President Lee Jae Myung speaks during a briefing on the vision and development strategy for Korean semiconductors in the AI era at the presidential office in Yongsan District, central Seoul, on Dec. 10, 2025.PRESIDENTIAL OFFICE/NEWS1
Lee Kwang-hyung
The author is the president of KAIST.
The U.S. government recently imposed export controls restricting foreign access to Anthropic's most advanced AI models, Mythos 5 and Fable 5. Both are cutting-edge cybersecurity systems that were unveiled on June 9. Mythos 5 is available only to verified institutions and cybersecurity professionals, while Fable 5 has been released for general users. Although they were originally designed for cybersecurity consulting, the models are reportedly capable of conducting hacking operations autonomously. Taken a step further, they could serve as offensive cyber weapons.
Washington appears to have concluded that these models constitute strategic technologies directly tied to national and cybersecurity interests, with the potential to threaten U.S. government systems. This is more than another export restriction. For the first time, the United States has effectively classified AI models themselves as strategic assets comparable to semiconductors or advanced weapons. The world is entering a new era of "technopolitics," in which technological capability, rather than geography alone, shapes national power and determines the future of states and societies.
Korea has experienced a similar strategic shock before. In July 2019, Japan announced export restrictions on key materials essential for semiconductor and display manufacturing. The move caught Korea off guard because the two countries had long maintained deeply interconnected industrial supply chains. But the nation responded quickly. Government, industry and researchers worked together to overcome the crisis, strengthening domestic capabilities in materials and components while diversifying supply chains.
Looking back, one may almost feel grateful for that painful lesson. Former Japanese Prime Minister Shinzo Abe unintentionally taught Korea the importance of technological self-reliance. It revealed how essential domestic capabilities are and reminded us that even close allies may one day place their own national interests first.
National competitiveness was once defined by steel and automobiles, then by semiconductors. Today, AI is emerging as the asset that will shape society, culture, the economy, national security and diplomacy. The Mythos episode, therefore, poses a critical question: If Korea can one day no longer freely access the United States' most advanced AI systems, how will the country respond?
Fortunately, the government has already begun preparing for this challenge through its "Top-3 AI Power" strategy and its sovereign AI initiative. Although these policies have attracted skepticism, recent developments suggest that the government has correctly recognized the strategic direction of AI competition.
The Top-3 AI Power strategy is far more than an industrial promotion plan. It represents a national strategy for securing technological sovereignty in an AI order increasingly dominated by the United States and China. At the center of that vision is sovereign AI, which refers to independent AI capabilities that understand Korea's history, culture, legal system and industrial structure, and that remain reliably available whenever the nation requires them.
Just like energy security, AI security now matters. And just as semiconductor sovereignty is indispensable, AI sovereignty is becoming equally important. The Mythos case demonstrates that AI sovereignty is no longer an abstract concept but a practical issue directly connected to national competitiveness. If a country entrusts its AI capabilities entirely to others, it risks surrendering aspects of its national security as well.
To become one of the world's three leading AI powers, Korea must make the sovereign AI project a success.
First, the country must be capable of world-class AI model development. Competing head-on with U.S. technology giants at the same scale would be unrealistic. Instead, Korea should continue building its own large-scale foundation models while accumulating technological expertise and specializing in sectors where it already possesses global advantages. Manufacturing, semiconductors, biotechnology and energy all present opportunities to develop industry-specific AI systems that could lead the world.
Second, Korea must expand its national AI computing infrastructure. The heart of AI competition lies in training and inference capacity. The government's National AI Computing Center Establishment Plan is therefore critically important. Korea's leadership in memory semiconductors should be connected with AI data centers and advanced supercomputing capabilities. At the same time, investment by private domestic firms and foreign companies in AI data centers should be actively encouraged.
Third, the nation must intensify efforts to cultivate talent in strategic technologies. Science and engineering universities should move beyond conducting AI research to become innovation platforms that design national AI strategies while connecting government, academia and industry. Universities will play indispensable roles in sovereign AI through basic algorithm research, cybersecurity, AI semiconductor development, talent cultivation and startup ecosystems. They should also lead entrepreneurship based on AI technologies.
The Mythos case also underscores the growing importance of AI security. Korea must strengthen education and research on AI cybersecurity, covering both attacks against AI models and attacks carried out using AI. As I conclude my tenure as KAIST president, this remains one of my greatest regrets. Years ago, I recognized the importance of strengthening information security, but I could not implement my vision as much as I hoped to. Looking back, I wish that I had pushed more forcefully.
The Mythos episode has delivered a warning to Korea. Just as Japan once used technology as leverage, the United States could someday exert pressure through AI. Sensitive government information could even become vulnerable through dependence on foreign AI systems. Rather than viewing this case as a setback, Korea should treat it as an opportunity. If we respond wisely, perhaps a few years from now, we will be able to look back and say, "Thank you, Mythos."
This article was originally written in Korean and translated by a bilingual reporter with the help of generative AI tools. It was then edited by a native English-speaking editor. All AI-assisted translations are reviewed and refined by our newsroom.
