Pentagon Red-Flags Anthropic as Supply Chain Risk

Defense Secretary Pete Hegseth's previous threats to sanction an artificial intelligence firm for its unyielding acceptable use policy have now materialized into official action.

Following weeks of unsuccessful negotiations, public ultimatums, and veiled legal threats, the Department of Defense (DoD) has officially designated Anthropic, a prominent AI company, as a "supply-chain risk." This move significantly escalates the ongoing dispute regarding Anthropic's acceptable use policies and could potentially lead to litigation.

As initially reported by The Wall Street Journal on Thursday, citing a knowledgeable source, this ruling will prohibit defense contractors from engaging with the government if they integrate Claude, Anthropic's AI program, into their offerings. Notably, while such designations are typically reserved for foreign entities linked to adversarial governments, this marks the first instance an American company has publicly received this classification.

Central to this dispute is Anthropic's steadfast refusal to permit the Pentagon to utilize its Claude AI for two specific applications: autonomous lethal weapons operating without human oversight, and widespread mass surveillance. The Pentagon contended that Anthropic's insistence on controlling government usage would grant undue influence to a private entity, while Anthropic expressed skepticism that the government would honor its established ethical boundaries. Negotiations deteriorated, with the Pentagon progressively threatening to invoke the "supply-chain risk" designation if Anthropic did not accede to its demands. After Anthropic publicly declared its non-compliance last Thursday, the Pentagon followed through on its threat. (The Pentagon declined to comment on the record, and Anthropic did not immediately respond to a request for comment.)

The full scope of the Pentagon's enforcement of this designation remains uncertain. On Friday, during his announcement of the intent to label Anthropic a risk, Defense Secretary Pete Hegseth asserted that any company engaging in "any commercial activity" with Anthropic – even beyond their direct work for the Pentagon – would face cancellation of their defense contracts. Anthropic promptly countered, stating that such a sweeping interpretation of the law would be unlawful. Secretary Hegseth, alongside President Donald Trump, has imposed a six-month deadline for Anthropic to remove Claude from all government systems. This directive, however, presents significant challenges, particularly for military operations. Reports following the weekend's U.S. attack on Iran, which resulted in the targeted missile strike killing Supreme Leader Ayatollah Ali Khamenei, indicate that Claude-powered intelligence tools were instrumental in the mission's success.