US Export Controls Force Anthropic to Take Claude Fable 5 Offline

The US government imposed export controls on Anthropic's Claude Fable 5 and Mythos models on Friday, less than a week after their public release. The controls restrict foreign nationals, including those employed by Anthropic within the United States, from accessing the models. In response, Anthropic took both models completely offline for all users.

As of Tuesday, Fable 5 remains unavailable. Users opening Claude see a notification stating "Fable 5 is currently unavailable."

What Fable and Mythos Are

Mythos is the underlying model that powers both Mythos 5 and Fable 5. According to Anthropic, Mythos represents a "potential cyber-weapon" that could exploit security vulnerabilities. The company initially released Mythos Preview in April exclusively to enterprise clients and governments — what Anthropic calls "cyber defenders" — rather than the general public.

Fable 5, released last week alongside the full Mythos 5, is described as the "safeguarded" version of Mythos with additional protections built on top. It was the first Mythos-class model made available to the public. Anthropic skipped directly to version 5 for both models without releasing versions 1 through 4.

The Regulatory Conflict

The situation creates a significant irony: Anthropic has spent years arguing that AI models could become dangerous and that government regulation is necessary. The company's CEO Dario Amodei has publicly advocated for the government to "get serious about regulating AI sooner rather than later."

Now that regulation has arrived, Anthropic finds itself unable to operate under the restrictions imposed. The company reportedly cannot comply with the export controls while keeping the models online because it lacks technical mechanisms to adequately restrict access by nationality.

Timeline of Events

• April 2025: Anthropic releases Mythos Preview to select enterprise and government clients only
• Last week: Anthropic releases both Mythos 5 and public-facing Fable 5
• Friday: US government imposes export controls on both models
• Friday-Monday: Scrambled negotiations between Anthropic and government officials over the weekend
• Tuesday: Fable 5 remains offline with no resolution announced

What This Means

This incident marks the first major clash between US AI regulation and a leading AI company's operations. The situation raises critical questions about how the United States will regulate AI models going forward and whether that regulatory framework will function as a coherent safety regime or as an ad-hoc political tool.

The international implications are significant. Foreign governments, particularly China, are observing whether the US regulatory approach represents a serious safety framework or political weaponization against companies that don't align with the current administration.

For Anthropic, the crisis represents a challenge to its business model. The company positioned Mythos as enterprise-grade technology requiring careful handling, only to have its safeguarded consumer version immediately restricted. Whether Fable 5 returns and under what conditions will set precedent for how American AI companies navigate government restrictions on advanced models.

The standoff also demonstrates the technical challenges of implementing nationality-based access controls for AI models distributed through web interfaces and APIs. Unlike physical technology subject to export controls, AI models pose unique compliance challenges that neither companies nor regulators have fully solved.