U.S. export controls on advanced AI highlight need for sovereignty

The U.S. government’s decision to restrict access to Anthropic’s most advanced artificial intelligence models has highlighted the growing importance of technological self-reliance in the AI era.

On Friday, Washington issued an executive order blocking foreign access to Anthropic’s cutting-edge AI models, Mythos 5 and Fable 5. The measure marks the first U.S. export control targeting AI models themselves. The restrictions apply not only to overseas users but also to foreign nationals residing in the United States and foreign employees working for Anthropic. As a result, the company suspended services just three days after their release.

Security concerns had intensified following the launch of the Mythos series in April. In response, Anthropic unveiled two new models on June 9: Fable 5, a version with built-in safety restrictions for general users, and Mythos 5, a security-focused model. Access to Mythos 5 was limited to selected companies and allied countries through a global cybersecurity partnership known as Project Glasswing.

The U.S. government, however, concluded that Fable 5 could potentially be “jailbroken,” meaning its safety mechanisms could be deliberately bypassed. Citing national security concerns, Washington moved to impose export controls.

The decision has heightened concerns that Korea could be relegated to the role of an AI infrastructure subcontractor rather than a leading AI power.

Rep. Ahn Cheol-soo of the People Power Party warned that while Korea possesses world-class semiconductor manufacturing capabilities, it remains significantly behind the United States and China in AI models and software competitiveness. He argued that the country risks becoming little more than a factory supplying chips and that the latest U.S. restrictions should serve as a warning.

AI capability is increasingly becoming a measure of national competitiveness. In a field where technological gaps can widen rapidly, countries that fail to develop their own capabilities risk long-term dependence on foreign technologies.

For that reason, securing sovereign AI — the ability to develop domestic AI models and core technologies independently — has become essential.

The Lee Jae Myung administration has repeatedly emphasized sovereign AI as a national objective. Yet progress has been slow. Ha Jung-woo, the former senior presidential secretary for AI policy planning who led many of the government’s initiatives, was reassigned during the election period, and a successor has yet to be named.

Korea must move quickly to rebuild a government-wide AI control tower and accelerate a national strategy encompassing technological development, talent cultivation, infrastructure expansion and regulatory reform. The latest U.S. export controls demonstrate that access to advanced AI technologies can no longer be taken for granted.