Which authorities and bodies exist at EU level and what are they responsible for?
The AI Act provides for four authorities and bodies at EU level:
- European Artificial Intelligence Office ("AI Office");
- European Committee on Artificial Intelligence ("AI Board");
- Advisory Forum; and
- Scientific Panel of independent experts.
We will briefly outline how they are organised and what their respective tasks are:
What is the European Artificial Intelligence Office ("AI Office") and what does it do?
The AI Office was established by a Commission decision and is part of the Commission's Directorate-General for Communication Networks, Content and Technologies (DG CNCT). The tasks assigned will be carried out by civil servants already entrusted with these tasks. The tasks assigned to the AI Office can therefore already be performed.
The AI Office performs a wide range of tasks. These are laid down in the AI Act on the one hand and specified in the Commission decision on the other. In carrying out its tasks, the AI Office must cooperate with relevant stakeholders (science, AI developers, civil society, social partners, etc.), EU and national authorities and bodies.
The tasks include, among others:
- Monitoring the implementation and application of the AI Act in general. This includes in particular the following tasks:
- Assisting the Commission in the preparation of relevant Commission decisions, and implementing and delegated acts; preparing guidance and guidelines to support the practical implementation of the AI Act; preparing standardisation requests, evaluating existing standards and drafting common specifications;
- Contributing to the provision of technical support, advice and tools for the establishment and operation of AI regulatory sandboxes;
- Conducting evaluations and reviews and preparing reports
- coordinating the establishment of an effective governance system, including by preparing the set-up of advisory bodies at Union level;
- Providing the Secretariat for the AI Board and its subgroups, as well as providing administrative support to the Advisory Forum and the Scientific Panel;
- Encouraging and facilitating the drawing up of codes of practices and codes of conducts .
- Monitoring the implementation and application of the AI Act in relation to GPAI. This includes in particular the following tasks:
- Market surveillance authority for GPAI schemes where the model and scheme originate from the same provider (Art 75 AIA);
- Developing of tools, methods and benchmarks to assess the capabilities of GPAI, especially for very large GPAI with systemic risks;
- Monitoring the emergence of unforeseen risks stemming from GPAI models, including by responding to alerts from the Scientific Panel;
- Coordinating the supervision and enforcement of legislation for which the Commission has supervisory and enforcement powers (e.g. Digital Services Act or Digital Markets Act);
- Supporting the implementation of the rules on prohibited AI practices and high-risk AI systems in coordination with the relevant bodies responsible under sectoral legislation, including facilitating the exchange of information and collaboration between national authorities, collecting notifications and establishing information platforms and databases, in particular when a GPAI model or system is integrated into a high-risk AI system.
- Cooperation with stakeholders, monitoring the implementation and application of the AI Act;
- International cooperation with third countries and international organisations to ensure a contribution to a strategic, coherent and effective Union approach to AI in coordination with Member States and in line with Union positions and policies;
- Promoting cross-sectoral cooperation within the Commission;
- Supporting the development, deployment, roll-out and use of trustworthy AI systems and applications that bring societal and economic benefits and contribute to the Union's competitiveness and economic growth. In particular, the promotion of innovation ecosystems through cooperation with relevant public and private stakeholders and the startup community;
- Market monitoring of AI markets and technologies.
What is the European Artificial Intelligence Board (AI Board) and what does it do?
The AI Board is made up of one representative from each Member State, the European Data Protection Supervisor and the AI Office. The European Data Protection Supervisor is acting as an observer and the AI Office is not participating in votes. Other national and European authorities, bodies or experts may be invited for relevant topics. The meeting is chaired by a representative of a Member State.
Representatives are elected by the member states for three years and can be re-elected once. The following criteria must be considered when electing the representative:
- The representative must have the relevant national competences and power to contribute to the tasks of the AI Board;
- The representative must have the power to facilitate the implementation of this regulation in their Member State, for example by collecting relevant data;
- The representative must be designated as the single contact point between the Member State and the AI Board; where appropriate, the representative should also serve as a single contact point for stakeholders in their Member State.
