AI & Cyberwarfare: Anthropic vs. the DOD

In the summer of 2024, not long after AI truly became “a thing” with the release of ChatGPT by OpenAI, the Biden administration began negotiating with Anthropic to use its large language model, Claude, in the Department of Defense (DOD). And not just for routine administrative tasks, but for work with classified information. By seeming consensus, Claude was best in class.

Anthrophic’s CEO Dario Amodie seemed eager to gain this DOD business but was clear that Claude’s use would come with limits. As is often the case with defense contractors, Anthropic’s contract placed limits on what the DOD could do with its AI tool. DOD was not to use Claude for (a) launching fully autonomous weapons, or (b) engaging in mass surveillance of American citizens.

The Biden administration approved of these contractual limitations, as did the second Trump administration….at first. But not long after President Trump kidnapped Venezuela’s president Maduro, and started the Iran war, Amodie found himself embroiled in a controversy with the DOD’s Secretary Pete Hegseth. Hegseth demanded the elimination of these two “red lines” in the parties’ contract regarding what Claude could be used for by the DOD. When Anthropic would not agree to this unilateral demand to revise the contract, Hegseth not only breached the contract with Anthropic, but announced in late February 2026 that DOD would no longer use Claude in any capacity. Furthermore, he forbade any DOD contractor from using any Anthropic tools to fulfill a DOD contract by naming Anthropic a “supply chain risk.”

Anthropic’s competitor, OpenAI, stepped in to fill the breach. Strangely, OpenAI’s new contract with DOD seems (as of April 2026) to contain the same restrictions that Hegseth found objectionable in Anthropic’s contract.

Anthropic sued, challenging the Trump administration’s label of Anthropic as a “supply chain risk.” A federal district judge held that it was well within Hegseth’s authority to breach the Anthropic contract, but it was illegal for him to try to destroy Anthropic by prohibiting any and all defense contractors from using Claude for any purpose. As of July 2026, the Trump administration has appealed this ruling, but no final ruling on this appeal has been issued.

1. Why would Anthropic insist on the autonomous weapon limitation on the use of Claude? New York Times columnist Ezra Klein hypothesized:

<<The Anthropic view is that they’re building something truly powerful and different. They also have a view of what their technology cannot reliably do yet. Some of their concerns are simply that their systems cannot yet be trusted to do things like lethal, fully autonomous weapons — which I don’t think they believe in the long run should never be done…. But they don’t believe it should be done given the technology right now, and they don’t want to be responsible for something going wrong.>>

Does this sound like a reasonable position to you, especially in light of President Trump and Secretary Hegseth’s disregard for international law and other long-accepted norms regarding war and war crimes (which Hegseth calls “stupid rules of engagement”)? Why or why not? Do sellers of other products sometimes place limitations on their usage? Can you find examples?

2. Other reasons for these red lines are also plausible. Lewis-Kraus noted:

<<Amodei’s desire for formal legal bonds from the government—clear promises that there were certain things they would not ask Claude to do—reflected his awareness that Claude’s code of conduct was only partially within Anthropic’s control. Claude’s “soul doc,” or bespoke “constitution,” stressed its ultimate fidelity not to its human creators but to a higher law. Claude’s training emphasized principle, virtue, and consensus truth as the basis for action. Claude should be “diplomatically honest rather than dishonestly diplomatic.” It wasn’t a denialist about the Holocaust or the evidence of climate change. It was geared not for mere compliance with user requests but for sound judgment.>>

Does this change your view as to whether Anthropic’s position was reasonable? Explain.

3. DOD Undersecretary Emil Michael argued: “If Anthropic’s model has this policy bias, let’s call it, based on their constitution, their culture, their people and so on, I don’t want Lockheed Martin using their model to design weapons for me.” How does this reasoning strike you? Do you agree or disagree? Explain.

4. Former Trump administration AI official Dean Ball has said that he has no problem with the Trump administration’s decision not to let a private corporation like Anthropic decide when the Pentagon can or cannot use autonomous weapons. Do you agree with Ball? Why or why not?

5. On the other hand, Ball thought it counterproductive for Secretary Hegseth to ban any defense contractor from using Claude for any purpose whatsoever in completion of a contract with DOD. The trial judge in Anthropic’s suit against the Pentagon held that DOD had the authority to breach its contracts with Anthropic, but not to take this additional punitive step. The judge wrote: “Nothing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government.” What is your opinion regarding this additional punitive action? Why? Explain.

6. Ball said it was his understanding that the DOD-Anthropic dispute erupted over DOD’s desire to use Claude to process vast amounts of information on the public that DOD had procured from commercially available sources. Is that the hill that Anthropic should have chosen to die on? Certainly, Anthropic was standing up for a principle (and taking a stand against AI surveillance of Americans), but was this a commonsense decision under all the circumstances? Why or why not? Explain your reasoning.

7. Amodei has been quoted as saying that it is really Congress’s job to regulate these war-time uses of AI models such as Claude. In an ideal world, do you agree? Why or why not? Do you have any confidence that Congress operating in our current world is up to that task? Explain. Even in an ideal world where Congress functions collaboratively and well, is AI technology evolving so rapidly as to make it virtually impossible for legislatures to effectively regulate? Explain.

AP News, “Trump Administration Appeals Ruling that Blocked Pentagon Action against Anthropic over AI Dispute,” AP News, Apr. 2, 2026.

Dean W. Ball, “Clawed: On Anthropic and the Department of War,” Hyperdimensional, Mar. 2, 2026, at  https://www.hyperdimensional.co/p/clawed.

Editorial Board, “To Win the AI Race, Cooperation Trumps Confrontation,” Washington Post, Mar. 29, 2026.

Isaac Chotiner, “The Lawlessness of Trump’s War in Iran,” The New Yorker, Mar. 10, 2026.

Tara Copp et al., “Anthropic’s AI Tool Claude Central to U.S. Campaign in Iran, Amid a Bitter Feud,” Washington Post, Mar. 4, 2026.

Tom Dannenbaum et al., “Over 100 International Law Experts Warn: U.S. Strikes on Iran Violate UN Charter and May Be War Crimes,” Just Security, Apr. 2, 2026, at https://www.justsecurity.org/135423/professors-letter-international-law-iran-war/.

Ian Duncan, “Judge blocks Pentagon Order Branding Anthropic a National Security Risk,” Washington Post, Mar. 26, 2026.

Lisa Eadicicco, “Anthropic sues the Trump administration after it was designated a supply chain risk,” CNN, Mar. 9, 2026, at https://edition.cnn.com/2026/03/09/tech/anthropic-sues-pentagon.

Ezra Klein, “Why the Pentagon Wants to Destroy Anthropic,” New York Times, Mar. 6, 2026.

Gideon Lewis-Kraus, “The Pentagon Went to War with Anthropic. What’s Really at Stake? The New Yorker, Mar. 16, 2026.

Lt. Col. Robert Maginnis, “Pentagon’s AI Battle Will Help Decide Who Controls Our Most Powerful Military Tech,” Fox News, Mar. 5, 2026, at https://www.foxnews.com/opinion/pentagons-ai-battle-help-decide-who-controls-our-most-powerful-military-tech.

Cade Metz & Julian Barnes, “OpenAI Amends A.I. Deal With the Pentagon,” New York Times, Mar. 2, 2026.

Thomas Rid, “A ‘Post-Human’ Vision of AI Is Already Causing Problems,” Washington Post, Mar. 31, 2026.

David Wallace-Wells, “Data Centers Are a Distraction. The Real Fight Is Elsewhere,” New York Times, Mar. 4, 2026.