Anthropic's top AI models set to return to Korea as U.S. drops controls

The U.S. government has lifted the ban on Anthropic's most advanced AI models, Mythos 5 and Fable 5, a decision expected to restore access to the two systems in Korea after weeks of restrictions.

According to Anthropic's X post on Tuesday, the U.S. Department of Commerce had "lifted export controls on Claude Fable 5 and Mythos 5." The tech company promised to begin restoring access the following day and provide a fuller update soon. 

Washington had ordered the models offline on June 12, citing national security concerns in a directive that cut off all access by foreign nationals as Anthropic pulled both models worldwide. On June 26, the government eased the ban and allowed Anthropic to offer Mythos 5 again to a group of vetted U.S. organizations. The reversal ends a suspension of about two and a half weeks.

In withdrawing the controls, the Commerce Department cited commitments the company had made.

"Anthropic has agreed to proactively detect and address security risks associated with the models; to work diligently with the U.S. government on protocols and standards and releases for Mythos, Fable, and future models; and to inform the U.S. government of any malicious activity," Commerce Secretary Howard Lutnick said in a letter to the company that was quoted by Reuters.

The department, he added, "reserves the right to re-evaluate the decisions made in this letter and the necessity of reimposing a license requirement, should circumstances change or should Anthropic fail to adhere to its commitments."

Anthropic is also working with other big tech companies on shared standards for assessing and fixing jailbreaks, or prompts that allow a model to bypass its safety guardrails.

The Korean government and firms have signed on to Project Glasswing — the program through which Anthropic provides Mythos 5 to a small set of vetted organizations — but the U.S. restrictions have stalled any actual use. With the controls gone, Mythos 5 access is now likely to resume in Korea through approved partners, though Anthropic has not confirmed when or for whom.

Fable 5 is the more widely available of the two, Anthropic's first top-tier model released to the public. Mythos 5 is the same underlying model with fewer guardrails, reserved for vetted institutions because of its higher-risk capabilities in areas such as cybersecurity.

The lifting could ease near-term uncertainty for Anthropic's business in Korea. But U.S. regulations remain a variable for Korean companies that want to use the most capable U.S.-made AI models for core operations, security and semiconductor research and development.

OpenAI took a similar route last month, when it previewed its new GPT-5.6 to a small group of trusted partners whose names it had shared with the government.

OpenAI said at the time that it did not believe "this kind of government access process should become the long-term default," adding, "It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them."

BY KWEN YU-JIN [[email protected]]