National artificial intelligence (AI) governance: a cohesive model of supervision and innovation advancement for Lithuania

Problem

European context 

The EU Artificial Intelligence Act (the Act) aims to regulate and govern the rapidly evolving AI technologies within the EU and beyond. It acknowledges the significant opportunities AI offers across various sectors but also addresses growing concerns regarding unregulated AI’s potential risks to fundamental rights and freedoms (Publyon, 2024). 

The Act, which will come into force in May, 2024, unifies AI regulation across the 27 EU Member States and applies to all AI systems impacting people in the EU, irrespective of their development or deployment location. A well-known example of that would be the ChatGPT model produced by OpenAI. Compliance obligations for these systems are determined by the level of risk the AI system poses, with a focus on safety, security, and fundamental rights (EY, 2024). 

Key features include a tiered compliance framework targeting prohibited, high-risk AI systems, and those posing systemic risks, with enforcement starting from late 2024/early 2025 and extending to nearly all AI systems by mid-2027. Significant financial penalties are in place for noncompliance (EY, 2024). 

The Act aims to achieve four main objectives (EY, 2024): 

• Ensuring AI systems placed on the EU market are safe and respect fundamental rights. 
• Providing legal certainty to foster investment and innovation in AI. 
• Enhancing governance and enforcement of EU law on fundamental rights and safety requirements for AI systems. 
• Facilitating the development of a single market for lawful, safe, and trustworthy AI applications, thereby preventing market fragmentation. 

Lithuania’s challenge

As some of these responsibilities are to be ensured by the EU Member States, and Lithuania’s breakthrough in the field of AI is seen as essential for a prosperous future, Lithuania faces a twofold problem: 

• First, under the EU’s AI Act, Member States must guarantee the supervision of businesses’ compliance with this regulation by setting up or designating at least one market surveillance authority and at least one notifying authority. The respective responsibilities of these authorities are currently not yet assigned. The assignment of functions and inter-institutional and inter-sectoral cooperation needs to be efficient at the governmental level and intuitive for businesses to perform their conformity assessment and business activities in Lithuania. 
• Second, in the public domain, one often hears that Lithuania is aiming for a breakthrough in the AI field, but experts often note that we do not have the capacity to invest in all areas of AI and that we should focus on prioritized solutions. However, the current governmental model for catalyzing an AI breakthrough is different: the promotion of AI innovation is fragmented, there is no institutionalized model for national priorities in AI, and the cooperation between different sectors and different public institutions is not vertically embedded. 

Goal

The aim of the project is to prepare a comprehensive regulatory and innovation promotion model for AI which would facilitate the implementation of the provisions of the EU AI regulation and leverage the opportunities it brings to ensure breakthroughs in the field of AI in Lithuania. 

Participants

2023/24

Benediktas Girdvainis