Anthropic Disabled Fable 5 And Mythos 5 After A U.S. Export-Control Order. Here’s What Happened

What happened?

If you’ve been using Claude in the last week, you may have noticed two new model names. Fable 5 was the headline launch — Anthropic’s first general-release version of its Mythos model family, which the company had previously described as too powerful in the cybersecurity domain to release publicly. Mythos 5, the version with fewer restrictions, was available only to a vetted group of organizations through Anthropic’s Project Glasswing program.

Both are now gone.

“The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees,” Anthropic said in its official statement posted shortly after the order arrived Friday evening. “The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.”

All other Claude models — Opus, Sonnet, Haiku — remain available.

The Big ‘Jailbreak’

The trigger seemed to be a reported technique for bypassing Fable 5’s safety guardrails — the filters used to stop the model from accessing the advanced cybersecurity capabilities baked into the underlying Mythos architecture. The concern was that if those guardrails can be defeated, a consumer-facing AI product effectively becomes an unrestricted cyber tool.

According to Axios, the administration had tried to stop Anthropic from releasing the models at all, and failed. The export control order was the government’s next move.

A day after the shutdown, White House AI adviser David Sacks laid out the administration’s account in a post on X.

“A highly credible trusted partner of both Anthropic and the USG who was testing Fable came forward with a jailbreak of those guardrails,” Sacks wrote. “The Admin asked Dario to fix the jailbreak or de-deploy the model. Dario refused.” He called the administration’s response “reluctant” and said resolution was straightforward. “The ball is in Anthropic’s court.”

X (Formerly Twitter)David Sacks on X (formerly Twitter): “I’ve had a number of conversations with folks inside and outside government about the current situation with Anthropic, and here is what I believe to be true: – As we know, Anthropic publicly released its Mythos class models earlier this week under the commercial name Fable…. / X”

Anthropic broadly disputes these accounts — both the severity of the jailbreak and the characterization of its response. In its public statement, the company said it had reviewed what it believes to be the underlying report and concluded, “the level of capability demonstrated is available from other publicly deployed models, including OpenAI’s GPT-5.5, and is used by cybersecurity defenders routinely.”

“We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people,” Anthropic said. “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”

A source close to the company reportedly contended that Anthropic was not presented with any details and never refused to fix issues. The two accounts have not been reconciled as of this writing.

Amazon’s Fingerprints

The identity of the “trusted partner” who flagged the jailbreak has not been officially confirmed, but reporting by Semafor and the Wall Street Journal points to Amazon.

CEO Andy Jassy reportedly told Treasury Secretary Scott Bessent and other government officials that Amazon researchers used Anthropic’s Claude Fable 5 to obtain information that could be used in cyberattacks. The government subsequently imposed the export control ban.

Amazon is one of Anthropic’s largest investors and provides much of its cloud infrastructure, making its purported role in triggering the shutdown one of the more striking and interesting elements of the story.

“It’s not uncommon for governments to seek our counsel on potential security risks,” an Amazon spokesperson told Semafor, declining to share specifics.

Another angle reported is that the White House acted partly over suspicion that a China-linked group had accessed Mythos, raising fears that the model could be reverse-engineered or distilled by a foreign adversary.

Equally Complex Backdrop

Since early 2025, Anthropic and the Trump administration have been locked in a thorny conflict.

The Department of Defense previously designated Anthropic a supply chain risk following the collapse of talks between the two, and Anthropic filed a lawsuit challenging that designation — a case still active in court.

“Three months ago, @DeptofWar kicked @AnthropicAI out of our building — forever,” Defense Secretary Pete Hegseth posted on X earlier this week. “Every passing day proves why that was the right move.”

Meanwhile, Sacks pushed back on the idea that old grievances were driving the export control order. “Those trying to misdirect and tie this action to the prior DoW/Anthropic issues are wrong,” he wrote on X.

There’s also a notable timing detail. Just one day after launching Fable 5, Amodei published a major policy essay calling on the U.S. government to hold legal authority to block or reverse frontier AI models that fail independent safety testing. Two days later, the government used that authority against Anthropic.

The shutdown also lands at a commercially sensitive moment. Anthropic filed a confidential IPO prospectus with the SEC earlier this month, disclosing a revenue run rate of $47 billion and a valuation of $965 billion.

Now that a sitting government has shown it can switch off a widely used AI product, mid-deployment, on the basis of a security assessment that the company says is inaccurate, how that precedent plays out is something the industry will be watching very closely.

In his policy essay published just days before the shutdown, Amodei opened with an analogy for the pacing mismatch between AI and regulation. “In one of the side plots to The Lord of the Rings, two of the Hobbits attempt to rouse Treebeard — a wise but ponderous sentient tree — to defend his forest from an army that is cutting it down. The problem is that Treebeard operates at a very different speed than the Hobbits.”

“The intersection of AI and our political institutions feels a bit like the Hobbits and Treebeard,” Amodei wrote.

This week, Treebeard moved fast.