Mapping the required competences for the EU AI Act–mandated regulatory authorities in Lithuania

Problem

The EU AI Act, a common legal framework for artificial intelligence within the European Union, came into force in August, 2024. The purpose of this legislation is to promote human-centric and trustworthy AI while ensuring a high level of protection of health, safety and fundamental rights against the harmful effects of AI (EU AI Act, 2024).

Within the Act, AI systems are classified using a risk-based approach, following the logic that “the higher the risk, the stricter the rules”.  Some AI systems are deemed to pose unacceptable risk (e.g. social scoring, manipulation) and are thus prohibited while high-risk AI systems (e.g. systems used for critical infrastructure, education, etc.) are allowed but required to pass a conformity assessment to make sure they are compliant with the Act. Limited risk systems, such as chatbots, are only required to comply with transparency obligations and there are no requirements for minimal-risk systems; hence the regulatory focus falls on high-risk AI systems.

Conformity assessments of high-risk AI systems will be carried out by conformity assessment bodies, designated and monitored by at least one notifying authority (in Lithuania – The Innovation Agency) in each member state.  Compliance with the regulations will be monitored by at least one national market surveillance authority (in Lithuania – The Communications Regulatory Authority, others – to be confirmed) (EIMIN, 2025).

The EU AI Act requires that these regulatory institutions have administrative, technical, legal and scientific personnel who possess experience and knowledge relating to the conformity requirements for AI systems, such as risk management, data governance, technical documentation and others.

Challenges

• The designated regulatory authorities do not currently possess sufficient competences to effectively evaluate and monitor compliance with the conformity requirements for AI systems.
• There is a lack of understanding around what constitutes a high-risk AI system and which sectors in Lithuania create such systems; thus, it is unclear where to focus competence-building efforts.

Goal

The aims of this project are to 1) understand which sectors in Lithuania create high-risk AI systems; 2) assess the institutional competences of the regulatory authorities for these systems; 3) propose a plan on how to fill the identified gaps in competences.

Participants

2024/25

Agnė Keršytė