TL;DR
The U.S. Commerce Department placed Anthropic’s Claude Fable 5 and Mythos 5 under export controls on June 12, 2026, forcing the company to disable both models worldwide. The government has cited national-security concerns, while Anthropic and more than 120 security professionals argue the action overstates the risk and harms defenders. A White House meeting is scheduled for June 22.
The U.S. Commerce Department ordered Anthropic on June 12, 2026, to place Claude Fable 5 and Mythos 5 under export controls, prompting the company to disable both models for all customers worldwide and turning a major AI product launch into a national-security dispute with consequences for enterprise buyers, cloud partners and U.S. AI labs.
According to the source material, Commerce Secretary Howard Lutnick sent Anthropic CEO Dario Amodei a letter at 5:21 p.m. ET on June 12 barring access to the models by any foreign national, including foreign-national employees inside Anthropic. Anthropic said there was no clean way to apply that restriction across its deployed systems, so it disabled both models globally by midnight.
The order came three days after Anthropic released the Mythos-class models. Fable 5 was the public, heavily guarded commercial version, while Mythos 5 was described as the more powerful underlying model, made available only to selected organizations for cyber-defense work under Project Glasswing.
Anthropic has called the order a “misunderstanding,” saying its understanding was that officials had learned of a way to jailbreak Fable 5. The company said the model had been tested for thousands of hours by internal teams, the U.S. government, the U.K. AI Safety Institute and third parties without discovery of a universal jailbreak.
Washington just switched off
a frontier model
On June 12, an export-control order forced Anthropic to disable Claude Fable 5 and Mythos 5 worldwide. The security merits are still contested. The lesson buyers took away is not: frontier AI can be turned off.
■ The government’s case
- A reported jailbreak pulled malicious, agentic outputs (UK AISI)
- Amazon told officials Fable yielded cyberattack-usable info
- Suspicion a China-linked group obtained the model
- Proliferation & reverse-engineering risk to national security
▲ Anthropic & 120+ experts
- Calls it a narrow, non-universal jailbreak — a “misunderstanding”
- Capability is real but not unique (GPT-5.5, Opus, Kimi 2.7)
- Controls remove tools from defenders, not just attackers
- Export rules built for chips & ore don’t fit software
The precedent is the story. Whatever the jailbreak’s true severity, the U.S. showed it can dark a commercial American model worldwide on ~90 minutes’ notice. Adoption was supposed to be the moat — this week it became the exposure, and the likely winner is the open, sovereign, self-hosted stack.
Reliability Shock for AI Buyers
The immediate cost to the AI industry is not limited to Anthropic’s lost access or customer disruption. The order showed that a commercial frontier model can be removed from global service on short notice through national-security authorities, even after customers have begun relying on it.
That matters for companies treating frontier AI as infrastructure. Banks, software vendors, security firms and government contractors now have a fresh reason to question whether a single proprietary model provider can be trusted for high-dependency workflows. The source material cites Deutsche Bank’s reported reaction as a warning that buyers may conclude they “can’t rely on it.”
The order may also push customers toward multi-model deployments, self-hosted systems and open-weight models that cannot be switched off by a single provider. That could weaken one of the main business cases for U.S. AI labs: that broad adoption of closed frontier systems creates durable commercial lock-in.
Agentic AI Unleashed: A guide to designing, building, and deploying autonomous AI systems (English Edition)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
A Launch Became a Security Case
Anthropic launched the models on June 9, presenting them as advanced systems for cybersecurity and biomedical work. By June 12, federal officials had moved to restrict access. By midnight, Anthropic had disabled them for all users, including U.S. customers who were not the stated target of the foreign-national restriction.
The government’s stated rationale has not been made public in detail. The source material says the letter cited national-security authorities but did not give Anthropic a specific explanation. Reporting summarized in the source points to several possible factors: a jailbreak found by the U.K. AI Safety Institute, a separate warning from Amazon, and concern that a China-linked group may have obtained access to the model.
The policy fight now centers on whether export-control tools built for goods such as chips can sensibly govern deployed AI software. Supporters of the order appear to be focused on proliferation and misuse risk. Critics argue that broad controls can also remove defensive tools from legitimate users while rival open models remain available.
“misunderstanding”
— Anthropic
Architecting Enterprise AI Applications: A Guide to Designing Reliable, Scalable, and Secure Enterprise-Grade AI Solutions
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Evidence Still Unpublished
It is not yet clear what evidence most directly drove the Commerce Department’s decision. The public accounts cited in the source material do not fully align: one centers on a jailbreak found during testing, another on Amazon’s warning, and another on suspected access by a China-linked group.
It is also unresolved whether the reported jailbreak represented a narrow failure, as Anthropic argues, or a broader national-security risk. The company says no universal jailbreak was found after extensive testing. Government officials have not publicly released the full technical basis for the order.
The financial cost to Anthropic, its customers and partners is also not yet known. The shutdown could affect contracts, customer trust, future model access terms and investor views if major AI labs move toward public listings.
PRO-LAB Asbestos Test Kit – You Collect 2 Samples, We Analyze Them. Emailed Results Within 1 Week (5 Business Days) Includes Return Mailer and Expert Consultation. Lab Fee Included
Easy and Safe Testing: Utilize our asbestos testing kit to safely collect 2 samples for analysis. Simple to…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
June 22 Talks Set Stakes
Anthropic and the White House are expected to meet on June 22, 2026. That meeting may determine whether access to Fable 5 and Mythos 5 can be restored, whether narrower controls are possible, or whether the order becomes a precedent for future frontier-model restrictions.
AI buyers will be watching for a policy signal as much as a technical ruling. If the controls remain in place, enterprises may speed up plans to diversify model providers, keep fallback systems ready and evaluate open or sovereign AI stacks for sensitive work.
Hands-On Industrial Internet of Things: Build robust industrial IoT infrastructure by using the cloud and artificial intelligence
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did the U.S. government order Anthropic to do?
The Commerce Department placed Claude Fable 5 and Mythos 5 under export controls on June 12, 2026. Because the order barred access by foreign nationals anywhere, Anthropic disabled both models worldwide.
Why were the models restricted?
The full government rationale has not been made public. The source material cites reports involving jailbreak concerns, Amazon’s warning about cyberattack-usable information, and suspicion that a China-linked group may have obtained access.
What is Anthropic’s position?
Anthropic says the action was a “misunderstanding” tied to a narrow jailbreak issue, not proof that the model was broadly unsafe. The company says extensive testing did not find a universal jailbreak.
Why does this affect the wider AI industry?
The shutdown showed that a deployed frontier AI service can be removed globally on short notice. That may change how customers evaluate vendor risk, redundancy and reliance on closed models.
When could the situation change?
A meeting between Anthropic and the White House is scheduled for June 22, 2026. Until then, access terms, possible reinstatement and any revised controls remain unsettled.
Source: Thorsten Meyer AI
