AI Act General Procedural Framework
Follow topic LLM context A cited markdown file you can paste into your AI assistant (ChatGPT, Claude, a RAG or project knowledge base) to ground it in this topic. Contains: the overview, key law text, case law, enforcement and guidance for this topic. Everything links back to its source on overview.legal — legal information, not advice.The source document is specifically titled 'Procedure' from the AI Act, suggesting a comprehensive procedural section that warrants a dedicated topic covering the general procedural framework, mechanisms, and requirements applicable across the regulation.
Overview
Legal Framework
The general procedural framework for the AI Act is established primarily in Articles 63 through 71 of the Regulation. These articles create a cohesive procedural architecture for market surveillance, conformity assessment, and enforcement. The framework mandates that Member States designate national competent authorities and market surveillance authorities, outlining their powers and obligations. Key procedures include those for investigating AI systems presenting a risk, handling non-compliant systems, and facilitating cross-border cooperation through an information-sharing system (AI regulatory sandboxes and real-world testing are governed by separate procedural rules in Articles 53 and 57, respectively). Recital 60, while focused on a specific domain, underscores the procedural imperative for authorities to ensure the accuracy, non-discrimination, and transparency of AI systems, given their profound impact on fundamental rights.
Practical Application
The procedural framework is designed for a coordinated EU-wide enforcement model. In practice, this means national market surveillance authorities act as the primary front-line investigators, utilizing their powers under Article 65 to require information, conduct inspections, and test systems. A critical procedural mechanism is the mandatory consultation and cooperation process outlined in Article 66. Before adopting measures against a non-compliant AI system, an authority must consult with the provider and the relevant notified body, if applicable, and notify other Member States through the dedicated information system. This prevents fragmented enforcement and ensures a consistent regulatory approach across the single market.
Key Considerations
- Map Your Supervisory Landscape: Providers must identify the relevant market surveillance authority in each Member State where their high-risk AI system is placed on the market or put into service, as procedural actions will originate from these national bodies.
- Prepare for Investigative Powers: Organizations must have internal procedures to promptly respond to formal information requests and potential on-site inspections from authorities, including providing access to data and documentation as required under Article 65.
- Engage in Mandatory Consultation: If notified of potential non-compliance, actively engage in the Article 66 consultation procedure with the authority and your notified body; this is a formal opportunity to provide corrective input before enforcement measures are finalized.