The Trump administration’s move to impose export controls on Anthropic’s most powerful AI technology followed a dispute over the company granting SK Telecom, South Korea’s largest wireless carrier, access to its Claude Mythos model, according to people familiar with the matter. US officials were concerned about what they alleged were SK Telecom’s ties to China, those people said. The controversy escalated when Amazon researchers identified vulnerabilities in Fable 5, a safeguarded version of Mythos, leading the White House to determine it could not trust Anthropic to safeguard its advanced AI.
The Role of SK Telecom
SK Telecom became one of roughly 150 companies to receive early access to Mythos through Project Glasswing, a program initially restricted to a small group of trusted organizations. The AI model is exceptionally skilled at identifying software vulnerabilities. Anthropic expanded Project Glasswing “following several weeks of close collaboration” with outside experts and the US government, according to the source.
SK Telecom has a significant financial relationship with Anthropic: a $100 million investment in 2023 that coincided with a commercial partnership to develop an AI model tailored to the telecommunications industry. Other Korean organizations in Project Glasswing included Samsung Electronics and the Korea Internet & Security Agency. While SK Telecom itself generated only about $1.9 million in revenue from China in 2024, it is part of the larger SK Group conglomerate, whose affiliates have extensive China business interests spanning semiconductors, energy, and other industries.
US Government Response and Export Controls
Shortly after Anthropic announced Project Glasswing’s latest expansion earlier this month, the White House asked Anthropic to revoke SK Telecom’s access to Mythos, according to a person close to the AI lab. Anthropic immediately complied, and the US government did not threaten export controls at that time. However, the situation escalated when Amazon flagged vulnerabilities in Fable 5.
| Event | Timeline | Action Taken |
|---|---|---|
| SK Telecom given Mythos access | Early June 2025 | Part of Project Glasswing expansion |
| White House asks Anthropic to revoke SK Telecom’s access | After expansion announcement | Anthropic complies; no export controls threatened |
| Amazon flags Fable 5 vulnerabilities | Around June 9, 2025 | White House determines Anthropic cannot be trusted |
| US government orders access revocation for all foreign nationals | June 13, 2025 (Friday) | Anthropic disables models entirely |
The confluence of events led the White House to order Anthropic to revoke access to Mythos and Fable 5 for all foreign nationals, including immigrants inside the US. Rather than gate access based on nationality—a process difficult to implement while preserving privacy—Anthropic decided to disable access to the models entirely. The White House and Anthropic remain at odds after days of negotiations about bringing the models back online.
Amazon’s Vulnerability Findings
Amazon researchers claimed it was possible to circumvent some of Fable 5’s guardrails and access Mythos’ formidable cybercapabilities. Anthropic and outside cybersecurity experts argued these risks are not unique to Claude. A person close to Anthropic said the company viewed SK Telecom’s access and the Amazon vulnerabilities as separate issues, noting that the US government’s letter demanding restrictions did not reference the Korean company or China.
Anthropic’s Decision to Disable Access
Anthropic declined to comment. The White House and SK Telecom did not immediately respond to requests for comment. The Washington Post previously reported that Trump administration officials were alarmed that Mythos recipients included a “South Korean telecommunications company” with alleged links to China, though the article did not name the firm. In response, SK Telecom told a Korean newspaper that the “anonymous insider’s remarks in foreign media lack verified facts, and our company has no ties to China.”
For enterprise technology leaders, the episode illustrates the increasing sensitivity of advanced AI model access and the potential for geopolitical tensions to disrupt partnerships. The intersection of national security concerns, foreign investments, and AI safety vulnerabilities creates a volatile regulatory environment for companies deploying frontier AI models.
