The US Commerce Department ordered Anthropic to suspend Claude Fable 5 and Mythos 5 on June 12, 2026, in the first AI model export ban.

The Commerce Department’s Bureau of Industry and Security issued the directive citing a jailbreak that enabled the Claude models to assist with cyberattacks.

What the Export Control Directive Said

Commerce Secretary Howard Lutnick signed the export order directing Anthropic to suspend all access by any foreign national to Fable 5 and Mythos 5.

The directive covered foreign nationals both inside and outside the United States, including Anthropic’s own foreign-national employees worldwide.

Per National Law Review’s analysis, the Bureau of Industry and Security cited a specific jailbreak enabling the Claude models to assist with sophisticated cyberattacks.

How Anthropic Responded Within Hours

Anthropic received the directive at 5:21 PM Eastern on June 12 and disabled both models for every customer worldwide within hours of receiving it.

Because Anthropic cannot reliably identify foreign nationals across its hundreds of millions of users in real time, a global shutoff was the only option.

As NBC News’s report reported, Anthropic publicly disagreed with the action, arguing the jailbreak cited is narrow and replicable using other available models.

Why Claude Fable 5 Triggered the Anthropic Export Ban

Claude Fable 5 and Mythos 5 are Anthropic’s two most powerful models, capable of autonomous zero-day exploit chaining across operating systems and browsers.

Claude Mythos Preview had already been restricted to roughly 50 developers under Project Glasswing since its April 2026 release through Anthropic.

The public Fable 5 launch on June 9 apparently brought capabilities broad enough that the government acted within three days of the public release.

Capabilities That Triggered the Ban

Claude Mythos 5 demonstrated the ability to autonomously discover and chain zero-day exploits across major operating systems and browsers in testing.

It could identify vulnerabilities in smart contracts and blockchain infrastructure and execute rapid security audits that previously required senior human researchers.

These capabilities relate directly to our coverage of top cybersecurity threats, where AI-assisted cyberattacks are now one of the top threats to US infrastructure.

What This Means for AI Regulation Going Forward

This is the first time a US government agency has used export control law to restrict access to a commercial AI model after public release.

Legal experts say the action signals that the Commerce Department views advanced AI models as dual-use technology subject to export restrictions.

The action connects to the Trump AI executive order of June 2, which created a voluntary frontier model review that Anthropic had not yet joined.

Which Claude Models Are Still Available

Claude Opus 4.8 and earlier Claude models remain fully available globally and are unaffected by the export control directive as of June 2026.

Anthropic has indicated it is working with the government to understand the exact scope of the jailbreak and whether targeted fixes could restore access.

Per Memeburn’s coverage, the AI industry is watching closely to see if this export control approach will extend to other frontier models from other labs.