Taiwan's government is pushing ahead with its own homegrown AI model, aiming to reduce reliance on English and Chinese-language systems built elsewhere.

• The National Science and Technology Council said its TAIDE AI system now has upgraded language and behavior models, with a new version to be released and opened to public and private use by the end of the year.
• The push responds to concerns that Taiwan has focused on hardware chips but fallen behind on large language models, especially ones that properly handle Traditional Chinese, Taiwanese Hokkien, and Hakka.
• The Ministry of Digital Affairs is building a sovereign AI language database using government data (excluding private or sensitive records) plus private-sector contributions that respect copyright; over 200 agencies have already supplied more than 3 billion tokens of data.
• A separate government-backed AI computing center project is moving forward with four companies, which will set aside 1-3% of their computing power for public use, such as schools.

Outlook: Expect the new TAIDE model to roll out by year-end, with Taiwan continuing to invest more heavily in building its own AI infrastructure and language data independent of foreign systems.