The US Government Just Pulled the Plug on Claude Fable 5 for Everyone

2 min read
All blog articles
The US Government Just Pulled the Plug on Claude Fable 5 for Everyone

Three days. That's how long Claude Fable 5 lasted between its public launch and a forced shutdown by the US government. On June 12 at 5:21 PM ET, Anthropic received an export control directive citing national security authorities, and within hours, its most capable publicly available model went dark.

Why the US government suspended Claude Fable 5

The directive bars all access to Fable 5 and Mythos 5 by any foreign national, including Anthropic's own foreign employees working inside the US. To ensure compliance, Anthropic had no choice but to disable both models for every customer worldwide.

The government believes someone found a way to jailbreak Fable 5. A researcher known as « Pliny the Liberator » had claimed a successful bypass shortly after launch, using multi-agent attacks, Unicode substitutions, and narrative framing techniques to defeat the model's safety classifiers.

Anthropic pushes back on the Fable 5 suspension

Anthropic is complying, but it is not staying quiet. The company argues the vulnerability is narrow and non-universal. The information obtained through the jailbreak was already publicly available and discoverable through other models. Applying this standard across the industry would, in Anthropic's words, "essentially halt all new model deployments."

Before launch, Anthropic conducted thousands of hours of red-teaming with the US government, the UK AISI, and private third parties. No tester found a universal jailbreak. Fable 5 shipped with a dedicated classifier layer that routes sensitive queries to Claude Opus 4.8 instead of responding directly.

A precedent that should worry every AI lab

All other Claude models remain fully accessible. But this story is bigger than Anthropic. If Washington can pull a commercial model used by hundreds of millions over a partial jailbreak, every frontier lab needs to ask: what stops this from happening to us next?

Anthropic had launched Claude Fable 5 as its flagship model on June 9. Microsoft had already restricted employee access over data retention concerns. This new suspension adds yet another layer of uncertainty for businesses that started building on the model. Anthropic says it will share more technical details within 24 hours and is working to restore access.

SN
Sarah Nakamura Sarah Nakamura writes about AI research breakthroughs, benchmarks, and technical advances for AIxploria.