Participation in the security-specific AI project of the Ministry of Science and ICT
Cooperation with SKT and Upstage among the elite teams of German wave capabilities
Demonstrated on-site 9 domestic security companies
Learning Korean security data and domestic regulatory environment
사진 확대
SK Telecom will join hands with Upstage to develop an artificial intelligence (AI) foundation model specialized for domestic cyber threats. The goal is to learn actual control data of domestic security companies into AI and verify it at commercial security sites to automate threat detection and response.
SK Telecom announced on the 27th that it has formed a consortium with Upstage, nine domestic security companies, and universities and associations and submitted an application to participate in the Ministry of Science and ICT’s “cyber security-specific AI foundation model” development project.
In this consortium, SK Telecom and Upstage will be in charge of developing AI models. The two companies are two of the three elite teams that recently passed the second phase of the AI foundation model project promoted by the Ministry of Science and ICT.
Based on SK Telecom’s “A.X K” and Upstage’s “Solar Open,” it plans to develop AI models tailored to the domestic cybersecurity environment. The goal is to combine Korean security data and models that have learned the domestic security environment with AI agents to automate blocking and measures as well as finding threats.
Actual field data held by domestic security companies are used for AI learning and verification. The consortium will include nine security companies including SK Shields, SecuI, AhnLab, Igloo Corporation, Raon Secure, Smart M2M, Am Intelligence, Genius, and Fioreink.
SK Shields and Igloo Corporation are in charge of security control and threat analysis. AhnLab and Genius participate in terminal security and malicious code analysis. Security I and Genius are in charge of network and access control, while Raon Secure and Smart M2M and Fioreink participate in simulated penetration and vulnerability checks. Aim Intelligence evaluates the safety of AI models themselves.
Security companies provide data accumulated at each company’s site for AI learning and apply the developed model back to commercial products and control environments. The plan is to create a circular structure of “development → field demonstration → evaluation → re-learning” by evaluating the results confirmed in the field and then re-learning the model.
SK Telecom also utilizes its own security operation experience. SK Telecom has the organization of the Chief Information Protection Officer (CISO) directly under the direct control of the CEO. At the CISO-affiliated Integrated Security Center, 129 dedicated personnel perform 24-hour detection and response tasks.
The model verification involves academics. Korea University and Soongsil University Industry-Academic Cooperation Group verify the performance and safety of the model separately from the development entity and train security and AI convergence personnel. The Korea Information Protection Industry Association (KISIA) is responsible for spreading the development results to the industry and linking them with the education of the incumbent.
Cho Dong-yeon, head of AI Convergence at SK Telecom, said, “We will create an actual security AI system by putting data and know-how from domestic security sites into domestic AI models. We will expand the security AI ecosystem so that we can compete in the global market with capabilities created by domestic industry-academic associations.”