13/06/2026
Right, so this is actually mad. The most powerful AI ever released to the public went live… and three days later the US government had it switched OFF. For everyone.
The model is Claude Fable 5. Anthropic launched it on June 9th and it instantly topped almost every benchmark out there. Then on June 12th, at 5:21pm, Anthropic got a directive from the US government , citing national security, an export-control order saying no foreign national, anywhere in the world, was allowed to access it.
Here's the catch: Anthropic can't realistically check the nationality of every single user. So the only way to actually comply… was to shut the whole thing down. For everybody. Hundreds of millions of people, cut off overnight.
So what triggered it? One demo the government was shown — a "jailbreak." Someone basically got Fable 5 to do the one thing it's meant to block: they handed it a codebase and asked it to find the security holes inside. (A jailbreak is just clever wording that sneaks past an AI's safety rules and gets it to do something it would normally refuse.)
Anthropic is pushing back hard. They say it's a narrow, non-universal jailbreak , not a master key that breaks every safeguard, and that finding flaws in code is something other AI models already do, used every day by security teams to keep systems safe. In their words, pulling a tool used by hundreds of millions over this one thing is "a misunderstanding," and they're working to restore access as fast as they can.
But here's why it actually matters: this has never happened before. The federal government has never reached in and pulled a live, publicly-deployed AI off the market like this. That's a huge line to cross, over who gets to decide which AI the world is even allowed to use.