Anthropic Disables Fable 5 and Mythos 5 Globally After Government Order Blocks Foreign Access

Since the Commerce Department's order earlier this week directing Anthropic to block foreign nationals from accessing Claude Fable 5, the company has taken a wider step: pulling both Fable 5 and Mythos 5 offline for every customer, domestic and foreign alike.

The government's order was aimed specifically at foreign access. Anthropic's problem is that it cannot surgically enforce that restriction. "The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance," the company wrote in a statement on its website, according to BBC News.

That means U.S. customers, including paying business accounts, lost access to the two most capable models in Anthropic's lineup. This is collateral damage from a geographically targeted directive.

The government's stated reason is a discovered jailbreak method. Anthropic says national security authorities have NOT identified specific vulnerabilities with any precision. "Our understanding is that the government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5," the company said in its statement, per BBC News.

Anthropic reviewed a demonstration of the alleged technique and reached a different conclusion. "These vulnerabilities all appear relatively simple, and we have found that other publicly available" — the BBC source was cut off at this point, but the company's public position has been that the risks are overstated and that comparable vulnerabilities exist across competing models.

That is a significant disagreement. The government says the jailbreak is serious enough to yank the models. Anthropic says it looked at the same demonstration and found minor, already-known issues.

Companies consistently underestimate the security implications of their own systems. The government may have intelligence about how adversaries intend to use the jailbreak that Anthropic does not have visibility into. A lab assessing its own vulnerability is an interested party. The national security community evaluating an AI model used globally is a different kind of evaluator.

Anthropic's self-description of Fable 5 as "too powerful" cuts both ways. If the model is as capable as the company claims, the gap between a "minor" jailbreak and a serious exploit may be narrower than Anthropic's engineers believe.

That said, as of June 13, 2026, U.S. national security authorities have not publicly identified the specific vulnerability, named the actors exploiting it, or presented evidence beyond the demonstration Anthropic itself reviewed. No charges have been filed. No public intelligence assessment has been released.

This shutdown is not happening in a vacuum. Anthropic and the Trump administration are already in active litigation over a separate order directing federal agencies to stop using Anthropic's AI tools, according to both BBC News and Observer BD. That lawsuit was ongoing before this week's foreign-access directive.

Two government actions against the same company within weeks, with an active lawsuit in between, raises a reasonable question about whether commercial competition and regulatory friction are factors alongside genuine security concerns. That question is currently unanswerable from public sources, and no investigation has been announced into the government's motives.

The third source provided, Republic World, covers the SpaceX IPO and contains no reporting on the Anthropic shutdown. It is not relevant to this story and should not be treated as corroborating or contradicting anything above.

BBC News approached the U.S. Department of Commerce for comment. No response had been published as of the BBC's reporting. Observer BD's June 13 coverage adds no new facts beyond what BBC News reported.

The unresolved question with real consequences: Anthropic's lawsuit over the federal-agency ban is still active, and this new shutdown gives both sides fresh material. Whether a federal court will weigh in on the Commerce Department's authority to order a private AI company to pull a domestic product from domestic customers — not just block foreign access — is an open legal question that has not yet been tested in court.