The White House compelled Anthropic to restrict SK Telecom’s access to its most advanced AI model, Claude Mythos, amid concerns about the telecom’s affiliation with China, sparking a standoff over AI export controls.
- White House orders revocation of SK Telecom’s AI access over China links
- Anthropic disables Claude Mythos access for foreign nationals to comply
- Security concerns raised by Amazon vulnerability report on AI guardrails
What happened
The U.S. government instructed Anthropic to revoke access to its Claude Mythos AI model for SK Telecom, a major South Korean telecommunications company. This directive came amid allegations about SK Telecom’s connections to China through its parent conglomerate, SK Group, which conducts extensive business in China, including telecommunications ventures dating back to 2004.
Anthropic had included SK Telecom in its Project Glasswing initiative, which grants early access to Claude Mythos selectively to trusted partners for testing and development purposes. Following the White House order and concerns raised by Amazon researchers who identified potential vulnerabilities in a safeguarded version of Mythos called Fable 5, Anthropic chose to disable access entirely rather than selectively restrict it by nationality.
Why it matters
The move highlights rising U.S. government vigilance over export controls related to cutting-edge AI technologies and potential risks tied to foreign entities with China affiliations. Claude Mythos is highly capable in cybersecurity tasks, increasing sensitivity around who can access this technology and under what conditions to prevent exploitation or misuse.
SK Telecom’s involvement underscores the complexities global tech partnerships face in a fraught geopolitical environment. While the company’s direct China operations are limited, its wider group’s significant China investments raise concerns about potential indirect access or influence. This incident also signals increased scrutiny of AI applications in telecommunications sectors worldwide.
What to watch next
Negotiations between Anthropic and the White House continue regarding the potential reinstatement of Claude Mythos and Fable 5 access under strict conditions. How the company balances compliance with government export controls and its global partner relationships will be critical for ongoing AI commercialization.
Monitoring developments around further regulatory actions, including potential export controls on advanced AI models, is key. Additionally, the impact this decision may have on South Korean AI collaborations and broader telecom-industry AI innovation will be closely watched by market participants and policymakers.