Système d’IA

The EU AI Act (Regulation 2024/1689) defines an AI system as a machine-based system designed to operate with varying levels of autonomy and that may exhibit adaptiveness after deployment. It infers, from the inputs it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments.

This definition is intentionally aligned with the OECD's definition of AI systems to ensure international coherence. It covers a broad range of technologies including machine learning, deep learning, logic-based approaches, and statistical methods.

The classification of a system as an "AI system" under the Act is the threshold determination that triggers all subsequent regulatory obligations. If a system does not meet this definition, the AI Act does not apply to it. Conversely, once classified as an AI system, the provider must determine its risk category (unacceptable, high, limited, or minimal) and comply with the corresponding requirements.

Understanding this definition is critical for compliance teams, as it determines the scope of applicability and triggers obligations ranging from transparency requirements for limited-risk systems to comprehensive conformity assessments for high-risk systems.

• Règlement européen sur l’IA