European AI Office (within DG CONNECT, European Commission)
- Type
- Government body
- Place
- European Union — European Union (extraterritorial: applies to providers placing models on the EU market wherever established)
- Last checked
- 2026-09-22
- Next check due
- 2027-03-21
Ways to reach them
- AI Office general inquiries
- AI Office recruitment
- AI Act Service Desk (Single Information Platform)
- Commission consultations on AI Act implementing acts, guidelines and codes of practice
- Homepage
- Live membership list
This route is retained as reference only. Its verification is missing, more than 180 days old, or the contact window is not currently open; do not rely on it until it is re-checked.
- Public submission route
What it does
The AI Act's central enforcement body for general-purpose AI. Exclusive supervisory and enforcement competence over GPAI model providers. Can conduct inspections, require technical documentation and access to models to assess capabilities, request corrective action, RESTRICT THE PUBLIC AVAILABILITY OF A MODEL, and impose fines directly (up to 3% of global turnover or EUR 15m for GPAI provider breaches; up to 7% or EUR 35m for prohibited-practice breaches enforced elsewhere). Develops evaluation tools, methodologies and benchmarks. Serves as secretariat to the AI Board. Under the Digital Omnibus on AI its supervisory powers were expanded. GPAI obligations have applied since 2 August 2025; the AI Office's enforcement powers over GPAI became exercisable from 2 August 2026.
Its AI responsibilities
The EU's central AI regulator and the enforcement body for the AI Act's rules on general-purpose AI models. Three functions: (1) AI Act implementation and enforcement, including 'developing tools, methodologies and benchmarks for evaluating capabilities' of GPAI models and investigating possible violations; (2) fostering trustworthy AI, including sandboxes and real-world testing; (3) international AI governance coordination. Over 125 staff across 6 units plus 2 adviser posts.
Its powers
fine
Honest assessment
The AI Office is the most consequential AI-safety regulator on earth by formal power — it is the only body that can order a frontier model off the market and fine its provider — and it publishes a working email and a genuinely staffed multilingual service desk. But the service desk answers COMPLIANCE questions; it is not a complaints channel and not a policy channel. The real input levers are (1) the multi-stakeholder drafting processes for codes of practice and guidelines, which were genuinely open and where civil-society drafting comments visibly changed text, and (2) the Have your say portal when an implementing act is out for feedback. Systemic-risk GPAI is explicitly within the AI Office's mandate, making this the ONLY jurisdiction where frontier risk is a statutory supervisory duty rather than a research interest.
Current business
GPAI model oversight, AI Act guidelines (including on high-risk classification and transparency obligations), the Code of Practice process, and implementation of the changes made by the Digital Omnibus on AI.
Public route
Four concrete routes: (1) general enquiries to CNECT-AIOFFICE@ec.europa.eu; (2) the AI Act Service Desk for regulatory questions; (3) the Expression of Interest form for collaboration, which is how outside experts get onto drafting and consultation groups; (4) respond to AI Office consultations on draft guidelines, which are announced on the Shaping Europe's Digital Future consultations page. The Advisory Forum is the formal civil-society seat.
Format rules
Consultations on draft AI Act guidelines are typically 4–8 weeks and run through structured online questionnaires. Responses may be published. Recent examples VERIFIED on the consultations page: 'Targeted consultation on the draft guidelines for the classification of high-risk artificial intelligence systems' (19 May – 23 July 2026); 'Consultation on the draft guidelines on transparency obligations under the AI Act' (8 May – 3 June 2026); 'European Commission survey: AI in healthcare and pharmaceuticals' (2 – 26 June 2026). All three have closed; the page is the live index to whatever is open now.
Key people
Governance bodies that carry outside voices into the AI Office: the AI Board (Member State representatives); the Scientific Panel of independent experts; the Advisory Forum, which is explicitly designed for balanced stakeholder representation including civil society; the AI Pact for business; and the Apply AI Alliance.
Notes
CRITICAL 2026 UPDATE: the Digital Omnibus on AI — Regulation (EU) 2026/1744 — entered into force on 27 July 2026. It DEFERRED high-risk obligations (stand-alone Annex III systems to 2 December 2027; Annex I embedded systems to 2 August 2028) but did NOT defer the GPAI/systemic-risk regime, which has applied since 2 August 2025, and did NOT defer the Article 50 transparency obligations, which apply from 2 August 2026. So the frontier-model part of the AI Act is fully live; the consumer high-risk part is delayed.
Research note
VERIFIED 2026-09-22 from the Commission's own AI Office page and consultations index. The head of the AI Office is NOT named on the official page and I am not guessing at it. NOTE: the three AI consultations I could verify have all closed — as at 22 Sep 2026 I found no open AI consultation, but the index page is dynamic and should be checked directly.