A senior official from the U.S. government claimed that the Chinese artificial intelligence (AI) startup DeepSeek has been supporting Chinese military and intelligence agencies.

The official added that DeepSeek circumvented U.S. regulations by mobilizing shell companies in Southeast Asia to secure high-performance AI semiconductors.

Despite strong semiconductor export controls from the U.S., DeepSeek has received attention in the industry for delivering outstanding performance. If this claim is true, there are analyses suggesting that illegally obtained U.S.-made chips were behind its technological growth.

DeepSeek logo./Courtesy of Yonhap News

Reuters reported on the 23rd, citing an anonymous senior U.S. State Department official, that "DeepSeek is contributing directly to the enhancement of military and intelligence capabilities on a national level in China beyond simple technological development."

The official noted, "There is a high possibility that DeepSeek will continue to willingly support Chinese military and intelligence operations," adding, "There are suspicions that user information and statistics are being shared with the national surveillance agency in Beijing."

This is the first time the U.S. government has publicly targeted a specific Chinese AI corporation in connection with military activities. Reuters assessed that the U.S.-China hegemony competition surrounding AI technology has entered a much more overt and dangerous phase.

In February 2025, the DeepSeek AI sign is visible on a startup office building in Beijing, China. /Courtesy of Yonhap News

According to Reuters, DeepSeek established shell companies in Southeast Asia to acquire NVIDIA's high-performance AI chip "H100" to evade U.S. semiconductor export controls. The H100 has been banned from export to mainland China and Macau since 2022 under U.S. export controls.

The official added, "There are indications that DeepSeek may have opted to use facilities in Southeast Asia instead of utilizing data centers in mainland China to access U.S. semiconductor chips remotely."

In January, DeepSeek unveiled its low-cost, high-performance AI inference model R1 in Hangzhou, China, shocking the industry. R1 has been evaluated as an innovative achievement realized through its own capabilities despite U.S. technological sanctions. Chinese media cited DeepSeek as a case symbolizing the rise of China's AI technology.

However, in reporting this allegation, Reuters analyzed that "there is a growing conviction (within the U.S. government) that DeepSeek has likely relied on U.S. AI chips and technology during its rapid growth" and added, "There are concerns that the capabilities possessed by DeepSeek may have been exaggerated."

Illustration=ChatGPT

DeepSeek also provides services through major cloud platforms such as Amazon Web Services (AWS) and Microsoft Azure in the U.S. This raises concerns that all user information using DeepSeek could potentially be utilized for Chinese government intelligence activities.

The Center for Strategic and International Studies (CSIS), a prominent U.S. think tank, recently warned directly, "The vast data collected by DeepSeek could be used to refine China's national surveillance system and to oppress ethnic minorities or support military decision-making."

The U.S. is poised to further intensify its pressure on DeepSeek. Italy has already blocked DeepSeek. In the U.S., both the Navy and NASA have prohibited the use of DeepSeek. In Europe, telecommunications authorities sensitive to data protection have begun a detailed investigation into DeepSeek's privacy policies.

The U.S. is likely to place DeepSeek on a sanctions list (blacklist) similar to Huawei in the future and exert pressure on domestic cloud corporations like AWS to immediately discontinue DeepSeek services.