The Pentagon Blacklisted Claude. Then It Became the #1 App in America.

The tool I run my entire business on got blacklisted by the U.S. government last week.

I want to tell you what happened, because I don't think people fully understand how big this is — not just for AI, but for every builder and business that depends on these tools.

The timeline

Friday, February 27th. Pentagon gives Anthropic a deadline: remove every usage restriction on Claude. All of them. Mass surveillance, autonomous weapons, the works. Anthropic's answer — no. They drew two hard lines: no mass surveillance on American citizens, and no fully autonomous weapons without a human in the decision chain.

That's it. Those were the conditions.

The government's response: label Claude a "supply chain risk" and order every federal agency to stop using it. Defense tech companies started dropping Claude within hours.

Then things got weird

ChatGPT uninstalls shot up 295% overnight. Not a typo — two hundred and ninety-five percent. Claude downloads jumped 51%. By Monday morning, Claude was sitting at #1 on the App Store. Above Instagram. Above TikTok. Above everything.

So many people flooded in that Anthropic's servers went down for 10 hours straight.

I felt that one personally. My content pipeline, my automation stack, my whole daily workflow runs on Claude. I've been building on it for six months — 18 slash commands, 12 API integrations, a visual dashboard, all of it wired through Claude Code. When it crashed I was just staring at my screen like... okay, I guess today's a day off.

The part nobody's talking about

Everyone's debating whether Anthropic made the right call. That's the wrong question.

The real story is that this was the first time millions of regular people actually got a vote on who controls AI. Not at the ballot box. At the app store. And they voted overwhelmingly for the company that said "there are lines we won't cross."

Same week, 900 employees across Anthropic, OpenAI, and Google signed an open letter backing the decision. Meanwhile OpenAI signed its own deal with the Department of Defense. Same question posed to two companies. Two completely different answers.

Why this matters if you're building on AI

I spent 20 years as a network architect. When someone labels a vendor a "supply chain risk" — that's not a slap on the wrist. I've seen that designation kill vendor relationships across entire enterprise ecosystems. It's the infrastructure equivalent of a kill switch.

Anthropic knew exactly what it would cost them. Government contracts, defense partnerships, maybe regulatory goodwill down the road. They walked into it anyway.

If you're building anything on AI tools right now — and honestly at this point who isn't — here's what I'd think about:

Your tools aren't neutral infrastructure. They carry the values of whoever built them. The APIs you wire into your products, the models you build your workflows around, the platforms you trust with your business logic — those choices say something about what you're building and who you're building it for.

That used to be abstract. It's playing out in real time now.

Pick your tools like you'd pick a business partner. Because that's what they are.