'The development project to create a Korean version of "Mythos AI," a security-specialized artificial intelligence (AI) foundation model (hereinafter referred to as "security-focused foundation model"), will proceed as a showdown between the "SK Telecom-Upstage" consortium and the "Naver Cloud-LG" consortium.
The Ministry of Science and ICT and the Korea Information Society Development Institute (NIPA) announced on the 26th that, following the deadline for business plan submissions from companies participating in the development of security-focused foundation models, a total of two consortia—SK Telecom Consortium and Naver Cloud Consortium—have submitted their applications.
The SKT Consortium includes Upstage, which advanced to the third evaluation round of the "security-focused foundation model" alongside SKT. It also comprises 13 companies including SK Shieldus, SecuAI, AhnLab, Glue Corporation, and Raon Secure, along with institutional investors. The submitted project title is "Development of a Comprehensive Cyber Security-Specialized AI Foundation Model Suite and Demonstration of Agent-Based Security Services Based on the Domestic Cyber Threat Environment."
Naver Cloud has allied with LG AI Research, LG CNS (LG C&S), and LG Uplus. A total of 32 institutional investors, including Eastsec, Logpresso, MarkAny, VDraft, and Korea Aerospace Industries, are also participating. The Naver Cloud Consortium has set the task as "Development of an AI Foundation Model for Internalizing National Cyber Security Capabilities."
The Ministry of Science and ICT plans to swiftly conduct suitability reviews of submitted documents and external expert evaluations (presentation assessments) to select the final winner around early September. Evaluation criteria consist of technical capability and development experience (30 points), development goals (30 points), and marketability and impact (40 points). If the final contractor is selected in September, after confirming the support task, 256 units of GPU (B200) will be provided step-by-step over a total period of 10 months until July next year.
This project was established amid the recognition that domestic independent AI security models are needed as cyber security threats grow due to advancements in AI technology. Previously, Anthropic's "Mythos AI" discovered security vulnerabilities that had gone undetected for over two decades and automatically generated attack code, highlighting increasing security threats from AI. In response, the Ministry of Science and ICT has launched the security-focused foundation model project to secure an independent security AI model.
in TECH
