So here's a fun situation: an AI company gets blacklisted by the Pentagon for being too ethical. Or maybe for not being ethical enough? It depends on who you ask.
Anthropic CEO Dario Amodei spent part of his Sunday defending his company's patriotic credentials after President Donald Trump blacklisted the startup's Claude AI for government agencies. In an interview, Amodei made the case that refusing the Pentagon's demands for unrestricted AI use doesn't make you unpatriotic—it might make you more American.
"I would say, we are patriotic Americans," Amodei said when asked what he'd tell Trump now. "Everything we have done has been for the sake of this country, for the sake of supporting U.S. national security. Our leaning forward in deploying our models with the military was done because we believe in this country."
That's a pretty direct rebuttal to the idea that refusing military work makes you anti-American. Amodei pointed out that Anthropic was actually the first AI company to assist the defense community in a classified capacity. He also emphasized the company's commitment to defending the U.S. against autocratic adversaries like China and Russia.
The Pentagon's Supply Chain Problem
So what exactly happened? The Pentagon labeled Anthropic a "supply-chain risk," which is bureaucratic speak for "we don't trust this vendor enough to let our contractors use their stuff." The label bars other contractors from using Anthropic's AI for military purposes.
The disagreement came down to control. The Pentagon wanted unrestricted use of Anthropic's AI. The company said no, citing concerns about domestic mass surveillance and autonomous weapons. Think about it: the military wants AI that can do whatever they need it to do, and a company founded by former OpenAI safety researchers is saying "but what about the ethical boundaries?"
Amodei acknowledged that while the company agrees with most military use cases, it maintains "red lines" on certain applications. He stressed the need for Congress to establish AI regulations, noting that technology is advancing faster than the law can keep up.
Collaboration With Conditions
Despite the blacklist, Amodei says he's still open to working with the government. He downplayed the blacklist's impact on non-defense operations, asserting that the company will continue to thrive. But he's not backing down on the ethical concerns.
The refusal to comply with the Pentagon's demand for unrestricted AI use led to the Department of War labeling Claude a "supply chain risk" and threatening to use the Defense Production Act to enforce changes. That's the law that lets the government compel companies to prioritize national defense needs—it's not an empty threat.
Meanwhile, Amodei has broader concerns about the AI industry. He's worried about the rapid concentration of AI power and wealth among a few companies, warning that this shift could lead to significant economic and political influence. It's a bit ironic: a company getting blacklisted by the government is also warning about too much power accumulating in private hands.
And here's the kicker: U.S. Central Command reportedly used Claude during the Trump administration's major air operation against Iran, just hours after the president ordered federal agencies to stop using the company's technology. So either someone didn't get the memo, or the military found Claude too useful to immediately abandon.
What you have here is a classic tech-government tension: how much control should companies maintain over their technology when it's used for national security? Anthropic is drawing lines in the sand about what its AI can and can't be used for, while the Pentagon wants tools without restrictions. Both sides claim to be acting in America's best interests—they just disagree on what that means.
