Pentagon Maintains Anthropic Ban Despite Lutnick’s Claim of a Washington Reset

The Pentagon says its restrictions on Anthropic remain in force, creating a direct clash with Commerce Secretary Howard Lutnick’s statement that the company is “back on the right side” of the Trump administration. Lutnick also said, “We trust Anthropic” and that the company had done what officials asked, according to Axios-based accounts published by Reuters and Livemint.

The conflicting messages leave Anthropic’s government status unresolved. Lutnick’s remarks signal a political thaw, while the Defense Department’s position keeps the practical barrier to defense work in place. Bloomberg described the dispute as a disagreement between the Commerce secretary and a senior Pentagon official, with no indication that the defense designation had been formally withdrawn.

The conflict centers on Anthropic’s limits for military use of Claude, including restrictions involving domestic surveillance and autonomous weapons. Defense Secretary Pete Hegseth designated Anthropic a national-security supply-chain risk on Feb. 27 after President Donald Trump directed federal agencies to stop using the company’s technology, according to Lawfaremedia and the New York Post.

Anthropic’s position put safeguards at the center of a procurement fight. The company had built defense relationships through Amazon and Palantir, and Pentagon technology officials had considered it an important AI supplier. CNBC reported that the blacklist unsettled experts because the department had previously made significant use of Anthropic’s systems and partnerships.

The legal picture is just as divided as the political messaging. A federal judge struck down the Pentagon’s designation, ruling that the action amounted to retaliation for protected speech, according to Roic and Lawfaremedia. That ruling challenged the government’s authority to punish a contractor for refusing particular military applications of its models.

A separate account from Bankinfosecurity said an appeals court allowed the Pentagon to enforce the supply-chain-risk designation while related challenges continue. That decision keeps Claude models out of defense systems during the litigation. The competing rulings explain why a favorable trial-court decision did not automatically restore Anthropic’s access to military contracts.

The dispute reaches beyond one vendor because it tests who sets operating limits for AI in national security. Anthropic has argued for constraints on uses that involve surveillance or autonomous weapons, while defense officials have sought broader access to commercial AI systems. Congress’s research service has framed the confrontation around autonomous weapon systems and the issues lawmakers face as the government incorporates AI into military operations.

For the Pentagon, maintaining a supplier ban supports its demand for control over defense applications and reduces dependence on a company that sets its own restrictions. For Anthropic, the case concerns whether safety commitments can survive government pressure. The outcome will shape how AI companies write military-use policies, negotiate contracts and respond when national-security officials reject those boundaries.

Anthropic’s immediate position depends on the courts and on whether the Defense Department changes its designation through a formal action. Lutnick’s comments alone don't lift the ban, and the Pentagon’s current stance means defense access remains constrained even as administration rhetoric softens.

The company also faces a broader commercial question. Amazon and Palantir helped connect Anthropic to defense work, while Pentagon officials have said the department can't rely on any single AI provider. That pressure favors a wider supplier base, including OpenAI, Google and Microsoft, but it also raises switching costs as military organizations standardize models, security controls and procurement rules.

The Anthropic fight has become a test of whether public-sector AI contracts reward maximum access or enforceable safety limits. A government victory would give agencies greater authority to exclude vendors over model restrictions. A court victory for Anthropic would strengthen the case that companies can retain boundaries on sensitive uses without losing federal business through political retaliation.

The contradictory statements also expose coordination problems across the administration. Commerce officials can describe restored trust, but defense procurement decisions operate through separate authorities and legal processes. Until those processes produce a final ruling or a formal Pentagon reversal, Anthropic remains caught between a diplomatic reset and an active defense ban.