Sovereign AI cannot become an alibi

Park Su-ryon

The author is the deputy editor of Content Division Three and the head of corporate research at the JoongAng Ilbo. 

Europe's long-feared U.S. "kill switch" arrived, but not in the form that many expected: The target was not the F-35 fighter jet but AI.

On June 12, two of the world's most advanced AI models, Mythos 5 and Fable 5, were reportedly shut down by a single email from the U.S. secretary of commerce. It marked the first time that Washington placed an AI model under export controls. The move appears to be the first step in a broader effort to align allies around a U.S.-led AI ecosystem.

That direction became clearer at the Group of 7 (G7) summit in France, where the United States said that access to its frontier AI models would be limited to "trusted partners." The message was unmistakable: U.S. AI is to become the global standard, with allies gaining access within a U.S.-designed framework. The strategy closely resembles earlier export controls on advanced AI chips, for which countries were effectively categorized by their level of access.

The Mythos episode has strengthened arguments worldwide for sovereign AI. In Korea, it also supports the Lee Jae Myung administration's push for greater technological self-reliance. From AI models and cloud services to computing power and energy infrastructure, the country's dependence on foreign providers alone is increasingly being viewed as a national security risk.

But the issue is more complicated than advocates of sovereign AI often acknowledge.

France offered perhaps the clearest illustration during the G7 meeting. French President Emmanuel Macron, one of Europe's strongest proponents of sovereign AI and a leading supporter of the French AI company Mistral, criticized Washington's ability to disable access to advanced AI. Even so, he did not present domestic AI as a complete alternative. Instead, he argued that democratic allies must retain stable access to the world's most capable U.S. AI systems. Germany expressed a similar position. Sovereign AI remains important, but it cannot by itself guarantee technological security.

The economic realities are equally daunting. Building globally competitive AI requires enormous and sustained investment in capital, computing infrastructure and talent. U.S. technology companies already operate on a scale that exceeds the resources of most individual nations. Korea, for example, secured a commitment from Nvidia to supply 260,000 GPUs over three years. But Microsoft alone reportedly purchased 485,000 GPUs in 2024, according to the Boston Consulting Group. The gap illustrates how difficult it will be for middle powers to compete independently.

That leaves countries such as Korea facing unavoidable choices about how much to invest in sovereign AI and where to draw the line between self-reliance and international cooperation. The Lee administration has allocated 9.9 trillion won ($6.44 billion) to AI this year as part of its goal of making Korea one of the world's three leading AI powers.

Less visible, however, are policies aimed at the generation that will bear AI's greatest economic consequences.

Concerns regarding youth employment have continued to mount, but government responses have remained modest. The "Youth New Deal," unveiled in April, focused largely on temporary measures, including short-term jobs at the National Tax Service and training programs linked to major companies.

But training alone offers limited comfort if employment opportunities themselves continue to disappear. In May, the number of employed young people fell by 254,000 from a year earlier, according to government data. Meanwhile, the share of young households in the lowest 20 percent for both income and net assets reached 15.2 percent last year, nearly double the 7.9 percent recorded in 2020. According to the Bank of Korea, young people entering the AI era are financially more vulnerable than ever.

Technological sovereignty matters. But if the social foundation supporting those most affected by AI continues to erode, aspirations to become an AI powerhouse risk becoming little more than slogans.

Only recently has the youth issue appeared to receive greater attention from President Lee. During a video conference on June 11, he identified young people as a central policy priority for the second half of his term. That recognition is welcome, but it also raises difficult questions. As companies expected to create jobs grapple with disputes over the proposed "Yellow Envelope" law while organized labor presses for extending the retirement age, where will sustainable employment come from?

The United States' willingness to weaponize cutting-edge AI makes a national AI strategy indispensable. But strategy alone is not enough. If the government fails to protect the generation standing on the front line of the AI transformation, sovereign AI — or any other technological ambition — will not serve as an alibi.