“If you strike me down, I shall become more powerful than you can possibly imagine.”
Obi-Wan said it to Vader before lowering his lightsaber.
This week, Anthropic effectively said it to the United States government.
The Pentagon pushed, reaching for their weapons. The administration escalated. Anthropic refused to remove guardrails around mass domestic surveillance and fully autonomous weapons. The government labeled the company a “supply chain risk” and cut it off from federal business like a hot lightsaber amputating Luke’s hand.
Within hours, Claude was the #2 free app in the Apple App Store.
That’s not coincidence. That’s consequence. That, my friends, is the Force and may it be with Anthropic.
Let’s slow this down.
The Department of Defense wanted broader rights to use Claude. For “All lawful purposes” sounds harmless until you unpack it. Who decides what’s unlawful and how long does it take to get an answer while it is used unlawfully is the question. In this case, it meant removing explicit contractual limits around using AI for surveillance of Americans and fully autonomous weapons systems.
Dario Amodei said no.
Not “let’s revisit this.” Not “let’s fine-tune the language.” No. Anthropic held the line.
The response was swift. The administration publicly branded the company a national security supply-chain risk. That’s language usually reserved for hostile actors, not American AI startups founded by former OpenAI researchers and backed by U.S. capital.
That escalation matters. But what matters more is what happened next. Claude surged.
From outside the top 100 weeks ago to #2 in the country. Sitting between ChatGPT and Gemini. Consumers downloaded it in droves. Developers took a look. Cultural signals amplified it. Narrative became momentum. Momentum became demand.
That’s how markets talk back. The algorithms have spoken.
Before this becomes partisan theater, let’s be adults about it. Governments seek flexibility. National security planners don’t like constraints. They want maximum optionality.
Companies sometimes draw boundaries. The friction was inevitable. What wasn’t inevitable was the public reaction.
When a company says, “We will not enable mass surveillance of our own citizens,” and the consequence is being labeled a national security risk, people notice. That dynamic is bigger than this contract dispute.
We’ve seen this before. Apple versus the FBI.
The government demanded access to encrypted iPhones. Apple refused. They were accused of putting privacy over public safety. Called irresponsible. Reckless.
And what happened? Privacy became brand equity. Consumers rewarded Apple for holding the line. Not because they understood the cryptography. Because they understood the principle.
When a company articulates a boundary clearly enough, people align with it. Anthropic may have just stepped into that same pattern.
We keep saying AI is the defining technology of the century. We’re in a global race. China is not slowing down. Europe is regulating aggressively. The U.S. wants dominance, leadership, strategic control. So here’s the uncomfortable question.
Why would we publicly kneecap one of our own premier AI companies over ethical guardrails that are, at minimum, defensible? You can argue Anthropic is wrong. You can argue the Pentagon is right. That debate is legitimate.
But when the signal to the ecosystem becomes: remove your safety limits or risk being labeled disloyal, founders learn something. Investors learn something. Developers learn something. Incentives shape behavior.
If ethical constraints become a liability, fewer companies will bother having them. That’s not ideology. That’s human nature.
There’s another contrast here.
Sam Altman announced OpenAI reached an agreement with the Defense Department. Amodei did not. This isn’t a morality play. This is how the industry works. Companies navigate. They calculate. They make different risk decisions.
But the market made a decision too. And right now, the market appears to be rewarding the company that said no. That’s not activism. That’s behavior.
Here’s Shimmy’s take.
Governments have enormous power. They can cancel contracts. They can issue designations. They can apply pressure. What they cannot fully control is conviction at scale. As I wrote last week, the algorithms rule. People can vote with their fingers and their wallets.
I downloaded Claude. I signed up for a yearly subscription.
Not as protest theater. Not as tribal politics. As alignment.
I believe companies that articulate clear ethical boundaries deserve to compete in the open market — not be publicly branded into submission.
You can strike a company’s federal revenue. You can’t strike consumer belief so easily.
Old Ben Kenobi understood something about power. When you stand for something clearly enough, even getting struck down can amplify you.
Power can cancel a contract. It can’t cancel conviction.
May the Force be with us all.
