The Korean Telecom Giant at the Center of Anthropic’s Mythos Controversy

The Trump administration ordered Anthropic to revoke access to its Claude Mythos and Fable 5 AI models for all foreign nationals, including immigrants in the U.S., following concerns about SK Telecom’s involvement. U.S. officials cited SK Telecom’s alleged ties to China as a key factor, though the company denied any connections. The decision came after Amazon researchers flagged vulnerabilities in Fable 5, raising White House skepticism about Anthropic’s ability to secure its advanced AI. Anthropic initially granted SK Telecom access to Mythos through its Project Glasswing program, which expanded to roughly 150 trusted organizations after collaboration with experts and the U.S. government. SK Telecom, South Korea’s largest wireless carrier, had previously invested $100 million in Anthropic in 2023. The White House demanded restrictions on Mythos and Fable 5 access to only U.S. nationals, but Anthropic chose to disable the models entirely rather than enforce nationality-based controls. Negotiations between the White House and Anthropic remain unresolved, with tensions persisting over whether to reinstate the models. Anthropic argued that SK Telecom’s access and Amazon’s findings were separate issues, as the government’s letter did not explicitly mention the Korean company or China. The company declined to comment further, while SK Telecom rejected claims of unverified ties to China in a statement to a Korean newspaper. The controversy highlights broader concerns about AI security and geopolitical risks, particularly as advanced models like Mythos demonstrate capabilities in identifying software vulnerabilities. The White House’s decision reflects growing scrutiny over foreign access to cutting-edge AI technology, raising questions about global collaboration in an increasingly competitive tech landscape.