Anthropic woke up Wednesday morning to find its services cut off across Europe, Asia, and most of the world. No warning. No explanation. Just gone.
The AI company behind Claude discovered the shutdown had come from the Trump administration's Commerce Department. They'd quietly added Anthropic to a restricted entities list—the kind of thing that actually kills your business overnight. The move prevents the San Francisco startup from serving customers outside the U.S., slicing off access to millions of users and potentially hemorrhaging millions in revenue. Anthropic found out about it the same way the rest of us did. Through news reports. Through regulatory filings. Nobody got a phone call.
Here's what makes this gnarly: Anthropic isn't a Chinese company. It's not stealing secrets. It's not even particularly cozy with the government in the way other AI firms are. The administration didn't accuse it of anything concrete or specific. The Commerce Department just... did it. (and yes, they really did this without warning). The order cited national security concerns around AI development, which is vague enough to mean almost anything. Basically a blank check for restricting whoever they want.
The timing. This comes right as the administration is pushing its own AI agenda. They're trying to position American companies as the dominant players in the global race. Anthropic's been growing fast—raised $5 billion last year, valued at $20 billion. It's also been more cautious than competitors about deploying its technology. That caution apparently didn't matter.
What mattered was that it exists.
The company's customers in the UK, Canada, Australia, and Japan are now locked out. Researchers who rely on Claude for their work can't access it. Startups building products on top of it have to scramble and hope for mercy. Anthropic has to figure out whether it can actually survive as a U.S.-only business or if this signals something larger—a pattern, a warning, a new normal.
Watch what happens next. Anthropic will probably sue the administration, arguing the restrictions are overreach or procedurally corrupt or both. The administration will probably double down, cite more national security mumbo-jumbo, maybe expand the list. And every other AI startup that thought it was operating in some kind of safe zone? They're going to start losing sleep, wondering if their turn comes next week or next month. You've got to wonder if anyone's actually safe anymore.