"It Manages 339 Trillion Won in National Assets"... Turns Out, the AI Employees Use Is Chinese [China Shock, K-Industry on the Brink | AI] (2)
80.2% cite data leaks as a concern over the use of Chinese AI
SW supply chain security risk emerges as a threat
Warning signs are flashing that China's technology is encroaching unchecked on the AI ecosystems of South Korea's public and financial institutions and IT companies. Beyond security vulnerabilities in the technology itself, the risk to SW supply chain security has emerged as a national threat, as control over data could slip overseas from the design stage.
At a recent parliamentary audit by the National Assembly's Strategy and Finance Committee, concerns were raised about the security complacency of the Korea Investment Corporation (KIC), a sovereign wealth fund that manages approximately KRW 339 trillion in national assets. According to documents obtained by Kang Min-kuk, a lawmaker from the People Power Party, all the core components of an AI assistant introduced for all KIC employees were found to be Chinese-made.
The language model, which handles tasks including summarizing and translating documents, preparing weekly reports, and searching for materials related to lawmakers' inquiries, was Alibaba's Chinese-made Qwen3. The embedding model used to search internal materials was BGE-M3, developed by the Beijing Academy of Artificial Intelligence (BAAI) in China. BAAI has already been placed on the U.S. Department of Commerce's export control list for supporting the modernization of the Chinese military.
Reliance on Chinese technology has also surfaced in government-backed projects. Naver Cloud was eliminated from the project to build an "independent AI foundation model," a government initiative led by the Ministry of Science and ICT to secure technological sovereignty. Naver Cloud was penalized after it emerged that the company had indirectly borrowed Qwen's model weights and other components in a way that did not align with the project's purpose.
The problem is that model selection has focused solely on performance and value for money, without basic security screening criteria such as the developer's country of origin or whether it is subject to sanctions. This creates a situation in which security concerns, such as the transfer of personal information overseas, are inevitable. In fact, in April 2025, it emerged that Chinese AI company DeepSeek had transferred personal information belonging to users in South Korea overseas without proper notice or consent. The Personal Information Protection Commission subsequently recommended that the company take corrective and improvement measures.
In a report, Rockbridge Network Korea said that data leakage was companies' biggest concern regarding the use of Chinese AI models, accounting for 80.2% of responses. Concerns about a lack of transparency in personal information processing followed at 67.7%. The use of Chinese products was not readily apparent on the surface. In a survey of 102 small and midsize companies, none said they used Chinese services for work. But when specific models were listed, five said they were using them.
Major countries around the world have begun reviewing their supply chains as a whole and updating their regulations. The U.S. Department of Homeland Security reportedly issued an order on October 19 (local time) to completely remove DeepSeek technology from federal civilian agencies' IT systems. The scope of the regulations has been expanded to track and fully block third-party services and software that use U.S. cloud platforms such as Microsoft or Amazon Web Services (AWS), or that indirectly incorporate DeepSeek models. The Japanese government has also recently introduced a certification system for the security of data centers and cloud services, effectively excluding Chinese products.
South Korea's industry is also calling on the government to provide safety guidelines for the use of AI models. Industry representatives say that because U.S. frontier models cost ten times as much, companies have little choice but to use Chinese models if they focus solely on value for money. They have also repeatedly emphasized the importance of the independent AI foundation model initiative. Some, however, say that the government may find it difficult to take action given the competing interests of different countries.
Hot Picks Today
"Two Spend 500,000 Won in a Flash, Yet Wait Two Hours?" ... 'Dreamland' Shunned by Kids and Foreign Tourists
- The Era of Getting 200 Million Won After 30 Years Is Over... A Japanese Bank's "Perpetual Department Head" Who Just Clocked In Cries Out
- Spain Hit by Its Worst Tragedy... Lee Kangin, Moreno and Real Madrid Join Soccer World in Mourning
- They Said, “Make It Your Home for Life” - Then Suddenly Kicked 200 Passengers Off... What Happened to the Around-the-World Cruise?
- "130 Rice Bowls Sold Out in a Day"... 'Unfiltered Face' Adds 8 Million Won to Night Market Sales: China's Bungeunwoo
Professor Lee Sang-geun of Korea University's Graduate School of Information Security said, "Code generation is essential for developing weapons and creating websites, and using Chinese models in this process could introduce vulnerabilities," adding, "The government needs a strategy to establish an independent AI foundation model while also supporting companies and institutions in building their own competitiveness."
© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.