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 systems for all customers worldwide. The government’s security rationale is disputed, but the shutdown has created a new concern for AI buyers: frontier models can be removed from service quickly by regulatory order.
The U.S. Commerce Department placed Anthropic’s Claude Fable 5 and Mythos 5 models under export controls on June 12, forcing the company to disable both systems worldwide and turning a frontier AI product launch into a national-security dispute with broad consequences for AI customers, investors and competing model providers.
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 any foreign national, including foreign-national employees inside Anthropic. Anthropic concluded it had no clean way to separate access fast enough and disabled both models for all users by midnight.
Anthropic had released the Mythos-class models on June 9. Fable 5 was the public commercial model, while Mythos 5 was a more powerful system routed to selected organizations for cyber-defense work through Project Glasswing. Anthropic said the Commerce letter cited national-security authorities but did not give a specific rationale. The company publicly described the episode as a “misunderstanding” and said it believed officials were reacting to a reported method for jailbreaking Fable 5.
The government’s exact basis has not been publicly settled. The source material cites reporting from The Wall Street Journal that Commerce was alarmed by a separate report from Amazon, whose researchers allegedly obtained cyberattack-usable information from Fable 5. It also cites Semafor reporting that officials suspected a China-linked group may have obtained the model, creating concern about reverse-engineering and proliferation.
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.
Model Reliability Takes A Hit
The shutdown matters because it changes how large companies may price the risk of relying on closed frontier AI systems. The main lesson for buyers is not only whether Fable 5 or Mythos 5 posed a security risk, but that access to a leading U.S. model can be withdrawn globally with little notice.
That risk lands at a sensitive moment for the AI industry. U.S. labs have argued that adoption, integrations and enterprise contracts will create durable value around their models. The Anthropic order points to a different exposure: the more customers depend on a centralized model, the more disruptive a sudden government restriction can be.
The source material says Deutsche Bank framed the concern as “Can’t rely on it,” a shorthand for enterprise customers now adding regulatory shutdown risk to procurement decisions. That could push more buyers toward multi-model contracts, self-hosted systems and open-weight models that cannot be switched off by one vendor.
Train It. Tame It. Teach It.: Build Your Personal AI Team and Get Every Model to Work Your Way (The AI Practitioner's Edge)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
A Three-Day Product Collapse
The timeline was unusually compressed. Anthropic launched the models on June 9, received the Commerce order on June 12 and had disabled access globally by midnight that same day. On June 14, more than 120 cybersecurity professionals signed an open letter asking officials to lift the controls, according to the source material.
Anthropic’s defense rests on the claim that the alleged weakness was narrow rather than universal. The company said Fable 5 had been tested for thousands of hours by internal teams, the U.S. government, the U.K. AI Safety Institute and outside reviewers without discovery of a universal jailbreak.
Critics of the ban, including signers of the open letter, argued that the models’ security capabilities are not unique. They pointed to other frontier and open-weight systems, including GPT-5.5, Claude Opus, OpenAI’s Daybreak and Kimi 2.7, as systems capable of comparable cyber work. That claim remains part of the policy dispute, rather than a settled technical finding.
“misunderstanding”
— Anthropic
UHPPOTE 2.4GHz WiFi Wireless RF Remote Control Door Access Control System
✅ The main feature of this kit is that it allows you to open the door simply by…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Security Case Still Disputed
It is not yet clear which allegation drove the Commerce order. The public record described in the source material includes several possibilities: a reported jailbreak found by the U.K. AI Safety Institute, a separate Amazon security report, concern that a China-linked group obtained the model, and broader fears about reverse-engineering.
It is also unclear whether the models pose a risk greater than comparable systems already available elsewhere. Anthropic and the open-letter signers say the relevant capabilities are not unique. Government officials and outside testers cited in the source material appear to believe the risk was severe enough to justify emergency export controls.
No public technical record has resolved whether the jailbreak was limited, patchable or broad enough to justify a global shutdown. The effect on Anthropic’s customers, contracts and possible public-market plans has also not been fully disclosed.
Hands-On Artificial Intelligence for Cybersecurity: Implement smart AI systems for preventing cyber attacks and detecting threats and network anomalies
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
White House Talks Set
Anthropic and White House officials are scheduled to meet on June 22, according to the source material. The immediate question is whether Commerce will lift, narrow or maintain the export controls on Claude Fable 5 and Mythos 5.
The larger policy question is whether U.S. export-control rules built for physical goods can be applied cleanly to frontier software models. Until officials clarify the standard, AI labs, cloud providers and enterprise buyers are likely to treat regulatory shutdown risk as a live factor in model selection.
AI-Powered Cybersecurity: AI Tools for Enterprise Security | AI for Network Security | AI Risk Management | AI in Cyber Policies | Cyber Threat Management AI | ML in Fraud Prevention
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. Anthropic then disabled access worldwide because the order barred access by foreign nationals and the company said it could not comply selectively in the time available.
Why did the U.S. government act?
The full rationale has not been made public. The source material cites concerns about jailbreaks, cyberattack-usable outputs, possible access by a China-linked group and reverse-engineering risk.
Is Anthropic saying the models were unsafe?
No. Anthropic described the action as a “misunderstanding” and said it believed officials were responding to a narrow jailbreak rather than a universal failure of model safeguards.
Why are AI customers concerned?
The shutdown showed that a frontier AI service can be removed from global use quickly by government order. That may push buyers to use multiple providers or self-hosted systems to reduce dependence on any one closed model.
When could the order change?
Anthropic and White House officials are scheduled to meet on June 22. Any change depends on Commerce Department action and the outcome of the security dispute.
Source: Thorsten Meyer AI
