Naver Selected to Lead South Korea’s Cybersecurity AI Project

Naver Cloud has been chosen to lead South Korea’s effort to build an AI system focused specifically on cybersecurity, with the company already using thousands of high-performance graphics processors to prepare the model.

The Ministry of Science and ICT selected the Naver Cloud consortium after evaluating two applicants, Naver Cloud and SK Telecom. The consortium includes LG CNS, LG AI Research, LG Uplus, Logpresso, S2W and East Security, as well as KAIST, Korea University, Seoul National University, Chung-Ang University and POSTECH. Thirty-two institutions will take part.

Before the formal launch, Naver Cloud had begun pretraining the security-focused model with 4,000 Nvidia GPUs it had secured. Following its selection, the project will also use 256 GPUs supplied by the government, 4,000 of Naver’s B200 high-performance GPUs and 256 H200 GPUs from LG.

The consortium plans to develop a massive 700-billion-parameter mixture-of-experts model using Naver’s HyperCLOVA X and LG’s EXAONE. It will build two independent models in parallel, strengthening HyperCLOVA X for defensive capabilities and EXAONE for offensive capabilities. The stated aim is to respond across the full cycle of cyberattacks and defense.

The resulting model is planned for release as open-source software that can be used commercially throughout South Korea’s security ecosystem. The project also plans to provide small businesses with free checks for vulnerabilities and cyberattacks, while helping domestic security companies commercialize AI services through Naver Cloud Marketplace. Naver Cloud chief executive Kim Yu-won said the company would work with “a sense of responsibility that we are creating, with our own technology, an AI system that protects the nation’s core infrastructure.”

“

With a sense of responsibility that we are creating, with our own technology, an AI system that protects the nation’s core infrastructure.