Cybersecurity-Specialized AI Foundation Model Development Program
Category: AI in Cybersecurity
A South Korean government R&D program (run by MSIT and NIPA) to develop a domestically-owned, cybersecurity-specialized AI foundation model, with SK Telecom and Naver Cloud consortiums competing for selection. Cybersecurity-Specialized AI Foundation Model Development Program was founded in 2026. Based in Sejong, South Korea. Latest round: Undisclosed.
- Founded
- 2026
- Headquarters
- Sejong, South Korea
Value proposition
Build Korea's own sovereign cybersecurity-specialized AI foundation model to reduce dependence on foreign models (notably Anthropic's Mythos) and accelerate AI transformation (AX) of the domestic cybersecurity industry, with the resulting model released as open source.
Products and solutions
A cybersecurity-specialized AI foundation model (open-source oriented) covering threat detection, vulnerability analysis, attack-path analysis, malicious-code generation defense, and agentic security services. Two competing consortium proposals: SK Telecom-led and Naver Cloud-led.
Unique value
Korea's first domestically developed, sovereign cybersecurity-specialized AI foundation model, backed by 256 Nvidia B200 GPUs (32 nodes) over 10 months for the selected consortium — a national strategic response to the 'Mythos shock' and US export restrictions on advanced AI models.
Target customer
Korean cybersecurity industry, public-sector and national-security organizations, and the broader domestic AI/security ecosystem.
Industries served
Cybersecurity, public sector / national security, AI infrastructure
Technology advantage
Government-backed sovereign AI program providing 256 Nvidia B200 GPUs (32 nodes) for 10 months; leverages Korea's leading telecom (SK Telecom) and cloud (Naver Cloud) AI infrastructure; open-source-oriented model designed for the domestic cyber-threat environment; consortiums aggregate Korea's top security firms, universities, and research institutes.
How they differentiate
Unlike commercial US models (e.g., Anthropic's Mythos) that face export restrictions, this program builds a domestically owned, open-source cybersecurity AI model optimized for Korea's specific cyber-threat environment, with government-backed GPU compute (256 Nvidia B200s) and a consortium of Korea's leading security and AI players.
Main competitors
Anthropic's Mythos (US cybersecurity AI model), other Korean sovereign AI foundation-model programs (Dokpamo, ModuAI), global security-AI models from Microsoft, Google, and CrowdStrike
Key partnerships
Ministry of Science and ICT (MSIT/과기정통부), National IT Industry Promotion Agency (NIPA), NVIDIA (B200 GPU supply), SK Telecom-led consortium (13 orgs: Upstage, SK Shieldus, Secui, AhnLab, Igloo Corp, RaonSecure, SmartM2M, Aim Intelligence, Genians, Piolink, KISIA, Korea University, Soongsil University), Naver Cloud-led consortium (32 orgs incl. LG CNS, LG AI Research, LG Uplus, Logpresso, MarkAny, SANDS LAB, Sparrow, S2W, AI Spera, ESTsecurity, Jiransecurity, Trillion Labs, Theori Korea, SNU, KAIST, POSTECH)
Major milestones
2026-07-22: Program announced and application opened by MSIT and NIPA, 2026-08-21: Application deadline (NIPA), 2026-08-26: Two consortiums (SK Telecom-led, 13 orgs; Naver Cloud-led, 32 orgs) confirmed as finalists, 2026-09 (early): Final selection of one consortium, 2026-09 to 2027-07: 10-month development with 256 Nvidia B200 GPUs
Market positioning
A flagship Korean sovereign-AI initiative positioning Korea to achieve cybersecurity AI independence, competing with US models like Anthropic's Mythos while strengthening the domestic security AI ecosystem.
Geographic focus
South Korea (national sovereign-AI program); indirectly competes with US and global cybersecurity AI models.
Latest news about Cybersecurity-Specialized AI Foundation Model Development Program
More AI in Cybersecurity companies
Official website: https://www.nipa.kr/home/2-2/16866
