This page offers a summary of the EU AI Act’s enforcement rules under Chapter V, which set out the obligations imposed on providers of general-purpose AI (GPAI) models. It also seeks to examine the part that other actors can play in enforcing the AI Act. Under the Act, providers of GPAI models are subject to both procedural obligations (including interaction with the AI Office) and substantive ones (relating to model development and documentation). Although these obligations have applied since 2 August 2025, the Commission’s supervisory and enforcement powers over GPAI model providers will only take effect on 2 August 2026 (see the full implementation timeline here). These powers include the ability to request documentation and information, carry out evaluations, require remedial measures (covering compliance, risk mitigation, market restrictions, recalls, and withdrawals), and impose fines. In addition to the Commission, several other actors play a key role in ensuring effective enforcement of the AI Act against GPAI model providers. For example, national market surveillance authorities (MSAs) can ask the Commission to use its enforcement powers against GPAI model providers, downstream providers can file complaints against them, and the scientific panel can notify the AI Office of any systemic or specific identifiable risks arising from a GPAI model.
Coming up in this post:
Introduction Substantive obligations for providers of general-purpose AI models. **Procedural obligations for providers of GPAI models.** When do providers of GPAI models fall under the Act’s jurisdiction? Oversight and enforcement authority of the Commission. Les voies d’exécution.
EU Artificial Intelligence Act