Anthropic forced to turn off its most powerful Ai models: global stop for national security reasons

The American artificial intelligence startup Anthropic announced the immediate suspension of access to its two most advanced models, Fable 5 and Mythos 5, just three days after their commercial debut. The decision was not made …

Jack Williams

Written by: Jack Williams

Published on:

Anthropic forced to turn off its most powerful Ai models: global stop for national security reasons

The American artificial intelligence startup Anthropic announced the immediate suspension of access to its two most advanced models, Fable 5 and Mythos 5, just three days after their commercial debut. The decision was not made by the company, but imposed by the US government through a national security directive.

According to the company, Washington has ordered it to block “any foreign national” from using the two systems, regardless of whether they are in the United States or abroad. The measure would even affect Anthropic’s foreign employees. Faced with the technical impossibility of immediately verifying the nationality of all users, the company chose a drastic solution: completely deactivating Fable 5 and Mythos 5 for all customers in the world.

The directive would have been sent by the US Secretary of Commerce, Howard Lutnick, with a letter delivered to the CEO of Anthropic, Dario Amodei. The government fears that the models could be misused after a company reportedly demonstrated the ability to bypass some of its security systems. The American administration in particular considers Mythos 5 to be a system with very advanced capabilities in identifying software vulnerabilities and IT flaws, characteristics that could transform it into a potentially useful tool for offensive operations in cyberspace.

The Anthropic protest

Anthropic openly contested the government’s decision, claiming it had not received detailed explanations of the alleged threats to national security. The company claims that the circumvention technique shown to the authorities would have allowed it to identify only “a limited number of minor and already known vulnerabilities”. “Based on the information we have seen, this is a misconception,” the company claims, that applying this standard across the entire industry would effectively block the release of any new advanced AI models. The company also said it was working to find a solution that would allow access to the models to be restored “as soon as possible”.

A precedent that can change the sector

The case represents one of the first cases in which the United States applies export controls not only on hardware, such as artificial intelligence chips, but directly on the Ai models themselves. For many observers, this is a historic step that could redefine the way Washington considers frontier technologies: no longer just commercial products, but truly national strategic assets. The decision also comes at a time of tension between Anthropic and Donald Trump’s administration, already divided on various aspects of the regulation and military use of artificial intelligence.

Jack Williams

Jack Williams

Jack Williams established himself as the editor-in-chief of The Vermilion with rigorous journalistic traits and an incessant quest for truth. His career began as an investigative reporter, where he quickly gained a reputation for his in-depth research and insightful articles. Williams takes his role at The Vermilion very seriously. Since he became editor-in-chief, he has set an editorial line based on impartiality, factual accuracy, and context relevance; this has established The Vermilion as a reference for political, economic, and global news. Under his leadership, the media continues its mission of reporting news with accuracy and equanimity, giving a voice to those who are often overlooked by large-scale media. Jack dedicates himself to fulfilling this mandate with passion, rigour, and unwavering determination.

Previous

Who owns the banks and who really lends money to Italians