Reports indicate that US officials became alarmed by alleged connections between the Chinese government and South Korean telecom giant SK Telecom. This concern reportedly triggered a crisis at AI firm Anthropic, leading to the shutdown of its experimental models, Claude Mythos and Fable. The White House instructed Anthropic to sever SK Telecom’s access to Mythos via the Project Glasswing partner program, a directive the company complied with immediately. While SK Telecom denied any direct links to China, the wider SK Group maintains significant business interests in the region and previously held a stake in the state-owned carrier China Unicom until 2009. Compounding the situation, Amazon and other entities subsequently flagged security flaws in Fable that could bypass safety restrictions. These combined issues eroded confidence in Anthropic’s risk management, forcing the White House to order both models completely offline.
This incident highlights how geopolitical tensions are increasingly influencing the deployment of advanced artificial intelligence systems within commercial networks. It underscores the difficulty for US tech firms in balancing innovation with national security concerns when dealing with international partners. The rapid shutdown demonstrates the extent of regulatory scrutiny facing AI developers and the potential for foreign policy to dictate technical access and safety protocols in real time.
- The White House ordered Anthropic to cut off SK Telecom access due to alleged ties between the carrier and China.
- Security flaws in the Fable model further reduced confidence in Anthropic’s safety measures.
- Both Mythos and Fable were taken offline completely as a result of these combined security and geopolitical concerns.
