SANDS Lab wins KRW 10 billion R&D project for cybersecurity-specific generative AI

SANDS Lab wins KRW 10 billion R&D project for cybersecurity-specific generative AI

• Joint development and implementation with LG Uplus, KAIST, 42Maru and Logpresso
• AI and cybersecurity both benefit from government investment in three game-changing and six core strategic technologies
• Microsoft collaboration on developed technologies aims to strengthen core digital competitiveness

Photo 1. From top to bottom: lead organization SANDS Lab and joint R&D partners LG Uplus, 42Maru, Logpresso and KAIST.

AI-based cyber threat intelligence specialist SANDS Lab (KOSDAQ 411080, CEO Kim Ki-hong) announced on the 3rd that it won a KRW 10 billion core information security technology development project led by the Ministry of Science and ICT and IITP.

The government project develops next-generation generative AI technologies and services for cyber threat response through retrieval-augmented generation (RAG)-based small generative language models (sLLMs) specialized in four core areas: vulnerability analysis, anomaly detection, security monitoring and threat intelligence. Its goal is a cybersecurity-specific LLM and a public-facing demonstration platform to curb harmful uses of AI.

R&D will run for four years from April 2024, with KRW 7 billion in government funding and a total budget of KRW 10 billion. SANDS Lab will oversee development and form a consortium with LG Uplus, KAIST, 42Maru and Logpresso for joint implementation.

Photo 2. Conceptual architecture for four core security services using RAG and sLLMs, and technologies preventing harmful AI uses

SANDS Lab plans to develop on-premises sLLM technology for internal enterprise security; RAG-based information retrieval and analysis; conversational security services; AI applications for core security domains; detection and response to harmful generative AI uses; and an integrated platform that connects heterogeneous systems.

Once the specialized generative AI and on-premises LLM services are developed, customized organizational solutions are expected to strengthen internal data protection and privacy, maximize analysis and detection accuracy, shorten decision-making time and deliver user-centered security.

AI applications are also expected to improve digital safety and reliability, stimulate economic and industrial growth through new solution markets and jobs, and promote a knowledge-based society. These contributions should strengthen domestic and international competitiveness and support social safety and economic growth.

A representative said, “Focused government investment in AI, one of the three game-changing technologies, and cybersecurity, one of the six core digital strategic technologies, has led to this integrated support project and gives our consortium growth momentum. Given the project’s scale and use of technologies drawing worldwide attention, we will work with Microsoft Korea to increase competitiveness.”

CEO Kim Ki-hong said, “We hope generative AI tailored to Korea’s threat environment will help ministries, incident response centers and other public institutions strengthen cyber defense through automated AI responses to increasingly sophisticated attacks.” He added, “We will also promote the fusion of LLMs and other next-generation generative AI with security across financial institutions and private industry, improve Korean companies’ competitiveness, support technology transfer and use, and pursue leadership in security approaches and global expansion.”

In March, SANDS Lab signed separate agreements with LG Uplus and 42Maru for cybersecurity LLM development, and with Microsoft Korea for next-generation AI security technology. Last year, it also successfully completed KISA’s KRW 5 billion cybersecurity AI dataset updating and enhancement project, helping drive Korea’s AI and big data-based security industry.