OpenEuroLLM Develops Open-Source Models for EU Digital Sovereignty

The European Union is advancing its digital sovereignty through the OpenEuroLLM initiative, aimed at developing open-source language models to counter the dominance of AI technology from Silicon Valley and China. This initiative, led by Jan Hajič and Peter Sarlin, focuses on creating transparent and fully open models, ensuring compliance with EU regulations and fostering innovation within European businesses. The project, funded by the EU’s Digital Europe Program, brings together 20 leading research institutions and companies to develop multilingual models tailored for European needs.

The consortium’s primary challenge lies in securing sufficient computing capacity, as Europe lacks a centralized facility comparable to those in the U.S. or China. However, the EuroHPC network and the AI Factories program provide substantial resources to overcome this hurdle. Additionally, creating multilingual models that serve all EU member states equally is a significant task, given the scarcity of data for low-resource languages.

The initiative also aims to address the competitive landscape shaped by China’s DeepSeek and U.S. giants like OpenAI and Google. By fostering an open, collaborative approach, OpenEuroLLM seeks to empower European businesses with customizable AI solutions while maintaining cultural and linguistic diversity. This project aligns with the EU’s AI Act, promoting innovation towards trustworthy AI.

As Europe navigates the evolving AI landscape, OpenEuroLLM represents a strategic effort to enhance the continent’s digital sovereignty. The consortium’s work will test existing AI regulations, with potential collaboration with the European Commission to amend any obstructive rules. This initiative highlights Europe’s commitment to leading in the global AI competition, ensuring its technological independence and competitiveness.

Source: EU supports AI challenge to Silicon Valley and China