Anthropic has disabled access to its most advanced AI models, Fable 5 and Mythos 5, following a directive from the U.S. government [1, 2].

This move signals a tightening of export controls on artificial intelligence, treating high-level machine learning capabilities as critical national security assets rather than commercial software. The restriction targets foreign nationals regardless of their physical location, including those residing within the U.S. [1, 3].

The company withdrew the models shortly after their launch to comply with the government order [1]. The directive focused on preventing foreign nationals from utilizing the flagship technology, citing national security and export-control concerns [1, 2].

"The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure …" a spokesperson for Anthropic said [2].

The scope of the ban is broad, affecting users globally [1]. Reports indicate the U.S. government specifically wanted these models blocked for all foreign nationals, even those currently in the U.S. [1].

Industry observers suggest this action reflects a shift in how the U.S. manages the proliferation of frontier AI models. A report from Yahoo News said the decision to block access shows the government increasingly views the technology as a national-security issue [4].

Anthropic has not provided a timeline for when or if these models will return to the global market. The company's decision to pull the models entirely, rather than implementing a granular filtering system, suggests the government's requirements for verification were too stringent for the company to maintain active service [1, 2].

“The U.S. government’s decision to block access to AI models for foreign nationals shows it increasingly views the technology as a national-security issue.”

This development marks a transition from voluntary safety commitments by AI labs to mandatory government intervention. By classifying specific model versions as export-controlled technology, the U.S. is treating AI capabilities similarly to advanced semiconductors or military hardware. This creates a fragmented global AI landscape where the most powerful tools are restricted by citizenship, potentially slowing international collaboration while accelerating a technological 'arms race' between superpowers.