Claude’s Fable 5 has been suspended by Anthropic after a direct order from Washington. The US government sent the directive late on Friday at 5:21 pm Eastern time. Officials cited national security authorities and singled out the company’s two most powerful models. You now see Claude Fable 5 and Claude Mythos 5 offline for all customers.
Why did the shutdown reach every single user? The directive applies to any foreign national, both inside and outside the United States. Anthropic cannot filter those users from American ones in real time across its systems. So the company turned off both models for everyone to stay within the law. Claude’s Fable 5 has been suspended worldwide, not only for the foreign users named. Access to its other models, including Claude Opus 4.8, stayed fully online without change.
Claude’s Fable 5 has been suspended over a reported jailbreak
Anthropic says the core worry centers on a possible Fable 5 jailbreak technique reported recently. The reported flaw lets a user push the model past a normal safety refusal. Engineers prompt the model to read a codebase and then point out software weaknesses. You find similar power in other public tools, including OpenAI’s GPT-5.5 model, the firm notes. Cybersecurity professionals rely on the same skills every day to defend their own networks. Anthropic argues the narrow risk does not match such a broad and sweeping recall.
An export control fight with real business stakes. The Anthropic export control directive lands at a delicate moment for the growing company. Anthropic plans a public stock listing this year and has built its name on safety. Claude’s Fable 5 has been suspended even though red teams found no universal jailbreak. OpenAI chief Sam Altman earlier mocked the whole approach as plain “fear-based marketing” in April. His point now stings a bit, since loud safety warnings drew sharp federal attention here.
What the Claude Mythos 5 shutdown means for you
The Claude Mythos 5 shutdown removes a tool many security teams used for defense. Anthropic had shared Mythos with about fifty vetted firms under a program called Glasswing. Those partners include Amazon, Apple, Google, Microsoft, and the cybersecurity firm CrowdStrike, among others. Many of those firms patch flaws in software you rely on every single day. Losing this access slows some defensive work, so the timing worries many security leaders. As I see it, the broad shutdown punishes ordinary users for a narrow technical risk.
The road back for the suspended models
Claude’s Fable 5 has been suspended for now, yet Anthropic wants access back fast. The company calls the order a misunderstanding and points to its strong layered safeguards. Independent classifier systems run apart from the model and still block the worst outputs. For now, you can move your workloads to Claude Opus 4.8 without much trouble. You should watch this Anthropic national security order, since it sets a wide precedent. The next federal ruling will shape how every AI maker ships powerful new models.
