Lee Gyu-lee is a business writer at The Korea Times, focusing primarily on IT & telecommunications, the Ministry of Trade, Industry and Energy and KOTRA. Prior to this, she has covered a wide range of cultural news, from film, television and K-pop to lifestyle and fashion.
Gov't to launch security-focused AI model for public sector this year
Science and ICT Minister and Deputy Prime Minister Bae Kyung-hoon speaks during a media conference marking his first year in office at Government Complex Sejong, Monday. Courtesy of Ministry of Science and ICT
By Lee Gyu-lee
Science and ICT Minister and Deputy Prime Minister Bae Kyung-hoon said Monday the government plans to develop a security-focused artificial intelligence (AI) model for the public sector this year as part of efforts to build domestic AI capabilities and reduce dependence on foreign AI providers.
"We have assessed that by leveraging the AI models we already have and fine-tuning them with security-related data and vulnerability scenarios, it is possible to build a security-focused AI model capable of identifying a wide range of vulnerabilities at the current level of technology,” Bae said during a media conference marking his first year in office at Government Complex Sejong.
"We plan to begin development and first apply the security-focused model in the public sector before the end of the year."
For the private sector, he said the government will evaluate whether the model could be deployed later this year or early next year.
Bae framed the effort as part of a wider push to secure sovereign AI capabilities amid intensifying global competition. He noted that leading AI models are increasingly treated as strategic assets, with governments moving to exert greater control over their development and deployment.
He stressed that the United States has shifted from promoting open AI models toward more closed systems, while China could eventually take a similar approach with its currently open models.
"The trend toward governments seeking to control frontier AI models is likely to become stronger. They're almost like nuclear weapons," he said
Regardless of imminent measures to develop security-focused AI models, Bae added Korea would need substantially greater investment to develop frontier AI models comparable to Mythos.
He explained domestic developers currently receive support equivalent to roughly 500 to 735 graphics processing units (GPU), while a Mythos-class frontier model would require about 10,000 Nvidia Vera Rubin GPUs, with hardware costs alone totaling roughly 3.5 trillion won ($2.37 billion).
“The ministry is in the process of persuading other ministries and fiscal authorities of the need for an investment of that scale, but nothing has been decided yet,” he said.
The minister said the government was pursuing two parallel AI strategies: developing domestic foundation models for public services and industrial AI adoption and exploring investment in frontier AI models capable of competing with the world's best.
He said the government was focused on delivering tangible outcomes from AI adoption while preparing for the emerging era of AI agents, which he said would become the next major battleground for competitiveness.