The last fourteen days have centered on a confrontation between Anthropic’s CEO, Dario Amodei, and Defense Secretary Pete Hegseth, as they debate the military’s application of artificial intelligence. Anthropic is holding firm against permitting its AI models to be utilized for mass surveillance of U.S. citizens or for fully autonomous weapon systems that execute strikes without human oversight. Concurrently, Secretary Hegseth has contended that the Department of Defense must not be constrained by a supplier’s regulations, maintaining that any application of the technology deemed "lawful" should be allowed.

This past Thursday, Amodei made it publicly clear that Anthropic has no intention of yielding, even in the face of warnings that his firm could be flagged as a supply chain risk. Given the rapid pace of current events, it’s important to re-examine the fundamental stakes of this dispute. Ultimately, this conflict is about authority: who gets to control powerful AI systems—the corporations that create them, or the government agencies that wish to employ them.

**What are Anthropic's primary concerns?** As noted, Anthropic aims to prevent its AI from being used in mass domestic surveillance or in autonomous weapons where humans are removed from targeting and engagement decisions. While traditional defense contractors often have limited influence over the end-use of their products, Anthropic has consistently argued that AI presents distinct dangers that demand special protective measures.

From the company’s viewpoint, the central challenge is upholding those safeguards when the technology is in military hands. The U.S. military already depends on highly automated systems, some of which are lethal. Although the decision to apply lethal force has traditionally rested with people, there are scant legal barriers to the military’s use of autonomous weapons. The Department of Defense does not impose an outright ban on fully autonomous weapon systems.

A 2023 DOD directive states that AI systems may identify and engage targets without human involvement, provided they satisfy specific standards and receive approval from senior defense officials. This policy is exactly what raises alarms for Anthropic. Military technology is inherently secretive; if the U.S. military were advancing toward automated lethal decision-making, the public might remain unaware until such systems were already active. And if those systems incorporated Anthropic’s models, the military could classify it as a "lawful use."

Anthropic’s stance is not that such applications should be prohibited forever, but rather that its current models are not sufficiently advanced to support them safely. Consider the risks: an autonomous system could misidentify a target, escalate a conflict without human approval, or make an irreversible, split-second lethal choice. Entrusting weapons to a less-capable AI creates a machine that operates with extreme speed and confidence but is poorly suited for high-stakes judgments.

Furthermore, AI has the potential to dramatically intensify lawful surveillance of American citizens to a troubling extent. Existing U.S. laws already permit surveillance through the collection of texts, emails, and other communications. AI shifts the balance by enabling automated, large-scale pattern recognition, cross-dataset entity resolution, predictive risk scoring, and continuous behavioral analysis.

**What is the Pentagon seeking?** The Pentagon’s position is that it should have the freedom to deploy Anthropic’s technology for any lawful purpose it considers necessary, rather than being restricted by the company’s internal policies on autonomous weapons or surveillance. Secretary Hegseth has specifically argued that the Department of Defense should not be bound by a vendor’s rules and that it would employ the technology for "lawful use."

In a post on X this past Thursday, the Pentagon’s chief spokesperson, Sean Parnell, stated that the department has no intention of conducting mass domestic surveillance or fielding autonomous weapons. "Here’s what we’re asking: Allow the Pentagon to use Anthropic’s model for all lawful purposes," Parnell said. "This is a simple, common-sense request that will prevent Anthropic from jeopardizing critical military operations and potentially putting our warfighters at risk. We will not let ANY company dictate the terms regarding how we make operational decisions."

He added that Anthropic has until 5:01 PM ET on Friday to decide. "Otherwise, we will terminate our partnership with Anthropic and deem them a supply chain risk for DOW," he said.

Despite the Department’s official stance that it simply should not be limited by a corporation’s usage policies, Secretary Hegseth’s criticisms of Anthropic have occasionally appeared linked to broader cultural grievances. During a speech at SpaceX and xAI offices in January, Hegseth denounced what he called "woke AI" in remarks that some interpreted as foreshadowing his clash with Anthropic. "Department of War AI will not be woke," Hegseth declared. "We’re building war-ready weapons and systems, not chatbots for an Ivy League faculty lounge."

**What happens next?** The Pentagon has threatened to either designate Anthropic a "supply chain risk"—which would essentially blacklist the company from government contracts—or invoke the Defense Production Act (DPA) to compel the company to adapt its model to military specifications. Hegseth has given Anthropic until 5:01 PM on Friday to respond.

As the deadline nears, it remains uncertain whether the Pentagon will follow through on its threat. This is a confrontation neither side can easily abandon. Sachin Seth, a venture capitalist at Trousdale Ventures who specializes in defense technology, suggests that a supply chain risk designation could mean "lights out" for Anthropic. However, he also noted that if Anthropic is cut off from the DOD, it could create a national security concern. "That leaves a window of up to a year where they might be working from not the best model, but the second- or third-best."

Meanwhile, xAI is preparing to achieve classified-ready status and potentially replace Anthropic. Given owner Elon Musk’s public statements on the issue, it is reasonable to assume that company would have no objection to granting the DOD complete control over its technology. Recent reports also suggest that OpenAI may uphold the same boundaries as Anthropic.