Anthropic has taken its two most advanced AI models — Claude Fable 5 and Mythos 5 — offline after the US government ordered limits on who could access them. The move bars foreign nationals, including some Anthropic employees, from using the systems, according to the New York Times.
At the heart of the dispute is a security vulnerability. Anthropic said it believed the US government had become aware of a potential means of jailbreaking Fable 5, according to the Associated Press. David Sacks, a senior White House AI adviser, said publicly that Anthropic CEO Dario Amodei had refused to "fix the jailbreak or de-deploy the model" after "a highly credible trusted partner" flagged the problem.
Amazon CEO Andy Jassy was among the tech industry figures who raised concerns with senior Trump administration officials about Mythos 5, helping set in motion the new export restrictions, according to The Information. The Wall Street Journal separately reported that Jassy's conversations with US officials triggered the crackdown.
The conflict has a military dimension as well. According to Reuters, the Pentagon clashed with Anthropic over the military's use of the company's AI technology. Defense Secretary Pete Hegseth warned Anthropic to allow the military to use its AI as it sees fit, AP sources say.
The White House is unlikely to extend the export restrictions to other AI companies, citing specifically Anthropic's refusal to fix the jailbreak, according to The Information — suggesting the action is targeted rather than a broad industry sweep.
The fallout is being felt globally. European political figures called the shutdown a "wake-up call" about the risks of depending on the United States for critical AI technology, according to the Associated Press. India, which accounts for Claude's second-largest user base after the US, is among the hardest-hit countries.
The episode sets a stark precedent: when a government decides an AI model is a security risk, it can effectively switch it off for the rest of the world.