The U.S. Commerce Department ordered export controls on Anthropic’s Claude Fable 5 and Mythos 5 on June 12, prompting Anthropic to disable both models worldwide within hours and turning a product launch into a test case for how quickly the government can remove a frontier AI system from global use.

At 5:21 p.m. ET on June 12, Commerce Secretary Howard Lutnick sent Anthropic CEO Dario Amodei a letter placing the two models under export controls, according to the source material. The order barred access by any foreign national, including foreign-national employees inside Anthropic. Anthropic shut off both systems for every customer by midnight after concluding it had no clean way to comply for only the affected users.

The shutdown came three days after Anthropic released the Mythos-class models on June 9. Fable 5 was the public, guard-railed commercial version; Mythos 5 was described as the more powerful underlying model, routed to selected organizations for cyber-defense work through Project Glasswing rather than released openly.

Anthropic said the letter cited national-security authorities but gave no specific rationale. The company called the action a misunderstanding and said its understanding was that officials had become aware of a jailbreak method affecting Fable 5. Anthropic argued that a narrow jailbreak was not enough reason to recall a deployed model, saying the system had already gone through thousands of hours of red-team testing by internal teams, government reviewers and third parties.

Switch-Off Risk Hits AI Buyers

The cost to the AI industry is less about one model outage than a new form of platform risk. Customers that built workflows on Anthropic’s latest systems saw that access could end globally with little notice, even for domestic users, if compliance cannot be narrowed by citizenship, location or employee status.

That precedent may push large buyers to split workloads across several model providers, keep fallback systems ready and treat regulation as part of uptime planning. It also gives open-weight and self-hosted systems a stronger pitch: once deployed inside a customer’s own infrastructure, access is harder for a vendor or regulator to revoke overnight.

The episode also lands as U.S. AI labs seek long-term enterprise contracts and investor confidence. If buyers view frontier models as politically exposed services rather than stable utilities, adoption can become a liability as well as a competitive moat.

From Launch To Shutdown

Anthropic released the Mythos-class models on June 9, 2026, positioning them for cybersecurity and biomedical work. Fable 5 was the public product, while Mythos 5 was tied to selected cyber-defense access through Project Glasswing.

The order arrived three days later. By June 14, more than 120 cybersecurity executives and engineers had signed an open letter asking Lutnick and National Cyber Director Sean Cairncross to lift the controls. Their argument was not that the models were harmless, but that they were not unique; they cited other U.S. and Chinese systems they said could perform similar security tasks.

“misunderstanding”

— Anthropic, June 12 statement

The Jailbreak Dispute Is Unsettled

The reason for the order is disputed. The U.K. AI Safety Institute’s red-team lead said his team produced a jailbreak that drew malicious responses from Fable 5 within hours and later extended it to multi-step agentic tool calls. The Wall Street Journal reported that Amazon warned officials after its researchers obtained information usable in cyberattacks, and Semafor reported that the directive was partly tied to suspicion that a China-linked group had obtained the model.

Anthropic disputes that those accounts justify a recall. The company says it had no evidence of a universal jailbreak after extensive testing. It remains unclear whether officials viewed the risk as model capability, a specific exploit, possible foreign access, reverse-engineering risk or a mix of those factors.

June 22 Talks Are Set

Anthropic and the White House are scheduled to meet on June 22, 2026. The immediate question is whether the controls will be lifted, narrowed or kept in place while Anthropic offers patches, access controls or new testing results.

AI buyers will be watching the policy signal that follows: whether export controls for frontier models remain a rare emergency tool or become a standing risk that procurement teams must price into contracts, continuity plans and provider choices.

Key Questions

What exactly did the U.S. order do?

It placed Claude Fable 5 and Mythos 5 under export controls and barred access by foreign nationals, according to the source material. Anthropic responded by disabling both models worldwide.

Why did Anthropic shut the models down for everyone?

The company concluded it had no clean technical path to block only the users covered by the order, including foreign-national employees, while keeping the models live for others.

Is the jailbreak confirmed?

A U.K. AI Safety Institute red-team lead said his team produced a jailbreak, and news reports describe separate concerns raised by Amazon and officials. Anthropic says the issue was narrow and not a universal jailbreak.

Does this affect other AI providers?

Directly, the order applied to Anthropic’s two models. Indirectly, it affects the wider market by showing that frontier AI access can be disrupted by export-control policy.

When could the models return?

No return date has been confirmed. The next known milestone is the scheduled June 22 meeting between Anthropic and the White House.

Source: Thorsten Meyer AI