Written By | Torch Media Africa
Claude Fable 5 was designed as the most capable public AI ever — leading benchmarks in software engineering, scientific reasoning, and complex multi-step tasks.
Search
NEW YORK
They were the most powerful AI models ever released to the public. They lasted exactly three days.
On June 9, Anthropic launched Claude Fable 5 and Claude Mythos 5 to massive fanfare. By June 12, the U.S. Department of Commerce ordered both models offline, citing national security. By 5:21 p.m., they were gone. The move was unprecedented — and the alleged “dangerous jailbreak” at its center, Anthropic argues, is something other freely available models already do without any bypass.
The Launch
Claude Fable 5 was designed as the most capable public AI ever — leading benchmarks in software engineering, scientific reasoning, and complex multi-step tasks. Claude Mythos 5, the same model without safeguards, was offered only to vetted researchers under a program called Project Glasswing. Hundreds of millions of users signed up. The industry was thrilled.
The Shutdown
On June 12, Commerce Secretary Howard Lutnick invoked emergency export control powers, ordering Anthropic to block access for any “foreign national” — including non-U.S. citizens living in America and even Anthropic’s own overseas employees. Unable to verify nationality at scale, the company had no choice but to pull both models for everyone.
The Jailbreak That Wasn't
The government cited evidence of a jailbreak — a method to bypass Fable 5’s safety controls. Anthropic’s own investigation found the technique essentially involved asking the model to read a codebase and identify security flaws.
“The potential jailbreaks disclosed to us are either entirely benign or minor findings that provide no special capability,” Anthropic stated.
Crucially, other freely available models — including OpenAI’s GPT-5.5 — can produce the same vulnerability information without any jailbreak at all.
Anthropic Fights Back
In a remarkable public statement, the company said it was complying but explicitly disagreed. “If this standard applied across the industry, it would halt all new model deployments,” Anthropic wrote, adding that the government’s action lacked transparency and technical justification.
Global Fallout
Non-U.S. users lost access overnight. Developers who integrated Fable 5 were left scrambling. Inside Anthropic, foreign-national engineers were barred from using the model they helped build. Analysts warned the precedent could fragment global AI, driving top talent to jurisdictions with fewer controls.
What’s Next
Anthropic is working to restore access and plans to share more technical details. Legal challenges are possible. Meanwhile, the rest of the AI industry is watching closely: if the U.S. can recall a model this quickly over such narrow evidence, no frontier AI is truly safe from a shutdown.
The age of artificial intelligence, this week made clear, will be shaped not just by code — but by borders and politics.
