With the arrival of the agent artificial intelligence (AI) era, a recommendation has emerged that, rather than which AI model one possesses, the benchmark for AI sovereignty should be the stable operation of the required technologies and services.
The Chey Institute for Advanced Studies said on the 23rd that it published a report titled "From possession to operation: National security in the agent AI era and sovereign AI 2.0" containing this message.
The report was drafted as a result of an expert roundtable, "New national security in the agent AI era: Global tech hegemony and Korea," co-hosted with the Seoul National University Center for Future Defense Studies at the end of last month.
The crux is that the sovereign AI strategy should be redesigned to focus on access sovereignty and operational sovereignty, beyond securing a proprietary model. Access sovereignty is the ability to reliably use the necessary AI technologies and services, and operational sovereignty is the ability to switch to alternative systems and sustain core functions even if a specific model or cloud service is interrupted.
Sovereign AI is not about developing and possessing all technologies independently; it is about building a structure that, while being closely connected with the outside, does not allow national core functions to become subordinate because of that connection.
Sung Nak-ho, head of hyperscale AI at Naver Cloud, analyzed, "As U.S. restrictions on AI access create limits for traditional product-centered export controls, the focus of controls is shifting to access rights to use AI models and services."
The need to respond to the changing landscape of cyber security driven by agent AI was also raised. Experts identified as a task the strengthening of individual defensive capabilities using AI while expanding coverage to apply it to more systems and assets.
Experts also noted that Korea's AI strategy needs to be redrawn. Given the practical difficulty of securing independent competitiveness in all strategic technology fields, including AI, they said Korea should select areas of strength—such as manufacturing, healthcare, and cultural content—and concentrate capabilities there.
There were also continued views that Korea should leverage its memory semiconductors and advanced manufacturing competitiveness as strategic assets in the AI hegemony race.
However, experts suggested that, because today's industrial advantages will not last forever, Korea should move beyond being a simple supplier of parts and infrastructure to participate in the global AI design process and become a country that accumulates at home the high added value created by AI.
Kim Yu-seok, president of the Chey Institute for Advanced Studies, said, "Sovereign AI 2.0 reflects the awareness that the concept of sovereign AI should be expanded from the mere possession of technologies and models to issues of operational capacity and norm design," adding, "For the Republic of Korea to become a participant, not an observer, in the process of norm formation, it must begin accumulating the necessary capabilities and experience now."