The objectivity and impartiality of the AI Board must be guaranteed.
The AI Board has the following tasks, among others:
- Contributing to the coordination between the national competent authorities for the application of this Regulation and supporting the market surveillance authorities
- Cooperating with other Union institutions, bodies, offices and agencies and competent authorities of third countries and international organisations
- Collecting and exchanging technical and regulatory expertise and best practices among Member States;
- Advising on the implementation of this Regulation, in particular with regard to the enforcement of the rules on GPAI;
- Contributing to the harmonisation of administrative practices (conformity assessment procedures, regulatory sandboxes and real-life testing) in the Member States;
- Providing recommendations and written opinions on all relevant issues related to the implementation and application of this Regulation.
The AI Board consists of two standing sub-groups to provide cooperation and exchange among market surveillance authorities and the notifying authorities. Further standing or temporary sub-groups may be established to fulfil the tasks.
The AI Office shall provide the Secretariat for the AI Board.
The Advisory Forum should not be confused with the AI Advisory Board (often also referred to as the “KI-Beirat”) in Austria!
What is the Advisory Forum and what does it do?
The Advisory Forum serves as an advisory body on technical matters for the Commission and the AI Board and consists of stakeholders from industry, start-ups, the SME sector, academia, think tanks and society. The composition is determined by the Commission, which must ensure a balance between commercial and non-commercial interests.
The Fundamental Rights Agency, the European Union Agency for Cybersecurity (FRA), the European Union Agency for Cybersecurity (ENISA), the European Committee for Electrotechnical Standardisation (CENELEC) and the European Telecommunications Standards Institute (ETSI) are permanent members of the advisory body.
The members must have proven expertise in the field of artificial intelligence. The term of office is two years and can be extended by a maximum of four years. It meets at least twice a year and must draw up an annual activity report, which must be published.
The Advisory Forum is responsible for the preparation of:
- Opinions;
- Recommendations;
- Written contributions.
Standing or temporary sub-groups may be formed to fulfil these tasks.
The Advisory Forum should not be confused with the AI Advisory Board (often referred to as “KI-Beirat”) in Austria!
What is the Scientific Panel of Independent Experts and what does it do?
The Commission shall set up a Scientific Panel of independent experts by means of an implementing act to support the enforcement measures under this Regulation.
The Commission shall select the members on the basis of up-to-date scientific or technical expertise in the field of artificial intelligence necessary for the tasks assigned to the Scientific Panel. The number of members shall be determined by the Commission in consultation with the AI Board on the basis of the required demand. When making appointments, it takes into account a balanced gender and geographical representation.
Experts must fulfil the following criteria:
- Particular expertise and competence as well as scientific or technical expertise in the field of Artificial Intelligence;
- Independence from any provider of AI systems or GPAI models or systems;
- Ability to carry out the activities diligently, accurately and objectively.
The Scientific Panel provides the following advisory and support services:
- Implementation and enforcement of this Regulation, esp. in relation to GPAI models, including in particular the following activities:
- Alerting the AI Office of possible systemic risks associated with GPAI models at Union level;
- Contributing to the development of tools and methodologies to assess the capabilities of GPAI models, including through benchmarks;
- Advising on the classification of GPAI models with systemic risk;
- Advice on the classification of different GPAI models and systems;
- Contributing to the development of tools and templates.
- Supporting the work of market surveillance authorities at their request;
- Supporting cross-border market surveillance activities;
- Supporting the AI Office in carrying out its duties in the context of the "safeguard clause procedure" to Art 81 AIA.
The members of the Scientific Panel fulfil their tasks independently, impartially, and objectively. They guarantee the confidentiality of the information and data they receive in the performance of their duties and activities. Each member submits a declaration of interests, which is made publicly available.
The AI Office implements systems and procedures to actively manage and prevent potential conflicts of interest.