TrustLLM Meeting in Iceland - JSC Contributes to Europe’s Multilingual AI Future
Reykjavík, Iceland – June 11-13, 2025
The JSC joined partners across Europe this month at the second annual consortium meeting of TrustLLM, a key EU initiative focused on building open, trustworthy, and multilingual large language models (LLMs). Hosted by the University of Iceland, the event brought together leading voices in AI research, language technology, and policy to shape a European approach to inclusive artificial intelligence.
Representing one of Europe’s leading high-performance computing institutions, the JSC team actively contributed to discussions on the technical underpinnings of multilingual LLM development, data infrastructure, and evaluation frameworks that promote trust and transparency in language models. As part of the project, JSC is in the lead for work package 6, which aims to provide the technical foundation for large-scale training within the European HPC environment. As part of this work package, JSC is pioneering advanced algorithms that significantly boost model capabilities while crucially reducing the computational footprint during both training and evaluation - laying the groundwork for scalable, sustainable AI in Europe.
The meeting began with a presentation from former Minister of Tourism, Trade and Culture Lilja Dögg Alfreðsdóttir, who reminded participants about the importance of supporting smaller languages, such as Icelandic, in digital tools. Iceland’s successful integration of its native language into platforms like ChatGPT served as a case study for what’s achievable when governments, researchers, and tech companies collaborate.
A major part of the agenda included presentations from PhD students across the consortium, addressing core challenges in multilingual modeling. These young researchers represent the future of European AI development, which is open, inclusive, and scientifically grounded.
Another highlight was the panel discussion featuring Morris Riedel and Jan Ebert of JSC, who joined experts from Almannarómur, Miðeind, TNO, and others to explore the implications of underrepresented languages in AI.
Throughout the three-day meeting, the JSC team engaged in technical workshops focused on alignment strategies, multilingual evaluation, data collection, and ethical training methods.
The event also provided opportunities for informal collaboration and exchange, enabling partners to strengthen relationships that are critical to open and coordinated progress in Europe’s AI ecosystem.
As TrustLLM continues its mission, the JSC remains dedicated to enabling ethical and high-impact AI by providing the computational foundations and interdisciplinary collaboration necessary for success.
Official Website: Home | TrustLLM
Contact at JSC: Stefan Kesselheim