[Edaily Reporter Han Kwangbeom ] Big data analytics and artificial intelligence (AI) company #S2W Inc. announced on the 14th that, as a member of the Naver Cloud consortium, it will participate in the “Specialized AI Foundation Model Development (Cybersecurity) Project” organized by the Ministry of Science and ICT and the National IT Industry Promotion Agency (NIPA).

The AI Foundation Model Development (Cybersecurity) Project was launched as AI has recently emerged as a core technology that not only increases the speed and sophistication of cyber threats but also enhances detection and response capabilities.

This initiative was driven by the growing importance of security-specialized AI technologies and model capabilities, as AI has advanced to the point where it can directly perform vulnerability scanning, attacks, and defenses—as exemplified by Anthropic’s “Claude Mythos.” Consequently, the government has launched this project based on domestically developed technologies and models to accelerate the AI transformation (AX) of the domestic security industry and internalize the capabilities for developing security-specialized AI models.

The Naver Cloud consortium is led by Naver Cloud as the lead organization and comprises leading domestic companies such as S2W, LG Corp., and LG AI Research; key national infrastructure organizations including KOREA AEROSPACE INDUSTRIES (KAI) and Korea Hydro & Nuclear Power; and major universities and research institutions such as Seoul National University and the Korea Advanced Institute of Science and Technology (KAIST).

Based on Naver Cloud’s “HyperCLOVA X” and LG Corp. AI Research’s “EXAONE,” the consortium plans to develop, validate, and refine a hyper-large, security-specialized model featuring a Mixed Expert (MoE) architecture optimized for the domestic environment—with a total of 700 billion parameters (700B)—to contribute to strengthening Korea’s competitiveness in security AI.

In this project, S2W will leverage the technical capabilities it has accumulated in the fields of threat intelligence (TI) and security-specialized AI. Drawing on its experience in building cybersecurity-specialized language models and its expertise in collecting and analyzing vast amounts of threat data to transform it into intelligence, S2W will participate in key development processes such as model training and validation. In particular, the company plans to focus on enhancing the model’s completeness and practical applicability by incorporating hidden channel data—which is difficult to obtain from the Surface Web.

Seo Sang-deok, CEO of S2W, stated, “It is a great honor to participate in a national-level project aimed at enhancing competitiveness in security AI,” adding, “By contributing to the establishment of a proprietary technological foundation, we will help respond more effectively to increasingly sophisticated cyber threats and strengthen security sovereignty.”