The AI Act and its impact on Large Language Models and the CLARIN Infrastructure

Abstract

The EU Artificial Intelligence Act (Regulation 2024/1689), which entered into force in August 2024, represents the world’s first comprehensive regulatory framework for AI. While it explicitly excludes AI systems developed solely for scientific research and development, its implications for the CLARIN community are far-reaching, especially given the growing reliance on large language models (LLMs) and general-purpose AI (GPAI). This paper discusses the scope of research exemptions in the AI Act (Section 2), and provides an overview of the obligations related to transparency of AI-generated text outputs (Section 3), as well as those imposed on providers of General-Purpose AI models, including those with systemic risk (Section 4). Given CLARIN’s role as a data provider, special attention is paid to the requirement for dataset documentation.