Should People Be Free To Deal with the Department of War?
Amy Zegart’s surprising criticism—and my response.
Introduction
In a February 27 post titled “You Should Have Moral Qualms about Anthropic’s Claims,” Hoover senior fellow and foreign policy scholar Amy Zegart challenged the ethics of a company named Anthropic. What I found refreshing is that a defense contractor’s CEO had a strong enough belief in his ethics that he was willing to forego a lucrative contract. According to Zegart, I should have moral qualms about that. I don’t and I’ll say why.
Anthropic had told the Department of War that it did not want its products used for either autonomous weapons or mass surveillance of Americans. According to Zegart, the Pentagon stated that it did not contemplate such uses. But that wasn’t enough for Dario Amodei, the CEO of Anthropic, who stated that he could not “in good conscience” accept the War Department’s assurances. Here’s Brendan Bordelon in a February 26 news item in Politico:
[Secretary of War] Hegseth met with Anthropic CEO Dario Amodei on Tuesday to deliver a warning — give the military unfettered access to its Claude AI model by Friday evening or else have the government label it a “risk” to the supply chain. The designation, typically reserved for foreign firms with ties to U.S. adversaries, could ban companies that work with the government from partnering with Anthropic.
Hegseth threatened Anthropic with designating it as a risk to the supply chain. With that label, Anthropic could be forbidden, as noted above, from working with companies that work with the government. Hegseth also, though, threatened to invoke the Defense Production Act to compel Anthropic to work with the Defense Department. A risk to the supply chain and, at the same, a firm that Hegseth wants to use? Hmmm. Bordelon quotes Dean Ball, whom he identifies as a former AI advisor in the Trump administration, noting the obvious contradiction. Said Ball, “You’re telling everyone else who supplies to the DOD you cannot use Anthropic’s models, while also saying that the DOD must use Anthropic’s models.”
Zegart cites the Politico article but doesn’t mention this contradiction. Instead, she goes after Anthropic and CEO Amodei. She writes:
There is a serious ethical question about whether one company, elected by nobody, with its own normative agenda as well as substantial global investors and customers, should be dictating the conditions of the most essential government role: protecting the lives of Americans.
But she misstates the issue. Anthropic isn’t trying to dictate the conditions of this essential government role. Anthropic is simply stating what its own limits are. If the Pentagon can find another supplier, it is free to do so and, indeed, has already done so. OpenAI has stepped up to take Anthropic’s place.
Moreover, why does Zegart think it’s important that Anthropic is elected by nobody? Does Zegart really think that companies that contemplate working with the Department of War should be elected by somebody.
Consider other people who contract with the Department of War: people who join the military. No one elects them. Imagine that someone is thinking of joining, say, the Army but is concerned about whether his services will be used in ways he disapproves. Right now, he is free not to join. That’s one of the huge benefits of having an all-volunteer military. Would Zegart be consistent and tell people who support that young person’s choice that they should have moral qualms?
Zegart writes that Anthropic “lets one billionaire have an extraordinary say in our nation’s defense that should give anyone concerned about democratic accountability pause.” But every non-joiner of the military has some say in our nation’s defense, as do joiners. What she misses is that the very freedom to choose adds an extra layer of accountability. Moreover, it can be more effective than voting. The old joke in 1965 was about the guy who said, “I was told that if I voted for Goldwater, we’d bomb North Vietnam. So I voted for Goldwater and, sure enough, we bombed North Vietnam.” What the joke illustrates is that even if the winner of an election takes one position, he is free to go against that position, as did Lyndon Johnson. Voting is not the be all and the end all. But if you fail to get enough volunteers, you can’t carry out the war. And that’s a good thing, not a bad one.
Zegart herself admits the problem with democratic accountability. She writes, “The system often works slowly and poorly—but that’s democracy in action.” That is democracy in action, often working slowly and poorly. And this is what she wants us to have confidence in?
One of the great benefits of a free society is that people get to choose which employers they work for and which firms they will contract to. Anthropic is exercising that freedom to choose, as is everyone who joins, or doesn’t join, the military. Not only should we not have moral qualms about letting people be free to choose, but also, even if people make choices that we wouldn’t, we should celebrate their freedom to make those choices. Let’s not let the defense of the United States undercut the very freedom that many of its defenders cherish.
Postscript: In a comment on antiwar.com, where this Substack post is reposted, Thomas L. Knapp informed me that I should not use the term “Department of War.” Why, I asked. He explained:
Look at the last National Defense Authorization Act. Congress dispensed money.
What organization received that money? The Department of Defense.
What organization spends that money? The Department of Defense.
With what organization does (or did) Anthropic have a contract? The Department of Defense.
The “Department of War” is an imaginary organization. It has a budget of $0.00 and not a single government official on its payroll or table of organization. It’s Pete Hegseth doing macho cos-play with Trump and some others beating their chests and cheering him on, and that is wholly and precisely what it is.


I really want to defend Amy's point of view given that she's a very attractive woman, but my superoragatory commitment to intellectual honesty compels me to full agreement with your position. Rats!
The fact that Anthropic has an ethics officer is a good thing. The fact that they did not trust Hegseth's minions and wanted concrete wording was also good thinking, the response from the Administartion was unsurprisingly petulant. That the Pentagon is severing ties with the leading AI company is more than stupid,