U.S. Blocks Foreign Access to Mythos: "Korean Telecom Suspected of Ties to China"

WP Reports Citing U.S. Officials

Foreign media have reported that a South Korean telecommunications company was behind the U.S. government's decision to block foreign access to the latest AI model from startup Anthropic. It is said that authorities determined this telecom company, which was included in the list of enterprises with access to the AI model, had ties to China.

The Washington Post (WP) reported on the 15th (local time), citing officials, that the U.S. government made this decision after reviewing the organizations that Anthropic had announced would be granted pre-release access to its latest AI model, "Mythos 5."

Anthropic submitted a list of 111 organizations and companies to the administration, which the government approved. However, the situation changed when the company later notified authorities that it had granted access to an additional approximately 50 organizations.

In particular, when Anthropic delayed releasing the names of these newly added organizations for several days, the Trump Administration began a review of export controls. Subsequently, the government reportedly confirmed that the additional list included a South Korean telecom company suspected of having links to China.

The access in question was granted through "Project Glasswing," Anthropic's global cybersecurity cooperation program. The company has offered pre-release versions of "Mythos 5" and "Fable 5" to select U.S. and overseas organizations and companies to test for security vulnerabilities.

A source familiar with Anthropic's internal affairs told WP that the South Korean company's access to Mythos was revoked immediately after it was discovered. However, the outlet reported that this incident has undermined the Trump Administration's confidence in Anthropic's ability to manage sensitive AI technology.

On June 12, U.S. Secretary of Commerce Howard Lutnick sent a letter to CEO Dario Amodei, notifying him of export control guidelines restricting foreign organizations and individuals from accessing Mythos 5 and Fable 5. In response, Anthropic temporarily suspended the service of these models for all users worldwide. Afterwards, the company dispatched a delegation comprised of its top technical executives to Washington to negotiate with the government on AI export controls, according to the Wall Street Journal (WSJ).

Meanwhile, after the U.S. imposed a complete ban on foreign access to Mythos 5 and Fable 5, foreign governments that had planned to use the models to develop cybersecurity measures were forced to scramble. Japan's Nihon Keizai Shimbun (Nikkei) reported that the Japanese government has begun expanding contacts with other AI model developers such as OpenAI and Google.

This content was produced with the assistance of AI translation services.