Calm. Methodical. Evidence-Based.

Norms Impact

Pentagon sets Friday deadline for Anthropic to abandon ethics rules for AI — or else

The Pentagon is threatening to blacklist a major AI contractor and wield the Defense Production Act to force a private company to abandon self-imposed limits on government use.

Executive

Feb 24, 2026

Sources

Summary

Defense Secretary Pete Hegseth threatened to terminate Anthropic’s Pentagon contract by Friday unless the company accepts the Trump administration’s terms of use, and to designate Anthropic a supply chain risk and invoke the Defense Production Act.
The Pentagon is asserting coercive leverage over a private contractor’s internal use policies by pairing contract termination and supply-chain exclusion with wartime-style statutory authority to compel cooperation.
If carried out, the move would pressure AI providers to discard stated limits on military and domestic applications or risk losing federal business and being effectively blacklisted.

Reality Check

This is the state using procurement power and emergency-style authorities to coerce a private actor’s speech and safety policies, a precedent that can be turned on any company—and, ultimately, on our rights—when officials dislike a constraint. On these facts, the more immediate legal exposure is less about classic bribery than abuse-of-power and coercion: if the threatened “supply chain risk” label is used as retaliation to force policy concessions, it raises grave constitutional concerns and invites scrutiny under federal anti-corruption and honest-services theories (18 U.S.C. §§ 201, 1346) even if proving a traditional quid pro quo is difficult. The Defense Production Act is described here as a tool to “compel the use of its models,” and using it as leverage to override lawful internal red lines—while simultaneously threatening exclusion from federal contracting—tears at core governance norms against weaponizing state power for compliance-by-fear.

Detail

<p>Defense Secretary Pete Hegseth met Tuesday with Anthropic CEO Dario Amodei and warned that Anthropic must set aside concerns about how the Defense Department may use its technology or face termination of its Pentagon work, according to a person familiar with the talks.</p><p>During the meeting, Hegseth set a Friday deadline tied to the Trump administration’s terms of use and threatened to label Anthropic a supply chain risk and invoke the Defense Production Act against the company, the person said. A senior Pentagon official told POLITICO Anthropic has until 5:01 p.m. Friday before the DoD invokes the Defense Production Act “to compel the use of its models,” and that Hegseth will also label Anthropic a supply chain risk at that time.</p><p>Amodei reiterated two usage “red lines” cited by Anthropic: no mass domestic surveillance of U.S. citizens and no physical attacks where AI makes targeting decisions without human input, per the person familiar with the conversation. Anthropic confirmed the meeting and said discussions continued about usage policy and supporting national security “in line with what our models can reliably and responsibly do.”</p>