Anthropic suspends access to Fable 5 and Mythos 5 after US government order, just one day after launch

Yesterday it was a launch story; today it is an access suspension story

The previous day, Game Sunakku first summarized the launch structure, limited access, and operational points for Fable 5 and Mythos 5 in Anthropic unveils Claude Fable 5, while Mythos 5 gets limited access. But less than 24 hours later, Anthropic issued a new statement saying it would suspend access to both models.

In other words, the situation changed from explaining “what is new” about the same models yesterday to asking today “whether those models can continue to be used.” That is a large enough shift to treat this not as a simple update, but as a separate follow-up article.

What Anthropic actually said

According to Anthropic’s statement, the US government issued an export-control directive to suspend access to Fable 5 and Mythos 5 on the basis of national security authority. As worded, the directive applies to all foreign nationals inside and outside the United States, including Anthropic employees who are foreign nationals.

However, Anthropic explained that, in order to comply with the rule, it ultimately had to suddenly disable access for all customers. At the same time, it drew a line by saying that access to other Anthropic models is not affected.

The government saw jailbreak concerns, while Anthropic sees the action as excessive

Anthropic wrote that it understands the government to have concluded that there are ways to bypass or jailbreak Fable 5. But the company argued that the demonstrations it saw were at the level of finding a few minor vulnerabilities that were already known, and that other public models could produce similar results.

Anthropic also emphasized that it conducted thousands of hours of red-team testing before the launch of Fable 5 with the US government, the UK AISI, and outside organizations, and that it has not yet identified a general-purpose jailbreak that broadly disables safety measures. In short, the risk assessments by the government and the company are now sharply diverging.

What general readers and developers should watch

For general readers, the important point is that news about the latest AI models no longer ends with a benchmark-score race. Even when performance is strong, the actual availability of a model can suddenly change because of policy, export controls, and safety judgments.

For developers, the issue is more direct. In the previous article, Fable 5 was already a model that required attention to refusal and fallback design. Today, the operational risk has gone one step further: the model itself can be taken down entirely. For services strongly tied to a specific model, this means preparing alternative models, reduced-function modes, and notice flows as well.

What still requires caution

The statement alone does not disclose the exact technical scope of the issue the government identified, or how long the directive will remain in place. Anthropic said it would share more details within 24 hours, but at this point it is not possible to say when access will be restored.

So rather than reading today’s news as “Fable 5 has been finally judged dangerous,” it is more accurate to see it as one of the first official cases showing that even a latest frontier model can be stopped by a regulatory shock immediately after launch.