SK Telecom and Upstage Join Forces for 'Security AI All-Star Team' as Elite 'Dopamo' Squad

[Financial News] A security-focused artificial intelligence (AI) all-star team has emerged, bringing together two of the three elite teams in the 'sovereign AI foundation model' project. The plan is to embed data and know-how from South Korea's security front lines into a homegrown AI model and build a practical security AI system that can compete in the global market.
On the 27th, SK Telecom and Upstage said they had formed a consortium with major domestic security firms, universities, and the Korea Information Security Industry Association (KISIA) and submitted an application to join the Ministry of Science and ICT's project to develop a 'cyber security-specialized AI foundation model.'
At the core of the consortium is a structure that links a national AI model, real-world security data, and field validation.

The elite team that has been developing the national AI model will build the model, while security companies will feed data from actual operations into training. The finished model will then be tested in the field again, allowing further improvements.
In the global security industry, small AI models specialized for security are increasingly being combined with AI agents to automatically carry out not only threat detection but also blocking and response. In South Korea, however, no dedicated AI model has yet been built to learn from Korean-language security data and understand the country's regulatory framework.
SK Telecom plans to help secure national cyber sovereignty by demonstrating AI security services tailored to South Korea's cyber threat environment.
As the lead company in the consortium, SK Telecom will oversee the overall project and work with Upstage on model development.
SK Telecom and Upstage recently passed the second-stage evaluation for the sovereign AI foundation model project led by the Ministry of Science and ICT. Two of the three teams that cleared the second round are now joining the same consortium, concentrating domestic capabilities in large-scale foundation model development on this project.
The two companies plan to develop a security-specialized model based on the A.X K and Solar Open model families they have been building as sovereign AI foundation models. When combined with AI agents, the model can automate tasks from threat detection to response.

Nine major South Korean security companies, including SK Shieldus, SECUI, AhnLab, Igloo Corporation, RaonSecure, Smart M2M, AIM Intelligence, Genians, and PIOLINK, are also taking part. The nine firms will contribute data accumulated from real security operations and help validate the model in their own commercial products and monitoring environments.
Universities and the industry association are also involved. The industry-academic cooperation foundations of Korea University and Soongsil University will independently verify the model's performance and safety, separate from the developers, while also training professionals in security and AI convergence. KISIA will work to spread the project's results across the industry and connect them with employee training programs.
Another strength of the consortium is its security expertise. SK Telecom said it has closely followed the latest technology trends as an early investor in global AI companies such as Anthropic and through ongoing partnerships. As overseas efforts to use high-performance AI models for cyber defense continue to grow, the company plans to apply those lessons to South Korea's security environment.
As a national telecommunications operator, SK Telecom has the systems and operational know-how to monitor telecom infrastructure around the clock.
It also has a strong security governance structure. The Chief Information Security Officer (CISO) organization is directly under the CEO and serves as the company's security control tower. The integrated security center under the CISO operates a 24-hour detection and response system, with 129 dedicated staff members currently working there. Including the Chief Privacy Officer organization, the total rises to 149.
Under the CISO, the company also runs a separate white-hat hacker unit called the Red Team, which conducts vulnerability assessments from an attacker's perspective. Based on these internal response and recovery capabilities, the consortium aims to build a trustworthy public security AI infrastructure.


[email protected] Jang Min-kwon Reporter