Washington’s move to lift curbs on Anthropic’s Fable 5 and Mythos 5 shows how fast advanced AI has become a national-security fight.
Quick Take
- The U.S. Department of Commerce had forced Anthropic to cut off access to the models for foreign nationals worldwide.
- Anthropic said the government acted after a reported jailbreak concern, but it called the problem narrow and not universal.
- The company said it disabled the models for all customers because it could not screen users by nationality in real time.
- Current reporting says the Commerce Department has now lifted those export curbs, ending the shutdown.
What Changed
Anthropic first pulled Fable 5 and Mythos 5 offline after a Commerce Department export-control directive hit the company on June 12. Reporting said the order covered any foreign national, including Anthropic’s own non-citizen employees, which meant the company had to block access for everyone to stay compliant. That decision turned a product launch into a federal policy dispute almost overnight.
Anthropic said the letter did not explain the full national-security case. The company said officials had pointed to a potential jailbreak method, but Anthropic called it a narrow issue tied to one specific use case. It also said similar capability could be found in other public models, which is why it disputed the need for a broad shutdown.
Why the Government Stepped In
The government’s action fits a wider pattern in Washington. Federal officials have already used export controls on chips, model weights, and other advanced technology when they believe national-security risks are high. In this case, the concern centered on access to powerful model behavior that could aid cyber abuse. That is the kind of risk that gets treated as a security issue, not just a business problem.
Still, Anthropic’s own account raises a key question about scope. The company said the government offered only verbal evidence of a “potential narrow, non-universal jailbreak,” and it said the same type of capability was already available in other models. For conservative readers frustrated by federal overreach, that matters. A government can protect security without making every new tool look like a public threat.
What the Lift Means Now
Current reports say Commerce has lifted the export curbs on the models, which suggests the immediate legal block is gone. That does not erase the bigger lesson. When the federal government can shut down access to frontier AI so quickly, companies and users are left living under rules that can change with little warning.
Three days after launch.
One government letter to Anthropic.
5:21pm on a Friday.Claude Fable 5, Anthropic latest AI model ever released to the public was pulled from every user on earth except United States.
No public explanation. No technical review. No appeal process. 👇… https://t.co/yP8vp6I2nH
— Nemo (@HeyNemoAI) July 1, 2026
For Anthropic, the result is a partial win. The company had been working to restore access and said it believed the order reflected a misunderstanding. For the public, the episode shows how much power Washington now claims over advanced AI systems, especially when officials invoke national security. That is why this case will keep drawing attention from lawmakers, regulators, and anyone worried about how far export controls can reach.
Sources:
businessinsider.com, labs.cloudsecurityalliance.org, digitalapplied.com, instagram.com, forbes.com, reddit.com, sanctionsnews.bakermckenzie.com, bis.gov, mintz.com
