Research Home Topics EU AI Act Countries Sources Calendar Methodology Authors

AI Office and AI Board Governance Structure Under the EU AI Act (Articles 64–70)

Four interlocking governance bodies — the AI Office, the European AI Board, the Advisory Forum, and the Scientific Panel — plus the national competent authorities they coordinate: who they are, what they do, and how they divide responsibility for implementing the Act.

AI Act — Value Chain Obligations
Provider
Develops, places on market
Art. 16
→
Importer
Places provider's system on EU market
Art. 23
→
Distributor
Makes available without modification
Art. 24
→
Deployer
Uses system in own processes
Art. 26
An institutional building bearing EU Member State flags alongside the EU flag, evoking the Board's composition of one representative per Member State.

Builds on AR-000200 (The EU AI Act Explained), AR-000270 (GPAI Obligations), and AR-000280 (Market Surveillance and Enforcement) — none of their content is re-derived here. This article details how the AI Office and its associated governance bodies are structured under Chapter VII, and how the Board and scientific panel feed into the enforcement mechanics those articles already cover.

Executive summary

What this means

Who this applies to

Regulatory context

The AI Office (Article 64)

The European AI Board (Articles 65–66)

The Advisory Forum (Article 67)

The Scientific Panel of Independent Experts (Article 68)

National competent authorities (Article 70)

Requirements

What organizations must do

  • Check your national competent authority’s published single-point-of-contact route — required to be public since 2 August 2025.
  • For GPAI obligations, note the scientific panel is the body that may issue “qualified alerts” on systemic risk to the AI Office under Article 90, feeding directly into Article 55 obligations.
  • Do not treat the Board as an enforcement contact — route enforcement-relevant matters to the AI Office (GPAI) or the national market surveillance authority (high-risk systems).

Common mistakes

Treating the AI Board as an enforcement body is a common mistake — the Board’s tasks are advisory, coordination-based, and opinion-issuing, not directly executive. The Commission (via the AI Office) and national market surveillance authorities are the enforcement actors.

Assuming there is a single "EU AI regulator" for all AI Act matters is a second mistake — GPAI supervision sits with the AI Office at Commission level, while high-risk AI system enforcement sits with national market surveillance authorities, per the jurisdictional rules AR-000280 covered in depth.

Overlooking the Advisory Forum’s five permanent members — FRA, ENISA, CEN, CENELEC, ETSI — is a third mistake: their permanent status gives those bodies a standing seat in the advisory process, not an occasional input.

Open questions

  • Whether Articles 64, 65, 66, 67, 68, or 70 are genuinely unamended by the Digital Omnibus. None carry an amendment disclaimer on their Service Desk pages (65–70 via full page fetch; 64 via URL-indexed snippet), but absence of a disclaimer is not confirmation.
  • Whether the Article 68(1) implementing act formally constituting the scientific panel has been adopted, and the panel’s membership and size determined. Not independently confirmed against a primary source on this pass.
  • Whether all Member States met the Article 70(2) deadline of 2 August 2025 to publish single-point-of-contact information. Not systematically verified against the Commission’s published list.

Frequently asked

FAQ
Did the Digital Omnibus amend any of Articles 64–70?
No disclaimer appeared on any of the pages checked (65–70 via full page fetch; 64 via URL-indexed snippet) — consistent with the pattern across much of Chapter VII, though not treated as confirmation they are definitively unamended.
FAQ
Has the Article 68(1) implementing act formally constituting the scientific panel been adopted?
Not independently confirmed against a primary source on this research pass.
FAQ
Have all Member States met the Article 70(2) deadline of 2 August 2025 to publish single-point-of-contact information?
Not systematically verified — the Commission’s published list of single points of contact exists but was not checked for completeness on this pass.

Go deeper

Apply this in practice

AGNUS Academy turns AI governance research into hands-on skills — exercises, case studies, and frameworks built on the research you just read.

Explore the Academy →