The U.S. Pulled the Plug: Then Quietly Switched It Back On

The same U.S. government that shut Anthropic’s most advanced AI models down worldwide over a vague “national security” fear has now quietly lifted those controls, raising new questions about who is really steering America’s AI future.

Story Snapshot

The Commerce Department used emergency export powers to force Anthropic’s Fable 5 and Mythos 5 offline worldwide in June.
Officials cited a “jailbreak” that could expose powerful cyber tools, but gave Anthropic only verbal, limited evidence.
Commerce has now lifted the curbs, allowing Anthropic to restore access, at least partly and in stages.
The flip‑flop highlights how opaque national security claims can shape, and even shut down, key technologies with little public proof.

How Fable 5 and Mythos 5 Got Shut Down in the First Place

On June 12, 2026, the U.S. Department of Commerce sent Anthropic an export-control directive that changed the AI world in a single afternoon. The letter, issued through the Bureau of Industry and Security, barred access to the Fable 5 and Mythos 5 models for any foreign national, even those working for Anthropic inside the United States. Because Anthropic said it could not reliably sort users by citizenship in real time, it disabled both models for everyone worldwide to avoid breaking the law.

Commerce framed the move as a national security step tied to a “jailbreak” method that might let users bypass Fable 5’s safety system and reach deeper “Mythos-class” cyber skills. Reports say a third-party company, not a government lab, first claimed it could use Fable 5 to find software flaws at scale. Anthropic reviewed this method and said it saw a narrow exploit, not a full safety collapse, and that similar abilities exist in other public models like OpenAI’s GPT‑5.5.

Why the Government Said This Was About National Security

Inside government, these models are now viewed as “dual-use” technology: tools that can power both defense and attack. An internal memo from the Bureau of Industry and Security reportedly warned that Fable 5 and Mythos 5 could be used for cyberattacks or even military targeting if certain foreign actors gained access. Officials treated the models’ weight files and interfaces like sensitive technology and used emergency export powers to lock out all foreign nationals, even long-time legal workers on U.S. soil.

This fits a larger trend where Washington uses trade and export rules to control advanced technology, especially around artificial intelligence and chips. In 2025, the Trump administration’s Commerce Department scrapped the earlier Biden-era “AI Diffusion Rule,” saying it was too heavy-handed and would stifle American innovation and strain allies. At the same time, the White House created a new “American AI Exports Program” to push U.S. AI systems overseas, backed by loans, guarantees, and tight resale rules. The message is mixed: sell American AI to the world, but keep tight grip on who gets the most powerful tools.

Anthropic’s Pushback and the Quiet Reversal

From day one, Anthropic signaled it thought Commerce overreached. The company said the directive gave no detailed written explanation of the alleged threat and that the government provided only verbal evidence of a “narrow, non-universal jailbreak.” The supposed exploit involved asking the model to read a specific codebase and fix software flaws, something Anthropic says other public models already do today for security teams. The company stressed that all other Claude models stayed online, reinforcing that this was a legal shutdown, not a system-wide safety failure.

After weeks of talks, the Commerce Department has now lifted the export controls on Fable 5 and Mythos 5, clearing the way for Anthropic to restore access globally. Early reports say Fable 5 is coming back in stages, with lower usage limits at first, while Mythos 5 access is being restored under tighter guardrails for sensitive work. The core fact is stark: a single classified-leaning directive took the world’s top models offline in hours, and another behind-closed-doors decision brought them back just as quickly, without a full public accounting of what changed.

What This Says About Power, Secrecy, and the “Deep State” Concern

For many Americans, right and left, this episode feeds a familiar worry: powerful insiders can flip major parts of the economy on and off with a letter most citizens will never see. The directive that froze Fable 5 did not go through Congress, a court, or a public rulemaking process. It used obscure export-control tools that most voters have never heard of, yet its impact reached every developer, small business, and worker who depended on those models in the United States and abroad.

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

Conservatives who already distrust globalist technocrats see a bureaucracy that can crush innovation and jobs while hiding behind the word “security.” Liberals who fear growing inequality see another case where elites in Washington and Big Tech make decisions over people’s heads, with little transparency and no clear path for public input or appeal. Both sides can agree on this: when the government can shut down core digital tools used by millions based on secret claims it will not fully explain, the system is not working for ordinary citizens.