Naver Cloud is preloading 4,000 graphics processing units (GPUs) into the government’s cybersecurity-specialised AI foundation model development and moving to develop a 700B-class large-scale security AI model.

Naver Cloud said on Sept. 3 its consortium has been selected as the final operator of the government’s “Specialised AI Foundation Model Development (Cybersecurity Field)” project and has begun full-scale development.

The consortium began pretraining a security-specialised AI model by deploying 4,000 B200 GPUs held by Naver Cloud even before the project officially starts. Separately from the 256 GPUs provided by the government, it will also add 256 H200 GPUs from LG.

Based on this, it will develop two 700B-class security-specialised models with a mixture-of-experts (MoE) structure using Naver Cloud’s HyperCLOVA X and LG AI Research’s EXAONE.

The HyperCLOVA X-based model is structured to strengthen defensive capabilities, while the EXAONE-based model strengthens offensive capabilities, to respond across the full cycle of cyber attack and defence. The models will be verified using 2 internationally accredited benchmarks.

Training will use about 830TB of real data. It will use data secured from national infrastructure and major industries, including operating data from Naver Cloud and LG CNS as well as data from KEPCO KDN, Korea Hydro & Nuclear Power, the Financial Security Institute, the Korea Institute of Science and Technology Information (KISTI) and LG Uplus.

The models will be tested in 7 areas: power, finance, science and technology, telecommunications, semiconductors, defence industry and aerospace. It will also develop closed-network operation technology so the models can be used in secure environments where access to external networks is restricted.

A total of 33 industry, academic, research and public institutions will participate in the consortium. It plans to release the models as open source for commercial use and to support free vulnerability and cyberattack detection checks for small and medium-sized companies, as well as commercialisation of security AI services through the Naver Cloud marketplace.

Kim Yuwon (김유원), chief executive of Naver Cloud, said, “With a sense of responsibility to create, with our own technology, AI that protects the nation’s core infrastructure, we will carry out this project.”