This is a Ristretto Post [Short and Intense]:

On June 9th 2026, Anthropic launched two of its most powerful AI models ever: Claude Fable 5, described as the most capable model the company had released to the public and Claude Mythos 5, a variant with select safety guardrails lifted for vetted cybersecurity and biology researchers. Three days later, both were gone and banned.

What the US Government Did

On June 12 at 5:21 PM ET, Anthropic received an export control directive from the US Government ordering the immediate suspension of all access to Fable 5 and Mythos 5 by any foreign national, including Anthropic’s own foreign national employees, anywhere in the world. The directive cited national security authorities but provided no specific details.

Anthropic’s understanding, however, is that the government believed it had identified a method to jailbreak Fable 5, a technique to bypass its internal safety guardrails. Because reliably enforcing a foreign-national restriction at commercial scale was deemed operationally impossible, Anthropic suspended access to both models for all customers globally.

What Anthropic Said

Anthropic issued an official statement disagreeing with the action. Its position: “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. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”

The company noted that no testers had yet demonstrated a universal jailbreak, only non-universal techniques, which are a known vulnerability in every model in the industry. Anthropic reviewed the demonstration and characterized it as involving a small number of previously known, minor vulnerabilities.

What Actually Happened

Fable 5 was, by every benchmark available, a generational leap. Massive gains in software engineering, long-horizon autonomous tasks, scientific research, and vision. Mythos 5 doubled down on that capability for specialized professional use. The concern wasn’t theoretical, regulators appear to have been briefed on a specific exploitation path before the public was.

The episode marks the first time the US Government has invoked export control authority to pull a commercial AI model from the market mid-launch. It sets a precedent that rattles the entire frontier AI Industry.

What Comes Next

Anthropic is expected to challenge the directive and negotiate the terms for a potential re-release, likely with an enhanced access verification layer. The deeper question: whether any frontier AI model can now be deployed globally without US government sign-off first. If so, the race to deploy at the frontier just got a new player at the table. We didn’t see this coming, and just like many out there testing the new model, we have to say that works quite well when checking our agents safety guardrails that we update in production in a few hours.

Sources: Anthropic Official Statement