Pentagon Labels Anthropic a Supply Chain Risk

The dispute traces back to contract renegotiations that began in late 2025, when the Pentagon sought broader terms for its use of AI models from all major providers. Anthropic, which signed a $200 million contract with the DOD in July and was the first AI lab to deploy models on classified networks, pushed back on granting unlimited access.

Anthropic wanted narrow exceptions preventing the use of Claude for fully autonomous lethal weapons and mass surveillance of American citizens. The Pentagon viewed these restrictions as a vendor attempting to "insert itself into the chain of command," according to a senior defense official. Talks continued through February, but broke down definitively on the eve of the [Iran conflict](/posts/2026-03-01/developing-khamenei-confirmed-dead-as-iran-retaliates-across-gulf-states-and-us-israel-launch-fresh-strikes-on-tehran).

On February 28, President Donald Trump posted on Truth Social directing all federal agencies to "immediately cease" using Anthropic's technology, writing "We don't need it, we don't want it, and will not do business with them again." Defense Secretary Hegseth followed with a post on X announcing the supply chain risk designation, calling Anthropic's stance "a master class in arrogance and betrayal." Anthropic said it received no advance warning that these statements were coming.

The supply chain risk label, codified in federal law, was designed to protect the U.S. military from infiltration by foreign adversaries such as companies beholden to Beijing or Moscow. Anthropic is the first American company to receive the designation. It requires defense vendors and contractors to certify they do not use Anthropic's models in Pentagon-related work.

However, the scope may be narrower than initial reports suggested. Amodei noted that the designation "doesn't (and can't) limit uses of Claude or business relationships with Anthropic if those are unrelated to their specific Department of War contracts." [Microsoft](/posts/2026-02-18/msft-analysis-the-most-under-owned-megacap-why-microsofts-28-pullback-from-highs-may-be-the-ai-eras-best-risk-reward-setup), which announced plans to invest up to $5 billion in Anthropic in November, said its lawyers concluded that Anthropic products can remain available to customers outside the DOD.

[Lockheed Martin](/posts/2026-02-21/lmt-analysis-lockheed-martin-touches-a-52-week-high-as-global-rearmament-reshapes-the-defense-sector-is-the-rally-priced-in) said it will "follow the President's and the Department of War's direction" and look to other AI providers, adding that it expects "minimal impacts" since it is not dependent on any single large language model vendor. Defense data analytics firm Palantir, which derives about 60 percent of its U.S. revenue from government contracts and has an existing integration with Claude, saw its shares dip on the news before recovering. Analysts at Piper Sandler warned that moving off Anthropic's technology could "pose some short-term disruptions" to Palantir's operations.

Hours after Anthropic was [blacklisted on February 28](/posts/2026-03-01/trump-bans-anthropic-as-openai-wins-pentagon-deal), OpenAI CEO Sam Altman announced a new agreement to deploy ChatGPT in classified military environments, effectively positioning OpenAI as Anthropic's replacement. Altman said the Pentagon displayed a "deep respect for safety and a desire to partner to achieve the best possible outcome."

However, OpenAI's rapid deal drew scrutiny. Altman later acknowledged the announcement "looked opportunistic and sloppy" and said he shouldn't have rushed it. OpenAI subsequently amended its agreement, claiming it now includes "more guardrails than any previous agreement for classified AI deployments, including Anthropic's."

Elon Musk's xAI also struck a deal with the Pentagon last week to deploy its Grok AI systems on classified networks. The swift entry of multiple competitors into the vacuum left by Anthropic underscores the high stakes of military AI contracts, which are expected to grow substantially as the DOD accelerates its integration of artificial intelligence across operations.

The designation has drawn broad opposition from across the political spectrum. Senator Kirsten Gillibrand, a Democrat on the Senate Armed Services and Intelligence Committees, called it "a dangerous misuse of a tool meant to address adversary-controlled technology" and "a gift to our adversaries."

A group of former defense and national security officials, including former CIA director Michael Hayden and retired military leaders, sent a letter to lawmakers expressing "serious concern" about the designation. They wrote that applying the supply chain risk label to a domestic company for declining to remove safety guardrails "is a category error with consequences that extend far beyond this dispute."

Neil Chilson, a Republican former chief technologist at the Federal Trade Commission, said the decision "looks like massive overreach that would hurt both the U.S. AI sector and the military's ability to acquire the best technology." Several legal experts have said the designation is unlikely to hold up in court, noting it was designed for foreign adversaries and has never been tested against a domestic company operating transparently under U.S. law.

Some observers have pointed to an apparent inconsistency: Chinese AI company DeepSeek, which has faced accusations of unfair practices, has not received a similar designation despite posing arguably greater supply chain concerns.

While Anthropic faces significant losses in government contracts, the public dispute has produced an unexpected windfall in consumer adoption. The company said more than a million people signed up for Claude each day during the past week, propelling it past OpenAI's ChatGPT and Google's Gemini as the top AI app in more than 20 countries on Apple's App Store.

The broader AI industry is watching the standoff closely. An influential tech advocacy group whose members include Nvidia and Apple sent a letter to Hegseth urging him not to apply the supply chain risk label, warning it would discourage private companies from working with the federal government. Tim Fist, director of emerging technology at the Institute for Progress, said the designation "hurts the AI industry and thus US national security for essentially no gain."

Michael Sobolik, an AI and China expert at the Hudson Institute, offered a blunt assessment: "We're treating an American AI company worse than we're treating a Chinese Communist Party-controlled AI company. The U.S. government risks cutting off the legs of one of our best AI companies in the early years of this AI race."

The legal challenge Anthropic has promised will test uncharted territory in defense procurement law and could establish precedent for how the government can compel private technology companies to provide unrestricted access to their products.