Naver selected to develop cybersecurity-focused AI foundation model

A consortium led by Naver Cloud has been selected to develop a cybersecurity-focused AI foundation model. Naver will now begin work on the project in earnest alongside LG Group's ICT affiliates, which are part of the consortium.

The selection comes amid growing global concern over AI-enabled cyberattacks, including the threat of AI systems capable of autonomous hacking. Development of a Korean-built security-focused AI model — dubbed "K-Mitos" — is expected to gain momentum.

The Ministry of Science and ICT announced Thursday that it had selected the Naver Cloud consortium as the final project operator for its "Cybersecurity-Focused AI Foundation Model Development" initiative, following a presentation evaluation of the two competing consortiums.

The Naver Cloud consortium comprises 32 institutions in total, including key LG Group ICT affiliates LG CNS, LG AI Research and LG Uplus.

The evaluation was conducted by a panel of external experts in AI and cybersecurity, who assessed each consortium's project proposal and presentation. The panel reviewed three criteria: technical capability and development experience; the quality and feasibility of development strategy, technology and goals; and market potential and broader impact.

The Ministry of Science and ICT said the selected operator was recognized for its strong technical capabilities and development track record, its large-scale industry-academia-research consortium structure, and the quality of its development strategy — including a multi-agent and harness configuration.

The selected consortium will receive 256 Nvidia B200 GPU units over 10 months starting this month. Support will be provided for the first five months, with the remaining five months contingent on the results of a midterm evaluation.

The ministry plans to sign an agreement with the selected operator in September and begin development of the cybersecurity-focused model through GPU support.

"We plan to immediately form a working-level consultative body with the selected operator and work toward producing results as quickly as possible," the ministry said.

With the selection of a developer for the cybersecurity-focused AI model now complete, expectations are rising that South Korea's AI ecosystem — built on homegrown technology — will grow more robust.

Three flagship government AI initiatives are now moving forward in earnest: the independent AI foundation model project, the Moduui AI program and the newly selected cybersecurity-focused AI project.

The independent AI foundation model project — known as "Dokpamo" — has completed its second evaluation round and narrowed the field to three teams. LG AI Research, SK Telecom and Upstage have advanced to the third evaluation round, with the final selection of two teams expected in early next year.

For the "Moduui AI" initiative, which aims to build an ecosystem where all citizens can use AI for free, three consortiums led by SK Telecom, Kakao and KT Corp have been selected. A general-purpose AI chatbot service set to launch before year's end will be available to all citizens free of charge with no usage limits. The government plans to launch a beta service later this month and roll out the full service in December.