European Union - Establish AI Office

Overview

The Commission Decision of 24 January 2024 establishes the European Artificial Intelligence Office (AI Office) as part of the administrative structure of the Directorate-General for Communication Networks, Content and Technology (DG CONNECT). The Decision was published in the Official Journal (OJ C/2024/1459) on 14 February 2024 and entered into force on 21 February 2024. The Office is tasked to accelerate preparation for the implementation and enforcement of the EU Artificial Intelligence Act and to act as a Union-level focal point for monitoring, evaluation, investigation and international cooperation. The Decision underscores that the Office will operate in line with Commission internal procedures and must not affect the powers of national competent authorities nor duplicate relevant activities of existing Union bodies and agencies. See the full text at EUR-Lex (C/2024/1459).

Definitions

Key terms used by the Decision follow the AI Act framing and Commission practice: 'Office' (the European Artificial Intelligence Office); 'Directorate-General' (DG CONNECT); 'forthcoming Regulation' (the EU Artificial Intelligence Act as referred to in the Decision); 'general-purpose AI models' (models with broad applicability, including very large models); 'scientific panel' (an independent panel referenced in the Decision to provide technical alerts and risk signals). The Decision also cross-references sectoral legislation (e.g. the Digital Services Act and Digital Markets Act) when describing coordination responsibilities.

Governance and Institutional Framework

The Office is established within DG CONNECT and is subject to its annual management plan and the Commission's internal processes. Structurally, the Office provides secretariat and administrative support to the AI Board and its subgroups and may establish advisory bodies at Union level. The Decision explicitly prevents the Office from superseding national competent authorities: supervision powers under national or sectoral law remain for those authorities, while the AI Office coordinates, assists and supports the Commission’s supervisory role. Collaboration with relevant Commission DGs, agencies and Union bodies (including the EuroHPC Joint Undertaking) is mandated. For the Commission’s public explanation and download of the Decision see Shaping Europe’s digital future and the Official Journal at EUR-Lex.

Key Focus Areas

The Decision sets out multiple focus areas: (1) development of tools, methodologies and benchmarks to assess capabilities of general-purpose AI models (notably very large models with systemic risk potential) including procedures for evaluation and testing; (2) continuous monitoring of application and compliance of rules for general-purpose AI models, and monitoring the emergence of unforeseen risks by responding to scientific panel alerts; (3) investigative functions including the collection of complaints and alerts, assistance in preparing Commission decisions and evaluations, and conducting targeted assessments; (4) coordination and supervision where AI systems fall within other Union supervisory frameworks such as the Digital Services Act or Digital Markets Act; (5) support for implementation of rules on prohibited AI practices and high-risk AI systems, including facilitating information exchange between national authorities and establishing information platforms and databases; (6) standardisation work, preparation of common specifications and support for regulatory sandboxes; and (7) stakeholder engagement including establishment of fora and consultation mechanisms. These activities are designed to be technical, advisory and facilitative, while leaving binding enforcement actions to designated authorities and Commission powers under sectoral legislation.

Implementation Framework

Operationally, the Decision outlines resource arrangements and procedural responsibilities. Officials will be resourced from staff assigned within DG CONNECT, supplemented by external statutory staff funded through redeployment from the Digital Europe Programme’s administrative support appropriations. Operational costs are to be met from the Digital Europe Programme (Specific Objective 2 ‘Artificial Intelligence’). The Office is tasked with preparing Commission implementing and delegated acts, guidance, and standardised protocols; producing templates or documentation to enable uniform application of the AI Act; and providing technical support for national and Union-level sandboxes. Administrative set-up for advisory bodies, secretariat duties for the AI Board, meeting organisation and document preparation are specified as core administrative functions. The Decision’s financing and staffing provisions can be reviewed in the Official Journal text at EUR-Lex.

Monitoring and Evaluation

The Office must monitor markets and technological developments, track implementation of AI-related rules at Union and national levels, collect notifications and complaints, and prepare evaluations and reports regarding the forthcoming Regulation’s application. It will respond to alerts from the scientific panel and undertake evaluations of general-purpose AI models and systems, with particular emphasis on uncovering unforeseen risks. The Office is also charged with coordinating the establishment of an effective governance system across the Union, monitoring the creation of national authorities and facilitating information exchange, databases and platforms to improve oversight and enforcement coordination.

Penalties, Liability, and Appeals

The Decision itself is an internal administrative measure and does not introduce new civil or criminal penalties. It establishes a coordinating and technical body to support implementation and enforcement of the AI Act and other sectoral legislation; any sanctions, administrative fines or liability regimes remain those defined in the applicable primary legislation (notably the AI Act, the Digital Services Act and sectoral rules). The Office will support investigations and the preparation of Commission decisions but will not unilaterally impose penalties. Procedures for appeals, liability and redress are governed by the underlying substantive Union and national rules where enforcement action is taken.

Relationship to Other Instruments

The Decision explicitly ties the Office’s role to the forthcoming Regulation laying down harmonised rules on AI (the EU AI Act) and cross-references other Union instruments, such as the Digital Services Act and the Digital Markets Act. It also links the Office’s resourcing and operations to the Digital Europe Programme. The Office will assist with standardisation requests, coordinate on sandboxes and provide secretariat services for the AI Board, ensuring that guidance and technical outputs are compatible with sectoral supervisory regimes and existing Union agencies and offices.

International Alignment

The Office has an explicit international vocation: it will contribute to the Union’s strategic, coherent and effective approach to international initiatives on AI in coordination with Member States and Union positions. The Decision references international instruments and principles (for example the G7 Code of Conduct and Guiding Principles for developers of advanced AI systems) and mandates the Office to contribute to implementation and dialogue with international partners, supporting the EU’s global policy objectives on trustworthy AI.

Implementation Timeline

Sources and References