The US government just forced Anthropic to switch off its two most powerful models. Here is what actually happened, and why it matters more than the headline.

On Friday evening, Anthropic disabled public access to its two most capable AI models, Fable 5 and Mythos 5, to comply with an export control directive from the US government. This is, as far as anyone can tell, the first time a leading AI company has pulled a publicly deployed model off the shelf because of direct federal intervention. We want to lay out exactly what happened, what each side is saying, and what it means for anyone who actually builds on these tools.

**What happened, in order.**

Anthropic released Fable 5 on June 9 as the first Mythos-class model available to the public, with safeguards that route high-risk queries in areas like cybersecurity and biology to a less capable model. Mythos 5, the same underlying model with some safeguards lifted, stayed limited to a small set of partners through a program called Project Glasswing. Three days later, on June 12 at 5:21pm Eastern, Anthropic received a letter from the government citing national security authorities, ordering it to suspend all access to both models by any foreign national, whether inside or outside the United States, including its own foreign national employees.

Here is the catch that turned a narrow order into a total shutdown. There was no clean way to block only foreign nationals, so to comply, Anthropic disabled both models for every customer. If you had a session open, it ended in errors, and new queries were quietly routed to older models like Opus 4.8. Every other Anthropic model stayed available. Amazon also revoked access to both models on Bedrock to support compliance.

**What the government is saying.**

The directive did not include specific details of the national security concern. Anthropic's understanding is that the government believes it found a way to bypass, or jailbreak, Fable 5. Notably, as of now the only evidence Anthropic says it has received is verbal, with no written technical detail provided.

**What Anthropic is saying.**

Anthropic is complying, but publicly disagrees. Its position is that the vulnerability is narrow, that the issues it was shown were minor and already findable in other public models, and that recalling a commercial model used by hundreds of millions of people over a narrow potential jailbreak is not a proportionate response. The company called it a likely misunderstanding, apologized to customers, and said it is working to restore access, with more technical detail promised soon.

**The WhatAI editorial take.**

The jailbreak is the part everyone will argue about, and it is the part that matters least to most of you. Whether the specific bypass is serious or trivial will get sorted out in the coming days. The real story is the precedent. A government can now switch off a deployed frontier model overnight, for every user at once, on the strength of a phone call. That is a new fact about the world that every company building on cloud-hosted frontier AI has to absorb.

If your product, your workflow, or your business depends on a single top-end model from a single provider, this weekend was a live demonstration of your exposure. The lesson is not panic, it is design. Keep a fallback model you can switch to. Know which provider and which jurisdiction your critical workloads sit in. Treat model access as something that can change without warning, because it just did.

There is a real and reasonable case on the other side too, and we are not going to pretend otherwise. A Mythos-class model has, by Anthropic's own description, the strongest cybersecurity capabilities of any model in the world, and the government has a legitimate interest in not having that misused. The honest disagreement is about process, whether a recall this sweeping, on verbal evidence, with no transparent standard, is the right way to exercise that interest. That is the debate worth having, and we would rather host it than shout a verdict.

**Sources, so you can read the primary material yourself:**

- [Anthropic's own statement on the suspension](https://www.anthropic.com/news/fable-mythos-access)

- [Anthropic's original Fable 5 and Mythos 5 launch post](https://www.anthropic.com/news/claude-fable-5-mythos-5)

- [CNBC report](https://www.cnbc.com/2026/06/12/anthropic-disables-access-to-fable-5-and-mythos-5-to-comply-with-government-directive.html)

- [VentureBeat, with the enterprise angle](https://venturebeat.com/technology/anthropic-blocks-all-public-access-to-claude-fable-5-mythos-5-following-us-government-order-what-enterprises-should-do)

- [NBC News on the original launch](https://www.nbcnews.com/tech/security/fable-5-anthropic-release-public-mythos-claude-model-rcna349104)

- [The New Stack, noting the evidence so far is verbal](https://thenewstack.io/us-gov-orders-anthropic-to-pull-fable-5-and-mythos-5-three-days-after-launch/)

Related on WhatAI: [Claude](/tool/claude) and [AI Models](/models)

Over to you. Is this the government doing its job and protecting people from a dangerous capability, or is it overreach that sets a frightening precedent for who controls the tools we all rely on? And separately, did this change how you think about building on a single model? Tell us below.