South Korea‘s spy agency has accused Chinese AI chatbot DeepSeek of excessive personal data collection and biased responses on cultural issues, as concerns mount globally over Chinese AI applications’ data practices.
The National Intelligence Service (NIS) warned government agencies last week about security risks associated with DeepSeek, citing the app’s unusual data collection practices, including keyboard input patterns and communication with Chinese servers, Reuters has learned.
This development comes as US lawmakers prepare bipartisan legislation to ban DeepSeek from government devices, mirroring previous actions taken against TikTok. Representatives Darin LaHood and Josh Gottheimer are spearheading the initiative, citing national security concerns.
The South Korean agency revealed that DeepSeek provides advertisers unrestricted access to user data and stores information on Chinese servers, potentially accessible to the Chinese government under local laws. The NIS also highlighted inconsistencies in DeepSeek’s responses to culturally sensitive questions, such as the origins of kimchi, providing different answers in Korean and Chinese languages.
Several South Korean government ministries have already blocked access to the application, joining Australia and Taiwan in implementing restrictions. Texas has become the first US state to ban DeepSeek on government devices, with Governor Greg Abbott citing concerns about Chinese Communist Party influence through data collection.
Cybersecurity experts have identified hidden code in the application that allegedly transmits user data to China Mobile, a state-owned company banned from US operations. Federal agencies including the Navy and NASA have already restricted the app’s use.
DeepSeek did not immediately respond to requests for comment. However, China’s foreign ministry previously stated that Beijing prioritizes data privacy and security, claiming it never requires companies to collect data illegally.