SK Telecom, Upstage launch AI cybersecurity consortium

11 organizations join, including 6 security firms and 3 universities; consortium to advance AI models using 17 billion daily threat analyses and 2.5 petabytes of security data

Participants pose for a group photo at the launch ceremony of the AI Cybersecurity Consortium held at SKT Tower in Jung-gu, Seoul, on Wednesday. Pictured from left: Choi Yong-jin, head of SK Telecom's AI Tech Center; Choi Dong-hee, head of SK Telecom's AI Strategy Planning Division; Choi Dae-sun, dean of Soongsil University's College of AI; Heo Shin-wook, professor at Pusan National University; Kim Tae-yun, head of SK Telecom's Foundation Model division; Lee Jae-yong, head of Upstage's public sector; Yoo Kyung-sang, head of SK Telecom's AI CIC; Jeong Ik-rae, dean of Korea University's Graduate School of Information Security; Kwon Heon-yeong, professor at Korea University; Kang Seok-gyun, CEO of AhnLab; Jeong Sam-yong, CEO of Secui; Jo Young-cheol, CEO of Piolink; Lee Dong-beom, CEO of Genians; Lee Jeong-a, CEO of RaonSecure; and Kim Byeong-mu, executive vice president of SK Shieldus. [SK Telecom]

SK Telecom and Upstage are mounting a full-scale effort to prevent AI-driven cyberattacks. The two companies have launched an AI cybersecurity consortium with leading domestic security firms and universities, aiming to accelerate AI-based information security.

SK Telecom announced Thursday that it had launched the consortium together with Upstage, major domestic cybersecurity companies and universities, and signed MOUs with the participating organizations to advance AI information security technology cooperation.

The consortium brings together 11 organizations in total: SK Telecom and Upstage, six cybersecurity firms — SK Shieldus, Secui, AhnLab, RaonSecure, Genians and Piolink — and three universities: Korea University, Soongsil University and Pusan National University.

SK Telecom and Upstage will train their respective AI models on security operations data generated in real-world settings by the participating organizations, including malware analysis, vulnerability assessment and security monitoring. The goal is to enhance the AI models' ability to understand and respond to security situations.

During the pre-training phase, the models' foundational knowledge will be strengthened using security data. SK Telecom's next-generation model, A.X K3, will be systematically trained on security domain data and supplemented with data to ensure the safe operation of its "AI for All" public service. Upstage will focus on training its next-generation Solar model — capable of analyzing large volumes of code and long-term logs in a single pass — on security data for vulnerability detection and analysis.

In the post-training phase, real-world security operations data will be used to sharpen the models' practical task performance. Key focus areas include software vulnerability analysis and patching, incident response at security operations centers, and the safe use of security tools.

The quality of AI security capabilities depends on field data accumulated through real threat encounters and response experience. The participating cybersecurity firms will contribute their domain expertise — including threat intelligence, malware analysis data and security solution operational experience — to model development and validation.

SK Shieldus analyzes about 17 billion threats per day in real time through its integrated security monitoring platform Securidium. AhnLab holds more than 2.5 petabytes of security data along with malware and endpoint threat detection technology.

Secui brings high-performance firewall solutions; RaonSecure contributes domain-specific security data and autonomous penetration testing technology; Genians offers operational data from real-world security solutions and post-training datasets; and Piolink provides autonomous attack verification and detection rule generation technology.

Yoo Kyung-sang, head of SK Telecom's AI CIC, delivers a welcoming address at the launch ceremony of the AI Cybersecurity Consortium. [SK Telecom]

The consortium plans to continuously incorporate the latest threat data accumulated from security operations into model training.

Participating companies will also apply A.X K and Solar to their own security products and services — including zero-day vulnerability analysis and real-time security monitoring — and conduct joint pilot deployments in live operating environments. During the pilot process, the models' performance and safety will be assessed, and the analytical and response insights gained in the field will be fed back into training. The consortium aims to build a virtuous cycle linking data acquisition, model training and performance validation.

In addition, the consortium will apply "mutual red-teaming," in which A.X K and Solar attack and evaluate each other to identify vulnerabilities. Weaknesses uncovered in the process will be incorporated as training data to strengthen the defensive capabilities of both models.

"The level of a security AI depends on how broadly and deeply the model can reason," said Yoo Kyung-sang, head of SK Telecom's AI CIC. "The models validated through our proprietary AI foundation model project will demonstrate their effectiveness in real security environments alongside Korea's leading cybersecurity firms and universities, proving the value of our national AI models in the cybersecurity domain as well."

ko@heraldcorp.com