The past two weeks have been marked by a clash between Anthropic CEO Dario Amodei and Secretary of Defense Pete Hegseth, as they battle over the military’s use of AI.
Anthropic refuses to allow its AI models to be used for mass surveillance of the American public or for fully autonomous weapons that carry out attacks without human intervention. At the same time, Hegseth argued that the Pentagon should not be limited by vendor rules and that any “lawful use” of the technology should be allowed.
On Thursday, Amodei publicly signaled that Anthropic would not back down despite threats that his company could be designated as a supply chain risk as a result. But as the news cycle moves forward, it’s worth revisiting exactly what’s at stake in this battle.
At the heart of this battle is who controls powerful AI systems: the companies that build them or the governments that want to deploy them.
What are Antropics worried about?
As mentioned above, Anthropic does not want its AI models to be used for mass surveillance of the American population or for autonomous weapons that do not involve humans in determining targets or firing shots. While traditional defense contractors typically have little say in how their products are used, Anthropic has argued since its inception that AI technology has unique risks and therefore requires its own safeguards. From the company’s perspective, the question is how to maintain these safeguards when the technology is used in the military.
The U.S. military already relies on highly automated systems, some of which are deadly. While the decision to use lethal force has historically been left to humans, there are few legal restrictions on the military use of autonomous weapons. The Pentagon does not explicitly ban fully autonomous weapons systems. According to a 2023 Pentagon directive, AI systems can select and attack targets without human intervention as long as they meet certain standards and pass review by senior defense officials.
That’s exactly what makes Anthropic so uneasy. Military technology is inherently secret, so even if the U.S. military takes steps to automate lethal decision-making, we may not know about it until it is operational. And if you use Anthropic’s models, that could count as “lawful use.”
tech crunch event
boston, massachusetts
|
June 9, 2026
Anthropic’s position is not that such uses should be permanently taken off the table. It’s just that the model isn’t yet powerful enough to safely support them. Imagine an autonomous system misidentifying a target, escalating a conflict without human permission, or making deadly decisions in the blink of an eye that no one can reverse. When you put a weak AI in charge of your weapons, you end up with a very fast and confident machine that isn’t good at making high-stakes decisions.
AI also has the power to increase legitimate surveillance of American citizens to an alarming degree. Under current U.S. law, surveillance of U.S. citizens is already possible through the collection of text messages, emails, and other communications. AI changes the equation by enabling automated large-scale pattern detection, entity resolution across datasets, predictive risk scoring, and continuous behavioral analysis.
What does the Pentagon want?
The Pentagon’s argument is that it should be able to deploy Anthropic’s technology for any lawful use it deems necessary, rather than being limited by Anthropic’s internal policies regarding things like autonomous weapons and surveillance.
More specifically, Secretary Hegseth argued that the Department of Defense should not be limited by vendor regulations and would be committed to the “lawful use” of technology.
Sean Parnell, the Pentagon’s chief spokesman, said in a post Thursday in the X that the Pentagon is not interested in conducting large-scale domestic surveillance or deploying autonomous weapons.
“This is what we’re asking: Allow the Department of Defense to use Anthropic models for any lawful purpose,” Parnell said. “This is a simple, common sense request to prevent Anthropic from potentially jeopardizing critical military operations and endangering our nation’s warfighters. We will not allow any company to dictate the terms of how operational decisions are made.”
He added that Anthropic must make a decision by 5:01 p.m. ET on Friday. “Otherwise, we will terminate our partnership with Anthropic and view it as a supply chain risk to Dow,” he said.
Despite the Pentagon’s stance that it should not be limited by corporate use policies, Secretary Hegseth’s concerns about Anthropic sometimes appeared to be related to cultural grievances. Hegseth lambasted “woke AI” in a speech he gave at the SpaceX and xAI offices in January that some saw as a prelude to a feud with Anthropic.
“The War Department’s AI will not wake up,” Hegseth said. “We are building weapons and systems for war, not chatbots for Ivy League faculty lounges.”
So what now?
The Pentagon is threatening to either declare Anthropic a “supply chain risk” (effectively blacklisting Anthropic from doing business with the government) or invoke the Defense Production Act (DPA) to force the company to adjust its models to the military’s needs. Hegseth gave Anthropic until 5:01 p.m. Friday to respond. But with the deadline looming, no one knows whether the Pentagon will follow through on its threat.
This is not a fight that either side can easily walk away from. Sachin Seth, a venture capitalist at Trousdale Ventures who focuses on defense technology, said the supply chain risk label for Anthropic could mean “lights out” for the company.
But if Anthropic were removed from the Department of Defense, it could become a national security issue, he said.
“[The department]will have to wait six to 12 months for either OpenAI or xAI to catch up,” Seth told TechCrunch. “So you could have up to a year and they might end up using the second or third best model instead of the best model.”
xAI is preparing to declassify and replace Anthropic, and given owner Elon Musk’s rhetoric on the subject, the company would have no problem giving the Department of Defense complete control over its technology. According to recent reports, OpenAI may stick to the same red lines as Anthropic.
