use of AI systems for national security, defense, and military purposes as well as those released under free and open-source licenses; and 

• deliver a comprehensive analysis of each prohibited AI practice. 

Although the two sets of Guidelines are non-binding, they are valuable to businesses and consultants in their struggle with AI Act compliance.  

For further context, you may refer to DGKV AI Governance Newsletter in English at https://www.dgkv.com/news/The-EU-Artificial-Intelligence-Act-was-published-in-the-Official-Journal-of-the-European-Union-ENG; and in Bulgarian at https://www.dgkv.com/news/The-EU-Artificial-Intelligence-Act-was-published-in-the-Official-Journal-of-the-European-Union-BG.