Taiwan's Ministry of Digital Affairs (MODA) has officially prohibited public sector workers from using DeepSeek, a recently launched Chinese artificial intelligence (AI) model. This decision, announced on Friday, 31 January, stems from growing concerns over potential data leaks and national security risks associated with the software.

DeepSeek introduced its R1 chatbot last month, boasting capabilities that rival top-tier AI models in the United States but at a significantly lower cost. Despite its advanced features, the Taiwanese government has determined that the use of DeepSeek poses a substantial threat to national information security, particularly within government agencies and critical infrastructure projects.

Scope and government justifications
MODA’s prohibition applies broadly to employees within central and local government agencies, public schools, state-owned enterprises, and semi-official organizations. Additionally, workers involved in critical infrastructure projects and government-endowed foundations are barred from using the AI model.

The decision aligns with Taiwan’s long-standing policy of restricting technology that may compromise national security. MODA cited the 2019 Executive Yuan regulation, "Principles on Restricting the Use of Products That Endanger National Cyber Security," as the legal foundation for the ban. This regulation explicitly prohibits government agencies from utilizing information and communications technology (ICT) products and services that could pose cybersecurity threats.

Broader global restrictions on DeepSeek
Taiwan’s move follows similar actions taken by other governments worldwide. In the United States, the Navy has strictly forbidden service members from engaging with DeepSeek in any form, highlighting concerns about the potential exploitation of sensitive military information.

In the United Kingdom, Secretary of State for Science, Innovation, and Technology Peter Kyle has initiated a national security risk assessment of the AI model, signaling possible future restrictions. Meanwhile, Italy’s data protection authority, Garante, has placed restrictions on DeepSeek’s use of Italian nationals’ personal data, demanding transparency from the company regarding its data collection practices. As a precautionary measure, DeepSeek has also been removed from Google and Apple’s app stores in Italy.

In Japan, Prime Minister Shigeru Ishiba has indicated that his administration is considering enacting a basic law on generative AI, potentially placing new regulatory barriers on the use of AI models like DeepSeek within the country.

On national cybersecurity
Taiwan’s decision to ban DeepSeek is part of its broader efforts to safeguard national cybersecurity and prevent foreign interference. Democratic Progressive Party (DPP) Legislator Wang Ting-yu emphasized that DeepSeek is just the latest in a series of Chinese applications, websites, technologies, and hardware that have been restricted from government use.