• South Korea’s NIS advised government agencies to block DeepSeek, a Chinese AI chatbot, over data security and espionage concerns.
• DeepSeek stores user data on Chinese servers and provides biased responses to sensitive questions, raising privacy issues.
• The NIS warned that DeepSeek’s invasive data collection methods could expose sensitive government information to the Chinese government.
• Australia, Taiwan and Italy have also banned DeepSeek from government devices, citing national security risks.
• The global trend of caution towards Chinese AI technology reflects growing concerns over privacy and sovereignty.

South Korea’s National Intelligence Service (NIS) has advised government agencies to block access to DeepSeek, a Chinese artificial intelligence (AI) chatbot, due to concerns over excessive data harvesting and potential Chinese espionage.

The move, which took effect this week, follows a security bulletin from the NIS that detailed DeepSeek’s practices, which include storing user data on Chinese servers and providing biased responses to sensitive questions.

The decision to block DeepSeek comes at a time when South Korea, along with other countries like Australia, Taiwan and Italy, is increasingly wary of the security risks posed by Chinese technology. The NIS warned that DeepSeek’s data practices could expose sensitive government information to the Chinese government, which has a legal right to access data stored within its borders.

Excessive data harvesting and biased responses

According to the NIS, DeepSeek’s data collection methods are more invasive than those of other AI services. The agency stated that DeepSeek “includes a function to collect keyboard input patterns that can identify individuals and communicate with Chinese companies’ servers such as volceapplog.com.” This capability, combined with the app’s terms and conditions, which allow for indefinite data retention and unrestricted access by third-party advertisers, has raised significant privacy concerns.

One of the most alarming aspects of DeepSeek’s behavior is its biased responses to questions about sensitive topics. For example, when asked about the origin of kimchi, a traditional Korean dish, DeepSeek provided different answers based on the language of the query. In Korean, it acknowledged kimchi as a Korean dish, but when asked in Chinese, it claimed the dish originated from China. This discrepancy is not isolated; the NIS also noted that DeepSeek’s responses to questions about the Northeast Project, a Chinese research initiative that claims ancient Korean kingdoms as Chinese territory, were heavily influenced by Chinese Communist Party (CCP) propaganda.

Global concerns and restrictions

South Korea is not alone in its concerns. Australia and Taiwan have also banned DeepSeek from government devices, citing national security risks. Italy’s privacy watchdog ordered a nationwide block on DeepSeek, giving the company 20 days to explain how it complies with European data protection laws. The United States, including agencies like NASA and the U.S. Navy, has also restricted the use of DeepSeek due to security and privacy concerns.

These actions reflect a growing global trend of caution towards Chinese AI technology. The U.S. has imposed strict export controls on advanced chips and chipmaking equipment to China, aiming to constrain its AI development. However, DeepSeek’s emergence as a low-cost, high-performance alternative to U.S. AI models has shaken investor confidence and raised questions about the future of the global AI competition.

The decision to block DeepSeek on South Korean government computers underscores the country’s commitment to protecting its citizens’ data and national security. As other countries follow suit, the international community is sending a clear message to China: the global AI landscape will not be dominated by technology that compromises privacy and sovereignty. The rise of DeepSeek has not only sparked a technological arms race but has also intensified the geopolitical tensions between the East and the West.

Sources for this article include:

Breitbart.com

Fortune.com

Reuters.com

France24.com