Reports indicate that a South Korean telecom suspected of ties to China has contributed to growing distrust within the White House as the U.S. government reviews export controls on Anthropic's latest artificial intelligence (AI) model.
The Washington Post reported on June 15, citing two anonymous White House officials, that the Trump administration had been considering export control measures weeks before requesting the withdrawal of Anthropic's latest AI model.
According to the report, Anthropic submitted a list of 111 organizations that would receive pre-access to its latest model, 'Mitos,' to the Trump administration several weeks ago. After reviewing the list, U.S. authorities approved it.
However, Anthropic later revealed that the number of organizations granted access had increased beyond the initial list, with about 50 additional entities already using Mitos. When the company did not disclose the additional list for several days, the U.S. administration began to consider applying export controls.
White House officials indicated that the late-submitted list included a South Korean telecom suspected of having ties to China. The Post did not disclose the name of the company.
A source familiar with the situation stated that Anthropic quickly revoked the telecom's access to Mitos. However, this incident has reportedly significantly undermined the Trump administration's trust in Anthropic's ability to protect sensitive technology. One official told the Post, "Access was expanded too broadly and too far."
While officials from the Department of Defense, the Central Intelligence Agency (CIA), and the National Security Agency (NSA) advocated for export controls, some within the administration expressed caution regarding such measures.
The discussion gained momentum after Amazon informed the administration last week about security vulnerabilities found in Anthropic's new model. White House officials viewed this as another instance of mismanagement by Anthropic.
As a result, on June 12, the U.S. government announced export control guidelines restricting access to Anthropic's top AI models, 'Mitos5' and 'Fable5,' for foreign entities and individuals. Following this, Anthropic immediately suspended services for Mitos5 and Fable5 to ensure compliance with the guidelines.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.