Government Selects Naver Consortium to Develop Security-Specialized AI

The Naver Cloud consortium has won the South Korean government’s competition to build a cybersecurity-specialized AI foundation model, beating a consortium led by SK Telecom. The Ministry of Science and ICT announced the selection on the third day of the month.

The government will support Naver Cloud with 256 Nvidia B200 graphics processing units for 10 months. It will provide the first five months of support before deciding whether to fund the remaining period after an interim evaluation.

The project seeks to create a locally developed “AI shield” as cyber threats become more sophisticated. Generative AI and autonomous AI agents are being used to search for vulnerabilities, analyze attack routes and produce malicious code. The government also wants to address a weakness in general-purpose AI systems, which may have broad public security knowledge but lack sufficient data from real-world attacks and defenses in South Korea.

Naver Cloud plans to combine its experience developing large-scale AI models with malware, vulnerability and cyber threat intelligence data from security companies. The consortium also intends to develop models that can operate on closed networks and on-premises systems used in sectors such as nuclear power and finance.

The 32-member consortium includes LG CNS, LG AI Research, LG Uplus, security companies S2W, SANDS Lab and Theori Korea, the Financial Security Institute, Korea Hydro & Nuclear Power, Seoul National University and KAIST. The ministry praised its broad industry, academic and research structure, along with its use of multiple AI agents and a control and verification “harness.” It plans to release the results as open source and pursue model upgrades and wider industrial deployment from 2027.