TAIPEI (Taiwan News) — Taiwan’s Ministry of Digital Affairs on Thursday said that government agencies are prohibited from using Chinese-made generative AI language models due to risks of information manipulation, per CNA.

Deputy Minister Hou Yi-hsiu (侯宜秀) delivered a report titled “Taiwan’s Future Competitiveness: How Can Taiwan Win the New Talent War in the AI Era?” during a legislative committee session and fielded questions from lawmakers.

Hou said the National Security Bureau evaluated five prominent Chinese AI platforms: DeepSeek, Doubao, Yiyan, Qwen, and Tencent’s Yuanbao. The tests found that content generated by these models contained biased and false information, including Chinese government messaging and historical inaccuracies.

She said MODA has issued cybersecurity and AI evaluation guidelines for government units to conduct their own testing, and the ministry will continue reminding the public and private sector of potential risks.

DPP Legislator Lai Jui-lung (賴瑞隆) pressed the ministry on whether it had fully assessed the cybersecurity threats posed by Chinese AI models. Lai also asked whether Taiwan would consider banning certain platforms altogether, noting that several countries have issued warnings or required their removal from app stores.

Hou said such measures fall within MODA’s legal authority. She reiterated that government agencies are already prohibited from using Chinese-made generative AI models and that the National Security Bureau’s evaluations are conducted under MODA’s established frameworks.

She added that MODA’s AI Product and System Evaluation Center continues to conduct related testing to ensure national cybersecurity standards are met.