European AI Office

What it is

The European AI Office is a body established inside the European Commission as the foundation for a single AI governance system ^[1]. Rather than a separate agency standing outside the EU executive, it works within the Commission itself, so that AI supervision sits alongside the rest of the institution's policy work. Its central task is to make the EU's AI Act function in practice.

Origins

The decision establishing the office entered into force on 21 February 2024 ^[2]. It was created as the organisational core of a single governance system, meaning that responsibility for AI oversight would be gathered in one place rather than spread across several bodies ^[1].

Supporting the AI Act

The office supports implementation of the AI Act, with particular attention to general-purpose AI models, and it also supports the governance bodies set up in the Member States ^[1]. General-purpose models are trained on broad data and can be adapted to many different tasks instead of one narrow one; because a single such model can end up inside a great many products, supervision is organised at the level of the model provider. For that layer the office is not only advisory: it enforces the rules for general-purpose AI models ^[1].

Powers

Under the AI Act, the Commission acting through the AI Office can carry out evaluations of general-purpose AI models, request information and measures from providers, and apply sanctions ^[1]. These three powers form a chain: the office can examine a model, ask its provider for material or for changes, and impose penalties where the rules are not met.

Structure and staff

The office employs more than 125 staff, among them technology specialists, administrative assistants, lawyers, policy specialists and economists ^[1]. It is organised into six units and two advisers, covering excellence in AI and robotics, regulation and compliance, AI safety, AI innovation and policy coordination, AI for societal good, and AI in health and life science ^[1]. The range of units shows that the remit goes beyond enforcement: it includes safety work, coordination of AI policy, and applications in health and the life sciences.

The AI Pact and the Code of Practice

The office oversees the AI Pact, which allows businesses to engage with the Commission and prepare for implementation of the AI Act ^[1]. This gives companies a route to work through the coming obligations with the Commission before they apply. On 10 July 2025 the office published the final version of the General-Purpose AI Code of Practice ^[3], a document setting out how providers of general-purpose models can demonstrate that they comply with the rules.

Place in the wider policy agenda

The office works alongside the Commission's broader AI policy packages: the AI Continent action plan, launched in April 2025, and the Apply AI Strategy, launched in October 2025 ^[1]. Its units on AI innovation and policy coordination, and on excellence in AI and robotics, reflect the same double role — enforcing the Act's obligations for general-purpose models while supporting the wider effort to build and adopt AI in Europe.

Taken together, the office concentrates in one body the tasks that usually sit apart: helping Member States apply a new law, testing the models the law covers, negotiating practice with industry, and feeding into the Commission's wider strategy. That combination is what makes it the reference point for EU AI governance.