Washington Orders Anthropic to Pull Fable 5 and Mythos 5

Anthropic's most capable models lasted three days in public before the federal government pulled them. On the evening of 12 June, the company received an export-control directive ordering it to suspend all access to Claude Fable 5 and Claude Mythos 5, citing national security. According to Anthropic's own statement, the order arrived at 5:21pm Eastern, and the company complied within hours.

The models had been the talk of the field since their public release on 9 June, when Anthropic described Fable 5 as the most powerful system it had ever put in front of users. By the night of the 12th, both were dark.

What the order does

The directive is sweeping in a way that left Anthropic little room to maneuver. It bars access to Fable 5 and Mythos 5 by any foreign national, whether located outside the United States or inside it, and that prohibition extends to Anthropic's own non-citizen employees. Because the company cannot cleanly separate that group from everyone else at the infrastructure level, it concluded it had to switch the models off entirely. As the statement put it, "the net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers."

The scope is narrow in one respect. Only the two Mythos-class models are affected. Opus 4.8, along with the Sonnet and Haiku lines, remains available and untouched. Anyone who built on those models can keep working. It is the new frontier tier, the one that drew the headlines last week, that has gone away.

The trigger: a jailbreak claim

The reason behind the order is where the story turns contentious. An administration official told Al Jazeera and other outlets that the Commerce Department acted after another company claimed it had jailbroken the model, getting it to analyse code for security flaws in a way that alarmed officials about cyber-weapon potential. Commerce Secretary Howard Lutnick is reported to have written to Anthropic chief executive Dario Amodei to inform the company the models would fall under export controls.

Anthropic says it reviewed the technique and was unimpressed. By its account, the method surfaced "a small number of previously known, minor vulnerabilities" that other AI models could find just as easily. The company called the finding a narrow jailbreak without broad impact, and said it disagrees "that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people."

Why this matters beyond Anthropic

Strip away the specifics and the precedent is the real story. If a single demonstrated jailbreak, of the kind that surfaces against every frontier model sooner or later, is enough to trigger an export-control recall, then no major lab can deploy a new model with any confidence that it will stay live. Anthropic made exactly that argument, warning that applying this standard across the industry "would essentially halt all new model deployments for all frontier model providers."

It also fits a tightening pattern around AI and national security. We saw the export-control lever pointed outward this week when Nvidia leaned on a gap in those same rules to sell CPUs into China. Here the lever is pointed inward, at an American company's flagship product. The common thread is a government increasingly willing to treat advanced AI as controlled technology rather than ordinary software.

Anthropic says it believes the order rests on a misunderstanding and that it is working to restore access as quickly as it can. For now, the most powerful models the company has ever shipped exist only in the past tense, and the rest of the industry is watching to see how fast, and on what terms, they come back.