Defense Companies Also Block Access to 'DeepSeek' [Yang Nakgyu's Defence Club]

Proactive Response to Online Security Incidents Despite Network Separation
Some Companies Began Blocking Generative AI Since Last Year

Domestic defense companies in South Korea have begun blocking access to the Chinese generative artificial intelligence (AI) service ‘DeepSeek.’ While defense companies generally maintain that their internal computer networks are separated and therefore secure, this additional measure aims to prevent various security and information leakage incidents, including potential confidential data leaks.

On the 6th, SNT Dynamics and SNT Motive, subsidiaries of the SNT Group, announced through an internal notice that “there is a concern that DeepSeek extensively collects user data and leaks it to China,” and stated that “access to DeepSeek will be blocked within the company starting from 1 p.m. today.” They also requested employees to refrain from using DeepSeek not only on company PCs but also on personal communication devices, advising caution to ensure that internal business information or technical data is not entered when using personal communication devices. Hyundai Rotem and STX Engine also issued internal notices on the same day to inform about the block and urge caution.

Some defense companies have already completely blocked generative AI on their internal computer networks. Hanwha Aerospace, Hanwha Systems, and Korea Aerospace Industries (KAI) have prohibited the installation of generative AI on company computers since last year.

This is due to the continuous occurrence of security incidents such as hacking despite the internal network separation system. ‘Network separation’ mainly refers to ‘physical network separation,’ which uses two PCs connected separately to the internal and external networks.

In April last year, a North Korean hacking group stole data from more than ten domestic defense companies. They exfiltrated data through malware infections and email account hacking, and most of the affected companies reportedly were unaware of these breaches.

Earlier, on the 3rd, the Ministry of the Interior and Safety and the National Intelligence Service sent security guidelines to all central government ministries and metropolitan local governments, advising not to input sensitive information when using generative AI services such as DeepSeek and OpenAI. Additionally, the Personal Information Protection Commission sent an inquiry letter to DeepSeek’s headquarters requesting confirmation on personal information, data collection items and procedures, as well as processing and storage methods, but has not yet received a response.

Subsequently, domestic security-related central administrative agencies such as the Ministry of Foreign Affairs, Ministry of National Defense, and Ministry of Unification have also started blocking access to DeepSeek. Although no specific cases of damage have been confirmed yet, this is due to numerous technical concerns raised both domestically and internationally regarding DeepSeek.