La Loi sur l'IA de l'UE is a comprehensive legislative proposal introduced by the European Commission in April 2021, designed to regulate intelligence artificielle technologies within the European Union. The Act aims to promote safe and éthique practices while fostering innovation and ensuring that systèmes d'IA respectent les droits fondamentaux.
La loi sur l'IA de l'UE classe les systèmes d'IA en quatre niveaux de risque : risque inacceptable, risque élevé, risque limité, and risque minimal. Unacceptable risk AI systems, such as those that manipulate human behavior or are used for social scoring by governments, are banned outright. High-risk AI systems, which include applications in critical infrastructure, education, and application de la loi, are subject to stringent requirements regarding transparency, accountability, and human oversight.
For limited and minimal risk AI systems, the regulations are less strict, encouraging innovation while still promoting transparency and user awareness. Developers of AI systems must ensure compliance with the Act through risk assessments, gouvernance des données, and documentation practices.
The Act also establishes a European AI Board to oversee implementation and provide guidance, as well as a framework for cooperation among EU member states. With the EU AI Act, the European Union aims to set a global standard for Gouvernance de l'IA, balancing technological advancement with ethical considerations and public safety.