TL;DR
The U.S. Commerce Department placed Anthropic’s Claude Fable 5 and Mythos 5 under export controls on June 12, forcing the company to disable both models for all customers. The government’s rationale remains disputed, but the episode shows that frontier AI services can be cut off quickly by national security orders.
The U.S. Commerce Department placed Anthropic’s Claude Fable 5 and Mythos 5 under export controls on June 12, forcing the company to disable both AI models for customers worldwide and raising new concerns about whether frontier AI can be treated as a dependable global utility.
Commerce Secretary Howard Lutnick sent the order to Anthropic CEO Dario Amodei at 5:21 p.m. ET on June 12, according to the source material. The directive barred access by foreign nationals anywhere, including foreign-national employees inside Anthropic, leaving the company without a clean way to limit access by geography or customer type.
Anthropic disabled both models for all customers by midnight. The move came three days after the June 9 launch of the Mythos-class systems. Fable 5 was the guarded commercial model, while Mythos 5 was a more powerful system made available to selected organizations for cyber-defense work through Project Glasswing.
Anthropic said the order cited national-security authorities but did not give a specific rationale. The company called the action a “misunderstanding” and said it believed officials were responding to reports of a jailbreak affecting Fable 5.
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.
AI Buyers Face Switch-Off Risk
The immediate cost is not limited to Anthropic. The order gives enterprise customers, banks, defense contractors and software firms a concrete example of a frontier model being removed from service with little notice. That changes how buyers may evaluate dependence on closed commercial models.
The source material says Deutsche Bank treated the episode as proof that switch-off risk is no longer hypothetical. For companies building workflows around one model provider, the lesson is direct: access can depend on government decisions as well as pricing, uptime and model quality.
The dispute may also help open-weight and self-hosted AI systems. If customers believe hosted U.S. frontier models can be withdrawn under export rules, some may favor systems they can run inside their own infrastructure, including foreign open-weight models.
AI Engineering: Building Applications with Foundation Models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Launch To Shutdown
Anthropic launched the Mythos-class models on June 9, positioning them for cybersecurity and biomedical work. The systems were described as among the most capable models the company had shipped.
By June 12, the models had become the subject of a national-security order. The source material says the U.K. AI Safety Institute, Amazon and U.S. officials raised separate concerns about Fable 5’s cyber capabilities and possible exposure to a China-linked group. Those accounts have not been fully reconciled.
More than 120 cybersecurity executives and engineers later signed an open letter asking U.S. officials to lift the controls. Their argument was that the models were powerful but not unique, and that blocking access could also remove useful tools from defenders.
“misunderstanding”
— Anthropic
Access Tools RCBMHD Heavy Duty Button Master
Item model number : ?RCBMHD
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Rationale Still In Dispute
It is not yet clear which concern drove the order. The source material cites several accounts: a reported jailbreak found by the U.K. AI Safety Institute, an Amazon warning about cyberattack-usable outputs, suspicion that a China-linked group obtained access, and a White House request for Anthropic to patch or pull Fable 5.
Anthropic disputes the apparent basis for the action, saying a narrow jailbreak is not the same as a universal failure. The company also says the model had undergone extensive red-teaming by internal teams, government testers and outside reviewers.
The financial cost to Anthropic and its customers has not been disclosed. It is also unclear whether the export controls will be narrowed, lifted or applied to other frontier models.
AI-Powered Cybersecurity: How Artificial Intelligence is Revolutionizing Digital Defense, Threat Detection, and the Future of Online Security
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
June 22 Talks Loom
Anthropic and the White House are scheduled to meet on June 22. That meeting is expected to shape whether Claude Fable 5 and Mythos 5 remain offline, return under new restrictions or become the first example of a broader export-control approach for frontier AI software.
AI customers will be watching for a practical compliance path. Without one, the industry may face a new procurement reality: model capability matters, but revocability now matters too.
AI Resilience: A Control Plane for Governing, Testing, and Proving Enterprise AI Survivability
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What happened to Claude Fable 5 and Mythos 5?
The Commerce Department placed both models under export controls on June 12, and Anthropic disabled them for all customers worldwide by midnight.
Why did the U.S. government act?
The stated legal basis involved national-security authorities, but the specific rationale remains disputed. Reported concerns include jailbreak results, cyberattack-usable outputs and possible access by a China-linked group.
Did Anthropic agree with the order?
No. Anthropic called the situation a “misunderstanding” and argued that a limited jailbreak would not justify recalling a model already in use.
Why does this matter beyond Anthropic?
The case shows that closed frontier AI models can be disabled quickly by government action. That may push customers to diversify providers or use self-hosted systems.
When could the models return?
No return date has been confirmed. Anthropic and the White House are scheduled to meet on June 22, 2026.
Source: Thorsten Meyer AI
