European Commission to establish European AI Office for EU AI Act enforcement

The European Commission is preparing to establish the European Artificial Intelligence Office, which will be crucial in enforcing the AI Act. According to a draft document obtained by Euractiv, the AI Office will be integrated into the Commission with its separate budget line. The extent of the Office’s autonomy is still under discussion, with uncertainty as to whether it will have its political objectives or function as an extension of the unit responsible for the AI Act. The decision to establish the AI Office is expected to enter into force in February before formally adopting the EU’s AI law.

Why does it matter?

The AI Office will primarily support national authorities’ enforcement of AI rules. Its primary focus will be on policing General Purpose AI (GPAI) models and systems, considered the most potent types of AI. Recent advancements in computing power, data harvesting, and algorithm techniques have led to the development of robust GPAI models like OpenAI’s ChatGPT-4. The Office will develop methodologies and benchmarks to evaluate the capabilities of such GPAI models, as the AI Act already includes a tiered approach to GPAI models, distinguishing those with potential systemic risks. The AI Office will monitor the application of rules on GPAI models and address any unforeseen risks based on alerts from independent experts.

The new regulatory office will be able to investigate possible infringements of regulations related to GPAI systems and models. It would also collect complaints and alerts, issue document requests, conduct evaluations, and request enforcement measures to mitigate violations. Additionally, the AI Office will coordinate the enforcement of the AI Act on AI systems already covered under other EU legislation, such as social media’s recommender systems and search engines’ ranking algorithms.

One of the purposes of the AI Office will also be to support the preparation of secondary legislation implementing the AI Act, ensure the uniform application of regulations, and provide guidance and standardised protocols. It will establish a forum for cooperation with the open-source community to develop best practices for the safe development and use of open-source AI models and systems. The Office will also promote innovation ecosystems, collaborate with public and private entities, and monitor the progress of initiatives like GenAI4EU.

In terms of financing, the AI Office will face budgetary constraints. The Commission’s digital policy department, DG CNECT, will assign human resources, while the hiring of temporary staff and operational expenses will be financed through the redeployment of the budget from the Digital Europe Programme.