Anthropic didn't just disable its most powerful AI model last week. It detonated a geopolitical bomb.
When the San Francisco startup announced it was "brutally deactivating" Claude Mythos 5 and Fable 5 — just three days after launch — the tech press treated it as a compliance story. A company following government orders. A bureaucratic hiccup in the AI race.
Europe saw it differently.
"La guerre de l'IA a commencé," declared Le Monde. The AI war has begun. Not between companies. Between continents. Between a superpower that controls the frontier models and everyone else who depends on them.
The Shutdown
The order came from Washington at 5:21 PM local time on June 12 — 11:21 PM in Paris. The Trump administration, citing "national security" and export controls, demanded Anthropic cut access to Mythos 5 and Fable 5 for "all foreign nationals, inside or outside the United States." Including Anthropic's own foreign employees.
The reason? A user had reportedly found a way to bypass some of Fable 5's safety guardrails. Not a breach. Not an attack. A workaround. And for that, the Commerce Secretary — reportedly Howard Lutnick — ordered a global shutdown of two models that hundreds of millions of users had just gained access to.
Anthropic called it a "misunderstanding." The company argued that if this standard were applied across the industry, "it would essentially halt all new deployments of frontier AI models." But Anthropic complied anyway. Within hours, Mythos 5 — the model touted as capable of detecting and exploiting security vulnerabilities with "unprecedented speed and acuity" — went dark.
For American users, it was an inconvenience. For European governments, it was a wake-up call.
Europe's Dependency Crisis
France's cybersecurity agencies had been counting on Mythos 5. Le Monde revealed that Orange and Crédit Agricole were in discussions to access the model for cybersecurity operations. No French government service had access yet — but the plan was clear: use American AI to protect French infrastructure.
Now that plan is in tatters. Not because the model failed. Not because it was hacked. Because the US government decided, unilaterally, that Europeans shouldn't have it.
The timing is brutal. Just days before the shutdown, Anthropic had released a "bridled" version of Mythos — Fable 5 — specifically designed for public use with safety guardrails around cybersecurity and biological/chemical risks. It was the first time Anthropic had opened its most advanced model class to the public. European companies were lining up.
And then the door slammed shut.
"The blocking of Anthropic's latest model by the United States worries France and Europe," Le Monde reported. Not because Europeans fear the model. Because they fear the power to take it away.
The Sovereignty Question
This is the nightmare scenario European policymakers have been warning about for years. American AI companies control the frontier models. American regulators control the American AI companies. And when American national security interests conflict with European economic or security needs, the US doesn't hesitate.
The EU's Digital Markets Act, the AI Act, the push for "digital sovereignty" — all of it looks like paper tigers now. Europe can regulate how AI is used. It can fine companies for non-compliance. But it can't force an American company to keep serving European customers when the US government says no.
France's response has been telling. The same week as the Anthropic shutdown, Le Monde reported that ChapsVision — a French data analysis company — is replacing Palantir at France's DGSI intelligence service. The message is clear: if American tech can be switched off at will, France needs alternatives.
But alternatives to Mythos 5 don't exist in Europe. Mistral is making progress. ASML dominates chip manufacturing. But frontier AI models? That's an American monopoly, with China racing to catch up. Europe is caught in the middle, dependent on one, threatened by the other, capable of building neither.
🔥 Hot Takes
1. The AI Cold War just turned hot. This isn't about trade policy or tariffs. It's about whether one country can unilaterally deny access to the most advanced technology on Earth based on a single user's workaround. The Trump administration didn't negotiate with allies. It didn't consult NATO. It issued an order, and a $20 billion company complied within hours. That's not a partnership. That's a vassal relationship. And Europe just realized it.
2. Anthropic's "bridled" model was a trap. By releasing Fable 5 with safety guardrails and then shutting it down when someone bypassed them, Anthropic proved that "safe AI" is an illusion. The guardrails were never the point. The point was demonstrating control — and then showing that control can be revoked at any time. European companies that built workflows around Fable 5 learned the hard way: American AI is a rental, not a purchase. And the landlord can evict you.
3. China's AI strategy looks smarter every day. While Europe debates regulation and America weaponizes export controls, China is building its own models, its own chips, its own ecosystem. The Zhipu AI trillion-dollar valuation, DeepSeek's $50 billion raise, ByteDance's Seedance — none of them depend on American permission. Europe's choice isn't between American AI and Chinese AI. It's between American AI that can be taken away, and Chinese AI that can't be stopped. That's not a choice. That's a trap.
The Bottom Line
The Anthropic Mythos 5 shutdown isn't a story about AI safety. It's a story about power. Who has it, who doesn't, and what happens when the people who have it decide to use it.
Europe spent years building regulatory frameworks for AI. It wrote the AI Act. It fined American tech companies. It talked about digital sovereignty. And then, in a single evening, it learned that sovereignty isn't about regulation — it's about capability. If you can't build the models, you don't control the future.
France's Le Monde called it "the AI war." But wars have battles, and battles have winners. Right now, Europe isn't even on the battlefield. It's watching from the sidelines as America and China fight for control of the most important technology of the century.
And the scariest part? The war started three days ago, and Europe already lost its first battle without firing a shot.