AI Office
EU enforcement body within the European Commission overseeing AI Act obligations on GPAI models.
Last refreshed: 23 April 2026 · Appears in 1 active topic
The AI Office gains GPAI enforcement powers on 2 August with no guidance letters issued: is it ready to regulate the models it's supposed to oversee?
Timeline for AI Office
AI Office gains enforcement powers in August
European Tech SovereigntyGained supervisory authority over GPAI deployed inside VLOPs and VLSEs
European Tech Sovereignty: AI Omnibus deal splits enforcement into two speedsThree EU-US deadlines collide in 9 days
European Tech SovereigntyMentioned in: DMA orders Google to open search data
European Tech SovereigntyMentioned in: DMA clock starts toward six-day AI Act collision
European Tech Sovereignty- When does the EU AI Office start enforcing rules on AI companies?
- The AI Office has had enforcement powers since August 2025 but gains full GPAI enforcement powers on 2 August 2026. Models placed on the EU market before August 2025 have until August 2027 to comply.Source: European Commission
- What fines can the EU AI Office impose?
- For GPAI violations, the AI Office can fine providers €15m or 3% of global annual turnover, whichever is higher. For failing to supply information, the ceiling is 1%. For OpenAI, a single enforcement action could exceed €500m.Source: EU AI Act
- Has the EU AI Office issued any pre-enforcement guidance?
- As of 23 April 2026, fewer than 100 days before GPAI enforcement activation on 2 August, no pre-enforcement guidance letters from the AI Office have been publicly issued to model providers.Source: Lowdown
- What is the GPAI Code of Practice?
- The GPAI Code of Practice sets the compliance standards for general-purpose AI model providers under the EU AI Act. Companies that placed GPAI models on the EU market after August 2025 must already comply; those before that date have until August 2027.Source: European Commission
Background
The EU AI Office became operational in February 2024 within the European Commission's Directorate-General for Communications Networks, Content and Technology. Its core mandate is to enforce the AI Act's rules on General Purpose AI (GPAI) models — the class that includes large language models from OpenAI, Google DeepMind, Mistral AI, and Aleph Alpha. In August 2025 the Office gained enforcement powers allowing it to investigate, audit, and fine providers of GPAI models that fail to meet transparency, risk assessment, and incident-reporting obligations. Full enforcement powers activate on 2 August 2026, when the GPAI Code of Practice compliance window closes for models placed on the EU market after August 2025; models placed before that date have until August 2027.
A significant pre-enforcement gap has emerged. As of 23 April 2026 — fewer than 100 days before GPAI activation — no pre-enforcement guidance letters from the AI Office have been issued to model providers. The 2 August activation date also falls just six days after the DMA binding decision on Google's FRAND search-data obligation is expected on 27 July, creating a regulatory collision week with no published coordination between the two instruments. The AI Office has stated it will adopt a "collaborative, risk-based approach" initially; the fine ceiling for GPAI violations is €15 million or 3% of global annual turnover, whichever is higher — for OpenAI, a single action could exceed €500 million.
The AI Office coordinates between 27 national supervisory authorities and maintains a GPAI model registry. Its creation reflects the EU's bet that central enforcement of foundational model rules is more effective than delegating to national bodies. Critics — including European AI startups — argue compliance obligations disproportionately affect smaller European providers relative to US hyperscalers with compliance teams numbering in the hundreds. Mistral AI and Aleph Alpha have both argued the burden was calibrated for OpenAI's scale, not theirs.