Anthropic pushes back on US military’s “supply‑chain risk” claim, cites guardrails need.

Anthropic’s legal team announced Friday that it will “challenge any supply‑chain risk designation in court,” after Secretary of Defense Pete Hegseth labeled the AI startup a security threat and barred all Pentagon‑linked contractors from commercial dealings with the firm, Wired reported. The move follows weeks of tense negotiations in which the Defense Department demanded unrestricted “all lawful uses” of Anthropic’s Claude models, while the company insisted its contracts must prohibit mass domestic surveillance and fully autonomous weapons. Anthropic’s blog post said it had received no direct communication from the DoD or the White House about the designation, underscoring the abruptness of the order.

The Pentagon’s supply‑chain risk label is a statutory tool that allows the department to exclude vendors deemed vulnerable to foreign influence or other security gaps, according to Wired. Hegseth’s tweet warned that any contractor, supplier, or partner doing business with the military must cease all commercial activity with Anthropic “effective immediately.” The designation could cascade across the defense industrial base, forcing tech firms that rely on Anthropic’s APIs to drop the service or risk losing DoD contracts.

Anthropic’s CEO Dario Amodei pushed back in a separate blog entry, arguing that the guardrails the company seeks are essential to prevent AI from being used in ways that could undermine democratic values. The Atlantic detailed that Hegseth had reportedly threatened to invoke the Defense Production Act or impose the supply‑chain risk label unless Anthropic stripped those ethical safeguards by Friday. Amodei rejected the ultimatum, stating that “in a narrow set of cases, we believe AI can undermine, rather than defend, democratic values,” and that the Pentagon’s threats would not change Anthropic’s stance.

Industry analysts note that the standoff could have broader implications for the U.S. AI market. TechCrunch highlighted that the dispute centers on whether the military can access powerful foundation models without the developer’s consent to limit certain applications. If the Pentagon proceeds with the supply‑chain risk label, contractors may be forced to choose between a lucrative defense pipeline and a leading generative‑AI platform, a dilemma that could reshape vendor relationships across the sector.

The outcome remains uncertain, but both sides signal a willingness to test the limits of government authority over private AI technology. Anthropic’s promise to litigate the designation sets a precedent for how AI firms might resist future security designations, while the DoD’s aggressive posture underscores the growing strategic importance of AI in national defense. The next legal filings will likely determine whether the Pentagon can enforce its “all lawful uses” demand or whether corporate guardrails will prevail in the emerging AI‑security landscape.