The Executive Yuan of the Republic of China, led by Premier Cho Rong-tai, announced today that the Deepseek AI service has been fully banned in government agencies to ensure national cybersecurity in consideration of preventing internal information breaches.
According to a press release by the Executive Yuan, concerns were raised regarding the DeepSeek AI service products released by the “Hangzhou Deepseek Artificial Intelligence Foundation Technology Research Limited Company” in China in January. The spokesperson for the Executive Yuan, Li Hui-chih, stated that during a policy meeting chaired by Premier Cho Rong-tai today, discussions were held regarding the potential cybersecurity risks associated with this service.
Li Hui-chih emphasized that under the principle of digital justice, all applications related to generative AI should be developed in compliance with legal frameworks and open-mindedness. The government has observed violations of copyright laws in the data sourcing of the DeepSeek AI service and potential biases in model language training due to ideological censorship and constraints.
Premier Cho Rong-tai stated that the cybersecurity risks of “data uploading to China” associated with using the DeepSeek AI service have not been clarified yet and cannot be ruled out. There are concerns about violations of personal data protection and privacy infringement. In view of national cybersecurity considerations, the Ministry of Digital Development has issued a special warning to government agencies and critical infrastructure to restrict the use of DeepSeek AI products to prevent the transmission of user-related data or information through such products with cybersecurity concerns, thereby posing a risk to national cybersecurity.
Li Hui-chih reiterated that since 2019, the Executive Yuan has implemented the principle of restricting the use of products that endanger national cybersecurity across central and local government agencies, public schools, state-owned enterprises, administrative entities, as well as venues for public events provided by self-operated or outsourced entities. Therefore, Premier Cho Rong-tai has specifically requested the complete prohibition of the DeepSeek AI service in government agencies.
Moreover, Li Hui-chih pointed out that the Ministry of Digital Development has indicated that public universities and research institutions may use the service if necessary, subject to approval through the stipulated procedures. Central government agencies overseeing key infrastructure providers and government-subsidized corporations should supervise compliance with the National Cybersecurity Management Law to ensure national cybersecurity.
Li Hui-chih mentioned that Vice Premier of the Executive Yuan Li-li Chun stated that under the policy direction of Smart Nation 2.0, Taiwan needs to further develop AI application technologies to enhance innovative services in various fields while remaining vigilant about potential cybersecurity risks associated with these services.
Premier Cho Rong-tai highlighted that Taiwan is a rule of law country, and the administration of public institutions in accordance with the law is a fundamental principle of the rule of law. To expedite the implementation of the Smart Nation 2.0 policy, he requested government commissioners Lin Ming-hsin, Wu Cheng-wen, and Liu Ching-ching to accelerate the reinforcement and supplementation of AI legislation, including data governance regulations, to provide guidance for government agencies and private sector entities in advancing the AI application industry further.