Pentagon Flags Anthropic as Supply Chain Security Threat

The Department of Defense (DoD) has formally informed Anthropic's leadership that the company and its products have been classified as a supply chain risk, according to a Bloomberg report citing a senior department official.

This designation follows weeks of tension between the AI research lab and the Pentagon. Anthropic CEO Dario Amodei has declined to permit military use of its AI systems for mass surveillance of American citizens or for operating fully autonomous weapons where humans are not involved in targeting or engagement decisions. The DoD has countered that its utilization of AI should not be restricted by a private contractor.

Supply chain risk designations are usually applied to foreign adversarial entities. This label mandates that any company or agency working with the Pentagon must certify that it does not utilize Anthropic's AI models.

The Pentagon's decision risks significant disruption for both Anthropic and its own operations. Anthropic has been the sole leading AI lab with systems certified for handling classified information. The U.S. military is currently employing Claude, Anthropic's AI, in its Iran-related campaigns, where American forces use AI tools to rapidly process operational data. According to Bloomberg, Claude is a primary component of Palantir's Maven Smart System, which is relied upon by military operators in the Middle East.

Several critics have described labeling Anthropic a supply chain risk over this policy disagreement as an unprecedented move by the Department. Dean Ball, a former AI advisor in the Trump White House, has called the designation a "death rattle" of the American republic, arguing the government has forsaken strategic clarity and respect for a "thuggish" tribalism that treats domestic innovators more harshly than foreign adversaries.

Hundreds of employees from OpenAI and Google have urged the DoD to rescind its designation and called on Congress to challenge what they see as an improper exertion of authority against a U.S. tech firm. They have also encouraged their own leaders to maintain a united front in refusing the Pentagon's demands to use their AI models for domestic mass surveillance and for "autonomously killing people without human oversight."

TechCrunch has contacted Anthropic for a statement.

Amid this conflict, OpenAI independently reached an agreement with the Department, allowing military use of its AI systems for "all lawful purposes." Some OpenAI staff have raised concerns about the vague wording of this deal, which they fear could enable precisely the kinds of applications Anthropic sought to prevent.

Amodei has characterized the DoD's actions as "retaliatory and punitive," and has reportedly suggested that his refusal to endorse or donate to President Trump influenced the dispute with the Pentagon. In contrast, OpenAI president Greg Brockman has been a strong supporter of Trump, recently contributing $25 million to the MAGA Inc. Super PAC.