The AI That Was Too Powerful for the World Just Got Too Powerful for Washington

The timing was extraordinary. On Tuesday, Anthropic launched Fable 5 to the public — its most powerful AI model ever made available commercially, accompanied by an extensive safety architecture and a 1,000-hour external red-teaming programme. On Friday afternoon at 5:21pm Eastern Time, the US government ordered it to shut down.

Not for everyone. Technically, only for foreign nationals. But since Anthropic cannot selectively block users by nationality in real time across a platform serving hundreds of millions of people, the practical result was identical: Fable 5 and Mythos 5, the two most capable AI models Anthropic has ever built, were disabled for every user on the planet — including Americans — within hours of the order arriving.

WHAT ACTUALLY HAPPENED

Commerce Secretary Howard Lutnick sent a letter to Anthropic CEO Dario Amodei saying that the Mythos 5 and Fable 5 models would be subject to export controls to any location outside the US and to all foreign persons within the country.

Anthropic said it received the US government export control directive on June 12, 2026 at 5:21pm ET ordering it to suspend all access to its two most capable models, Fable 5 and Mythos 5, for any foreign national.

The letter did not provide specific details of its national security concern. The company could not selectively block only foreign nationals in real time, so the practical result was a hard shutoff of both models worldwide.

The unexpected move comes just days after Anthropic announced Fable 5 and Mythos 5, two powerful models that the company touted as state-of-the-art across a number of different industry benchmarks.

THE JAILBREAK CLAIM THAT TRIGGERED EVERYTHING

An administration official told Axios the Commerce Department decided to take the action after another company claimed it was able to jailbreak Mythos, alarming the administration about possible national security risks. The administration tried to get Anthropic to pause releasing the latest models but was unsuccessful, the official said, prompting the export control letter.

Anthropic's response to this reasoning was pointed and specific. The company pushed back hard on the characterisation — arguing that the evidence presented does not justify a full commercial recall.

Our understanding is that the government believes it has become aware of a method of bypassing, or "jailbreaking" Fable 5. We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well.

Anthropic said: "Our understanding is that the government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5. However, 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."

THE COMPLIANCE THAT CAME WITH OBJECTION

Anthropic did not resist the order. It complied within hours. But it made its disagreement public and explicit — an unusual move for a company that has built its brand on working constructively with government.

In its statement, Anthropic said the government did not provide specific details about its national security concern. The company apologised to its customers for the disruption. "As we have stated publicly, we believe the government should have the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts. This action does not adhere to those principles."

The administration official said the model needs to remain locked down until the US government's national security apparatus is hardened, adding that could happen in the next few weeks.

The "few weeks" timeline is the most practically important detail for Anthropic's commercial business. Fable 5 and Mythos 5 had barely begun generating the enterprise revenue that their launch was supposed to unlock. Every week they remain offline is a week of foregone revenue — and a week where enterprise customers reconsider whether building on Anthropic's most capable models is a commercially stable decision.

THE DOD CONFLICT: A PATTERN, NOT AN ACCIDENT

The announcement marks Anthropic's latest run-in with the US government after a high-profile clash with the Department of Defense spilled into public view earlier this year. After negotiations between the two organisations collapsed, the DOD declared Anthropic a supply chain risk, meaning the company purportedly threatens US national security. The label has historically been reserved for foreign adversaries, and requires defense contractors to certify that they do not use Anthropic's technology.

The DOD conflict and the Commerce Department export control order are not unrelated events. They are part of a pattern — a government that has become increasingly uncomfortable with Anthropic's willingness to publish its most capable models despite official concern about what those models can do.

Anthropic's position throughout has been consistent: it believes it has built the safety architecture to make these models deployable, and that a narrow jailbreak finding does not justify a full commercial withdrawal. The government's position is equally consistent: it does not share that confidence, and it has the legal authority to act on its own assessment.

WHY THIS MATTERS BEYOND ANTHROPIC

The export control order creates a precedent that every AI company — OpenAI, Google DeepMind, Meta AI — is watching with considerable attention.

Watch for formal documentation from the Commerce Department, any narrowing or clarification of the directive, and how rival labs respond now that a frontier model can be pulled from the market by government directive rather than by the company's own choice.

That last clause is the most consequential part. Before Friday, the question of whether a frontier AI model was safe to deploy was answered by the company that built it, with input from external red-teamers and regulators. After Friday, it is clear that the US government can and will exercise unilateral authority to remove a commercial model from global markets — without detailed explanation, without a formal process, and with immediate effect.

For international users and enterprises — including the hundreds of millions of people who were using Fable 5 between Tuesday's launch and Friday's shutdown — the practical implication is that access to frontier American AI can be revoked at any moment for any national security rationale. The model that was launched with the promise of making advanced AI accessible to everyone is now inaccessible to everyone, because of a concern about some of them.

THE IPO SHADOW

The timing carries an additional layer of significance that financial markets will process carefully. Anthropic has confidentially filed for an IPO, targeting a public listing at a $965 billion valuation. The combination of the DOD supply chain risk designation and the Commerce Department export control order — both happening within weeks of each other — creates a regulatory risk profile that no investment bank's IPO roadshow was expecting to have to explain.

Being described as a national security risk by the Department of Defense while simultaneously having your most capable models pulled by the Department of Commerce is not the pre-IPO narrative any company would choose. The resolution of both disputes — and the timeline of that resolution — will now be among the most watched variables in the lead-up to what would otherwise be one of the decade's most anticipated technology listings.

Privately-held Anthropic, which has long positioned itself as a more responsible AI developer, first released its Mythos model in April to a limited group of business and government partners, while warning of its abilities to find cybersecurity bugs. The company that built its entire identity around responsible AI development is now in a public disagreement with the US government about what responsible deployment actually means.

The answer to that disagreement — and who wins it — will shape not just Anthropic's IPO but the entire future of how frontier AI is regulated in America.