U.S. eases Mythos 5 curbs as Korean access remains unclear

by Kim Hee-su Posted : June 27, 2026, 12:34Updated : June 27, 2026, 12:34
Anthropic logo a keyboard and a robotic hand in this illustration taken June 5 2026 Reuters-Yonhap
Anthropic logo, a keyboard, and a robotic hand in this illustration taken June 5, 2026. Reuters-Yonhap
SEOUL, June 27 (AJP) - The U.S. government has eased restrictions on Anthropic’s advanced artificial intelligence model Claude Mythos 5, allowing the company to provide access to more than 100 approved companies and organizations.

The Commerce Department authorized Anthropic on Friday to make Mythos 5 available to selected “trusted partners,” according to a letter obtained by Reuters.

The approved list includes a number of Fortune 500 companies and institutions, a person familiar with the matter said.

Foreign nationals working for approved companies, as well as vetted foreign employees of Anthropic, will also be allowed to use the model without obtaining individual export licenses.

The decision marks a partial reversal of restrictions imposed by the Trump administration two weeks earlier.

On June 12, the administration ordered Anthropic to block all foreign nationals from accessing Mythos 5 and Fable 5, another advanced model equipped with safety safeguards, citing national security concerns.

Because Anthropic did not have a system capable of verifying every user’s nationality, the company suspended access to both models for all users.

The unusually broad export controls raised concerns within the technology industry that Washington could weaken U.S. leadership in artificial intelligence by restricting access to advanced models rather than only controlling semiconductors and computing equipment.

Commerce Secretary Howard Lutnick said Anthropic had made substantial progress in addressing the government’s concerns.

“Since June 12, Anthropic has worked with the government to address the relevant risks and has made significant progress,” Lutnick said in the letter. “In just two weeks, it has worked diligently to protect our security while ensuring that America remains the global leader in AI.”

It remains unclear whether any South Korean companies or institutions are included on the newly approved list, which has not been made public.

Before the restrictions were imposed, South Korea had secured access to Mythos through Anthropic’s Project Glasswing, a cybersecurity initiative that uses advanced AI to identify and address software vulnerabilities.

The Korea Internet & Security Agency joined the project alongside major South Korean companies, including Samsung Electronics, SK Hynix and SK Telecom.

Their access under the revised policy, however, has not been confirmed.

The earlier restrictions reportedly followed U.S. government concerns about Anthropic granting Mythos access to SK Telecom over the company’s alleged links to China.

SK Telecom has rejected the allegations, saying there was no basis for describing the company as connected to China.

The latest authorization applies only to Mythos 5 and approved partners. Restrictions on Fable 5 remain in place.