Taiwan’s Ministry of Digital Affairs (MODA) plans to release the initial version of its “sovereign AI language database” in the fourth quarter of 2025.

The database will use a training corpus based on licensing terms created by MODA to address copyright concerns.

It will primarily consist of open government datasets, policy reports, and government publications.

Only about 1,000 of the over 50,000 open datasets are suitable for language model training, according to MODA’s Chuang Ming-fen.

Agencies including the Hakka Affairs Council, Ministry of Education, Council of Indigenous Peoples, and Ministry of Culture are reviewing their language data for inclusion.

Access to the database will be open to both public and private sectors.

🔗 Source: Focus Taiwan

🧠 Food for thought

1️⃣ Taiwan’s initiative addresses critical language representation gaps in AI

Taiwan’s sovereign AI language database initiative is tackling a well-documented global problem: the extreme underrepresentation of non-English languages in AI systems.

Despite over 7,000 languages existing worldwide, most AI models are trained primarily on English data, creating significant linguistic disparities in technology development 1.

The Ministry’s focus on including agencies like the Hakka Affairs Council and Council of Indigenous Peoples directly addresses the preservation of linguistically diverse communities, similar to initiatives like FLAIR, which works to revitalize Indigenous languages using AI 2.

This approach aligns with global recognition that linguistic diversity in AI is not just a cultural concern but an economic one. The World Economic Forum has identified inclusive AI technologies as potential drivers for economic growth, particularly for underrepresented communities 3.

Taiwan’s initiative joins similar efforts worldwide, such as AI4Bharat and Masakhane, which are creating AI tools tailored to specific linguistic communities while engaging local populations to develop culturally nuanced datasets 4.

2️⃣ “Sovereign AI” reflects growing national data autonomy movements

Taiwan’s sovereign language database initiative mirrors a global trend where countries are establishing independent AI capabilities that reflect their specific cultural and linguistic contexts.