Sweden Launches National AI Language Model Rooted in Swedish Values

Sweden Unveils AI Model Reflecting National Identity

The Swedish government has officially launched a significant initiative to develop a national artificial intelligence (AI) language model, specifically engineered to reflect Swedish values, history, culture, and traditions. The announcement was made at a press briefing on February 20, 2026, attended by Prime Minister Ulf Kristersson, Minister for Public Administration Erik Slottner, and Minister for Upper Secondary School, Higher Education and Research Lotta Edholm.

Prime Minister Kristersson emphasized the importance of a high-quality Swedish language model, stating that 'Language models are not just translated words, but they carry history, culture, traditions, values, everything that is embedded in a language.' This strategic move aims to ensure Sweden's digital sovereignty and foster national competence across the entire AI lifecycle.

Collaborative Development and Funding

The ambitious project is being spearheaded by the Wallenberg AI, Autonomous Systems and Software Program (WASP), Sweden's largest individual research program. WASP is establishing a new research arena, WARA AI Training and Inference Compute at Scale (WARA AI-TRICS), which will focus on developing this open, large-scale Swedish language model.

Funding for the initiative includes a substantial contribution of up to SEK 30 million from the Knut and Alice Wallenberg Foundation. Additionally, the project will leverage the immense computational power of the Berzelius supercomputer at Linköping University, also supported by the Wallenberg Foundations. The development is a collaborative effort, involving:

• Researchers from eight of Sweden's top universities
• Authors and book publishers
• News media organizations, including the Newspaper Publishers Association

These partners are co-funding the effort by contributing high-quality, editorially reviewed training data that embodies Swedish values and norms, with a clear commitment to safeguarding copyright.

Purpose and Ethical Considerations

The primary goal of the AI model is to not only process the Swedish language effectively but also to deeply understand Swedish contexts, ranging from government decisions to the education system. The long-term ambition is to create a 'safe, ethical, and sustainable solution of high quality for government agencies, the public sector, media companies, and other Swedish organizations,' a matter deemed of national importance.

This initiative also addresses the broader national AI strategy, which seeks to strengthen Sweden's welfare and competitiveness by harnessing AI's potential while tackling societal challenges, enhancing national security, and promoting a prosperous democratic society. By developing its own foundational language model, Sweden aims to reduce dependence on external, often foreign, private sector models for such a critical resource.

Training and Future Outlook

The AI model will be trained on extensive datasets, including a large amount of data from public sector organizations, alongside material from Swedish literature and Swedish news media. Most of the model's training is projected to occur throughout 2026.

This national undertaking positions Sweden at the forefront of developing AI technologies that are deeply integrated with national identity and ethical frameworks, aiming to serve as a model for responsible AI development globally.