On the 12th (local time), the U.S. government issued guidance restricting overseas access to Anthropic's latest artificial intelligence (AI) models "Mythos 5" and "Fable 5," refocusing Korea's science and technology community on the need for sovereign AI (AI capabilities that a country can directly develop, operate and control). That is because it has become a reality that cutting-edge AI models can be controlled as national strategic assets.
The move was reportedly prompted by concerns that the models could be "jailbroken," allowing users to bypass prompt controls. Anthropic said the issue stemmed from a misunderstanding, but it blocked all access first to comply with the guidance.
Experts in the science and technology community see the incident as a direct warning for Korea as well. It showed that AI access could be restricted depending on a given country's policy decisions in a situation of high reliance on overseas foundation models. Foundation models refer to large, general-purpose models that underpin a range of AI services, including chatbots, translation, search, coding, and medical and scientific analysis.
Choi Ki-young, former Minister of the Ministry of Science and ICT (professor at Seoul National University's Department of Electrical and Computer Engineering), said, "There has long been considerable concern that AI could be treated as an important national strategic asset and that access by other countries could be controlled, and this incident appears likely to further heighten that awareness."
He added, "High-performance AI models can be used for both cyberattacks and defense, and it would not be anything but a major threat to other countries if one country monopolized such models," noting, "Taking this as a cue, Korea should accelerate efforts to develop top-tier, independent foundation models."
However, developing independent models requires massive compute resources, talent, data centers, and investment. The former Minister said, "For Korea, which lacks talent, graphics processing units (GPUs), data centers, and investment—indeed all resources—this is not easy," adding, "Industry, academia, research institutes, and government must work together more closely."
Physician-scientist Kim Nam-kuk, a professor in the Department of Convergence Medicine at Ulsan University College of Medicine and Asan Medical Center, predicted the move will further stoke global AI development competition. Kim said, "Paradoxically, in the long run, I expect this to become a boomerang that weakens the U.S.-centric global AI ecosystem," adding, "Beyond the U.S.-China bipolar order, it will be reorganized into a new equilibrium such as a tripolar system in which excluded countries band together."
He went on, "Right now, Korea's sovereign AI development strategy tends to insist too much on building fundamental technologies," adding, "For security, we should first develop high-performing AI from an applied-technology perspective, and we must prepare comprehensively through institutional support for big data discovery, data use and compensation, and investment in AI infrastructure."
Lee Sang-yeop, distinguished professor in the Department of Chemical and Biomolecular Engineering and vice president for research at KAIST, said it is more important for Korea to secure AI capabilities that operate without overseas access in key areas than to immediately catch up with U.S.-level, ultra-large models across all domains.
Lee said, "Ultimately, it is good to pour vast resources into developing sovereign AI, but resources are not always so abundant," adding, "Even if usage controls are imposed by AI-leading countries, we should build models that run well for defense, healthcare, research, and administration."