General Purpose AI Model Obligations Under the EU AI Act
Who this applies to
This applies to
providers of general-purpose AI models (GPAI), including developers, distributors, and importers placing such models on the EU market or putting them into service.
Deployers using GPAI models in high-risk systems are also subject to downstream obligations. Scope is defined under
Articles 3(1), 3(44), and 51–55 of the AI Act.
What is required
- Draw up and maintain technical documentation (including training/validation data, model architecture, and risk assessment) per Article 53(1).
- Provide a sufficiently detailed summary of the content used for training (e.g., datasets, copyright compliance) under Article 53(2).
- Publish a policy to respect EU copyright law (including opt-out mechanisms for rights holders) as required by Article 53(3).
- Demonstrate compliance with transparency obligations (e.g., disclosing AI-generated content, model capabilities/limitations) per Article 52(1).
- Conduct and document a conformity assessment (self-assessment or third-party, depending on systemic risk classification) under Article 55(1).
- Register high-impact GPAI models (those with systemic risk) in the EU database per Article 54(1).
- Implement a post-market monitoring system to track and report serious incidents per Article 51(2).
Key deadlines
The primary deadline for this obligation is
August 2, 2026.
Enforcement patterns
AI Act enforcement begins August 2, 2026. No precedent currently exists. This page will be updated as enforcement cases emerge.
Cross-border considerations
Implementation references appear most frequently in
Greece (GR) and
Austria (AT) for
Article 51 (post-market monitoring), and in
Cyprus (CY) for
Article 53 (technical documentation). No jurisdiction-specific deviations from the AI Act’s harmonized rules are yet documented.