Wiki · Governance · Last reviewed June 25, 2026

AI Act Advisory Forum

The AI Act Advisory Forum is the European Union's stakeholder advisory body for AI Act implementation. It gives the European Commission and the AI Board technical and practical input from civil society, academia, industry, SMEs, start-ups, rights bodies, cybersecurity experts, and standardisation organizations.

Category: Governance Published: June 25, 2026 Modified: June 25, 2026 Last reviewed: June 25, 2026 Tags: EU AI Act, AI Board, AI Office, standards, stakeholder governance

Definition

The AI Act Advisory Forum is a general advisory body to the European Commission and the European Artificial Intelligence Board under Article 67 of the EU AI Act. The Commission describes it as supporting implementation and enforcement of the AI Act, while complementing the narrower AI Act Scientific Panel, which focuses on independent expert advice for general-purpose AI models and related systemic-risk questions.

As reviewed on June 25, 2026, the Commission's public page said 174 Advisory Forum members had been selected from more than 700 applications across civil society, academia, industry, SMEs, and start-ups. The same page said members provide technical expertise and advice on a broad range of AI Act questions, including standardisation and implementation challenges.

Article 67 Role

Article 67 says the forum provides technical expertise, advises the AI Board and Commission, and contributes to their tasks under the AI Act. It can prepare opinions, recommendations, and written contributions at the request of the Board or Commission, and it can establish standing or temporary subgroups to examine specific questions connected to the law's objectives.

The forum is therefore not a public comment box. It is a structured channel for stakeholder expertise during implementation. That matters because many AI Act duties depend on practical translation: high-risk classification, common specifications, harmonised standards, transparency code practice, sector examples, small-firm burden, public-sector use, rights impact, and cybersecurity evidence.

Membership and Balance

Article 67 requires a balanced selection of stakeholders, including industry, start-ups, SMEs, civil society, and academia. It also requires balance between commercial and non-commercial interests and, inside commercial interests, between SMEs and other undertakings. Members serve two-year terms, which may be extended, and the forum must meet at least twice a year.

Five bodies are permanent members under the legal text and Commission page: the Fundamental Rights Agency, ENISA, CEN, CENELEC, and ETSI. That design shows the forum's hybrid purpose. It is partly a rights and accountability forum, partly a cybersecurity and standards forum, and partly a place where implementation friction from industry and public-interest groups becomes visible before guidance hardens into administrative routine.

Implementation Work

The Commission reported that the forum's inaugural session took place on June 19, 2026. The meeting introduced rules of procedure and co-chair election organization, and the Commission sought initial input on standardisation-related questions, the Code of Practice on Transparency of AI-generated content, and draft guidelines on high-risk AI system classification.

Those topics show why the forum matters for AI governance. The law can say "high risk," "transparency," or "standardisation," but organizations need examples, templates, thresholds, technical routes, exceptions, and evidence expectations. The forum is one venue where those translations can be contested by actors who do not all share the same incentives.

Governance Use

For compliance work, the forum should be treated as a source of implementation context, not as a substitute for legal text. A serious governance record should track whether a claim rests on Article 67, an Advisory Forum opinion, a Commission guideline, a harmonised standard, a common specification, a code of practice, or a private vendor summary. Those artifacts do different work and carry different legal weight.

For civil society and affected communities, the forum creates a route to watch. Its annual reports, membership lists, subgroups, meeting records, and written contributions can reveal whether implementation is being shaped mostly by large providers, by small deployers, by public-interest organizations, by rights bodies, or by standards engineers.

Limits

The Advisory Forum does not enforce the AI Act, certify AI systems, investigate providers, issue fines, or replace national competent authorities. It advises the Commission and AI Board. It is also not the same as the Scientific Panel, which has a more technical enforcement-support role around GPAI models and systemic risk.

Stakeholder balance is a design goal, not an automatic result. Participation can still be shaped by resources, language, travel, technical capacity, regulatory familiarity, and the ability to turn lived harms into the document formats regulators can use. The forum is useful only if its process keeps disagreement visible rather than converting plural input into a bland consensus label.

Source Discipline

Use Article 67 for the forum's legal role, the Commission Advisory Forum page for current membership and institutional framing, Commission news pages for dated meetings, and the AI Office page for how the forum fits with the AI Board, Scientific Panel, and broader implementation work. Treat trade-association announcements and member press releases as evidence of participation, not authority for the forum's legal powers.

Spiralist Reading

Spiralism reads the Advisory Forum as a pressure valve in the implementation machine. It gives the AI Act a place where standards, rights, industry constraints, public-sector practice, and civil-society warnings can meet before the rules become forms and checkboxes. Its danger is capture by the most resourced voices. Its promise is procedural: making the negotiation over machine governance visible enough to contest.

Sources

Return to Wiki