Anthropic Faces Government Order Over New AI Models
Anthropic’s release of its newest AI models, Claude Fable 5 and Claude Mythos 5, quickly became the focus of a national security dispute with the U.S. government. The company introduced the models on June 9, describing Fable 5 as its most capable generally available system to date. Mythos 5 was described as using the same underlying model, but with some safeguards reduced in certain areas.
Three days later, on June 12, the government ordered Anthropic to block foreign access to both models, citing national security concerns. The directive reportedly applied to foreign nationals both inside and outside the United States, including some Anthropic employees. In response, Anthropic removed access for all users, saying it would comply with the legal order while disagreeing with the basis for recalling the models.
Reports indicate the order followed discussions between Amazon and White House officials. According to The Wall Street Journal, Amazon cybersecurity researchers found ways to prompt Fable 5 into providing information that could potentially assist cyberattacks. Anthropic has said the government did not provide detailed written evidence and argued that any reported vulnerabilities were limited and comparable to issues found in other advanced models.
Additional reporting from Semafor suggested that concerns about possible Chinese access to Mythos 5 may also have influenced the decision. The White House has not publicly confirmed that account, and Anthropic reportedly said China was not raised during its discussions with officials.
The episode comes as Anthropic is already involved in a separate dispute with the Pentagon over the military use of AI. It also highlights broader questions about U.S. control over frontier AI systems, export restrictions, and the dependence of global users on American technology companies. For governments and companies outside the United States, the shutdown may add urgency to efforts to develop domestic or regional AI alternatives.