Just days after a much-hyped launch, US AI company Anthropic was forced to announce a drastic measure on Friday evening: models Fable 5 and Mythos 5 have been disabled to comply with an export control directive from the US government citing “national security authorities.” The move hits the company in the middle of an already strained relationship with the Trump administration – and has a months-long backstory.
Fable 5 is currently the most powerful, if also the most expensive, AI model on the market and was supposed to become Anthropic’s new bestseller. That has now been put on ice.
Anthropic stated that on June 12, 2026 at 5:21pm ET it received an export control directive from the US government suspending access to Fable 5 and Mythos 5 for all foreign nationals – regardless of whether they are located inside or outside the United States, including foreign national Anthropic employees themselves.
Since the company cannot separate foreign nationals from the rest of its user base in real time, the only practical consequence was a complete global shutdown of both models. All other Anthropic models remain unaffected.
The official trigger: according to a government official who spoke with Axios, the Commerce Department decided to take action after another company claimed it had been able to jailbreak Mythos – putting the administration on high alert. The administration had previously tried to get Anthropic to pause the release of its latest models, but was unsuccessful.
Anthropic itself disputes the severity of the incident. The company says it received only verbal indications of a narrow, non-universal jailbreak. It is convinced that the same jailbreak could be applied to other publicly available models – including OpenAI’s GPT-5.5 – without those being subject to comparable national security export controls.
The confrontation between Anthropic and the Trump administration did not begin on Friday. The Pentagon designation came after negotiations to update Anthropic’s contract broke down over two red lines: the company insisted that its AI tool would not be used for mass surveillance of US citizens or for autonomous weapons systems. The Pentagon, however, insisted on being able to use the technology for “all lawful purposes.”
On February 27, 2026, the Department of Defense designated Anthropic a “supply chain risk” to national security and barred any US military contractor, supplier, or partner from conducting any commercial activity with the company. Defense Secretary Pete Hegseth went further, formally classifying Anthropic as a “supply chain risk” – a designation typically reserved for foreign adversaries and compromised technology vendors.
Shortly after the order, OpenAI announced it had struck a deal with the Department of Defense to deploy its AI models in classified networks. It was the first time an American company had ever been designated a supply chain risk – and the first time the designation appeared to be used in retaliation for a company refusing to agree to certain contractual terms.
Anthropic fought back: the company filed two lawsuits – one in a federal court in California, one in the federal appeals court in Washington.
The company is complying with the directive, but makes clear it believes it is wrong. In its statement, Anthropic says it pursued a “defense in depth” strategy for Fable 5 and tested its safeguards for thousands of hours together with the US government, the UK AISI, and private third-party organizations. The vulnerabilities found were all relatively simple and could be reproduced using other publicly available models.
Anthropic writes that it believes “the finding of a narrow potential jailbreak should not be cause for recalling a commercial model deployed to hundreds of millions of people.” If this standard were applied across the industry, it would effectively halt all new model deployments by every frontier provider.
The company apologized to its customers for the disruption and said it is working to restore access as soon as possible.
Aus Datenschutz-Gründen ist dieser Inhalt ausgeblendet. Die Einbettung von externen Inhalten kann in den Datenschutz-Einstellungen aktiviert werden:
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。