SK Telecom (SKT) has launched an 'Independent AI Cybersecurity Consortium' in collaboration with Upstage, major domestic cybersecurity companies, and universities.
On October 7, SKT announced that it had signed a memorandum of understanding for AI cybersecurity technology cooperation with these partners. The agreement ceremony included participation from six cybersecurity companies: SKT, Upstage, SK Shieldus, SecureI, AhnLab, RaonSecure, and Genians, along with three universities: Korea University, Soongsil University, and Pusan National University, totaling 11 organizations.
This agreement aims to expand the models developed by SKT and Upstage, which are part of the government's 'Independent AI Foundation Model Project' (Dopamo), into the cybersecurity sector.
SKT and Upstage plan to train their AI models using real-world security data provided by participating organizations, including malware analysis, vulnerability assessments, and security monitoring. This initiative is designed to enhance the AI's ability to understand and assess security situations.
Participating companies will apply the SKT and Upstage models to their security products and services, focusing on vulnerability analysis and real-time security monitoring, even before security patches are available, and will conduct practical demonstrations in real environments.
The consortium will also implement a 'mutual red teaming' approach, where the SKT and Upstage models will attack and evaluate each other to identify vulnerabilities. Findings from this process will be used as training data to strengthen the defensive capabilities of both models.
Additionally, the participating organizations will engage in regular technical exchanges to identify joint AI cybersecurity projects and explore participation in AI and cybersecurity business initiatives.
Yoo Kyung-sang, head of SKT's AI CIC, stated, "The models validated in Dopamo will demonstrate their effectiveness in real security environments alongside leading domestic cybersecurity companies and universities, showcasing the achievements of national AI models in the cybersecurity field as well."
Meanwhile, the third evaluation of Dopamo is scheduled for February 2027.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.