Trump Administration Imposes Export Controls on Anthropic's Claude Mythos After SK Telecom Access Dispute
Key Takeaways
- ▸The Trump administration imposed export controls on Claude Mythos and Fable 5 following concerns that SK Telecom—a major Anthropic investor with alleged ties to China through its parent SK Group—gained access to the advanced AI model
- ▸Amazon researchers identified exploitable vulnerabilities in Fable 5 that could circumvent guardrails and access Mythos's powerful cybersecurity capabilities, amplifying US government concerns about technology safeguarding
- ▸Anthropic disabled access to both models entirely rather than comply with nationality-based gatekeeping, placing the company at odds with the White House over how to bring the technology back online
Summary
The Trump administration has ordered Anthropic to revoke access to its Claude Mythos and Fable 5 models for all foreign nationals, including immigrants within the US, following geopolitical concerns and security vulnerabilities. The export control decision was triggered by Anthropic granting South Korean telecom giant SK Telecom access to Mythos through its Project Glasswing program, which the US government expressed concerns about due to alleged ties between SK Telecom's parent company, SK Group, and China. The concerns were compounded when Amazon identified vulnerabilities in Fable 5 that could potentially bypass guardrails and access Mythos's formidable cybersecurity capabilities.
Anthropc chose to disable access to both models entirely rather than implement nationality-based restrictions that would be difficult to enforce while preserving privacy. The company has not publicly commented on the action, but sources indicate the White House and Anthropic remain in negotiations to bring the models back online. SK Telecom, which has invested over $100 million in Anthropic and participated in Project Glasswing alongside Samsung Electronics and South Korea's Internet and Security Agency, has denied having ties to China but has not elaborated further on the situation.
- The controversy reflects the collision between national security interests and the global nature of AI development, forcing companies to navigate impossible compliance scenarios
Editorial Opinion
The Mythos export control reveals the fundamental tension in AI governance: while restricting access to powerful technology makes intuitive national security sense, nationality-based gatekeeping is both technically impractical and commercially devastating. More troubling is how identified security vulnerabilities—which Anthropic and outside experts argue are not unique to Claude—triggered outsized geopolitical responses that ultimately suspended service rather than resolving the underlying issue. As AI capabilities advance, this pattern of choose-your-own-disaster compliance scenarios will repeat until governments and companies develop more nuanced frameworks for managing both innovation and security.
