The US government's ban on Anthropic's latest AI model, Fable 5, extends beyond American borders, prohibiting its use anywhere on Earth outside the United States. According to a report by Wired, the Trump administration's decision to enforce strict export controls on Anthropic's most advanced AI technology was triggered by a dispute over the startup granting South Korean telecom giant SK Telecom access to its powerful Claude Mythos model.
Dispute Over SK Telecom Access
Sources familiar with the matter revealed that US officials were unhappy with alleged ties between SK Telecom and China. The situation escalated after Amazon CEO Andy Jassy flagged vulnerabilities in Fable 5. Amazon reportedly alerted the Trump administration that its researchers had managed to "jailbreak" Fable, bypassing its safety guardrails in a way that could threaten national security.
Background of the Conflict
In its report, Wired noted that the situation builds upon earlier reporting by The Washington Post, which highlighted administration alarms over a then-unnamed South Korean telecom company. In response to those initial reports, SK Telecom told a Korean newspaper that the anonymous allegations lacked verified facts and denied any ties to China.
Furthermore, a source close to Anthropic emphasized that the company viewed the SK Telecom access and the Amazon security flaws as entirely separate issues, pointing out that the official government mandate restricting access exclusively to US citizens made no explicit mention of China or the Korean carrier.
White House Distrust and Consequences
A person close to the administration told Wired that the series of events led the White House to conclude it could not trust Anthropic to secure its cutting-edge technology. Consequently, the Trump administration ordered Anthropic to revoke access to Mythos and Fable 5 for all foreign nationals, including those residing within the United States. Rather than implementing a logistically complex and privacy-infringing nationality-based gatekeeping system, Anthropic chose to take the models entirely offline, leaving the company and the White House deadlocked after days of tense negotiations.
SK Telecom's Link with China
The underlying model, Claude Mythos, possesses exceptional software vulnerability detection capabilities, which initially prompted Anthropic to limit its availability to a select group of trusted entities under an initiative called Project Glasswing. When Anthropic expanded the program to roughly 150 companies, SK Telecom was included alongside other prominent Korean institutions like Samsung Electronics and the Korea Internet and Security Agency.
SK Telecom's involvement stems from a deep financial relationship, including a $100 million investment in 2023 aimed at building a telecom-specific AI model. However, shortly after Project Glasswing's expansion, the White House ordered Anthropic to cut off SK Telecom. Sources told Wired that Anthropic complied immediately, and the government did not threaten export controls at that specific juncture.
Historical and Financial Ties
While SK Telecom itself maintains a minimal footprint in China, generating only $1.9 million there in 2024 and employing just seven people, it belongs to the massive SK Group, which has extensive Chinese interests in energy and semiconductors. Wired detailed that SK Telecom's own history in China spans over two decades, highlighted by a 2004 joint venture called UNISK with state-owned China Unicom and a subsequent $1 billion bond investment in 2006.
Although SK Telecom divested the majority of its stake back to the Chinese carrier for $1.3 billion in 2009, a 2025 SEC filing shows it still holds a $17 million financial interest in UNISK. This historical connection is particularly sensitive given that the first Trump administration blacklisted US investment in China Unicom due to alleged military links, and the FCC recently proposed a ban preventing US telecom companies from interconnecting with the Chinese carrier over national security concerns.
