The Pentagon blacklisted Anthropic on February 27, 2026, marking the first time a U.S.-based AI company has been declared a “supply chain risk” to national security. Defense Secretary Pete Hegseth’s designation came after Anthropic CEO Dario Amodei refused to remove AI safety guardrails preventing mass domestic surveillance and autonomous weapons—two “red lines” the company would not cross. Hours later, OpenAI signed a competing Pentagon deal with reportedly similar restrictions, exposing contradictions in the government’s stance. Then came the irony: Claude shot to #1 in the U.S. App Store on March 1, overtaking ChatGPT as downloads surged 60%.
First American AI Company Declared “Supply Chain Risk”
Hegseth’s designation carries unprecedented weight. The “supply chain risk” label is typically reserved for adversarial foreign companies like Huawei and ZTE. Legal experts at Just Security note there is no public record of any American company receiving this designation from the Pentagon before Anthropic.
The blacklist cancels Anthropic’s $200 million Pentagon contract immediately and forces all military contractors to certify they don’t use Claude in their workflows within six months. Contractors face a stark choice: work with the Pentagon or work with Anthropic. Not both.
This weaponizes a national security tool against a domestic company for taking an ethical stance. It raises questions about government overreach when companies refuse contracts on safety grounds.
Anthropic’s Non-Negotiable Safety Guardrails
Anthropic demanded explicit contractual prohibitions against two specific uses: mass domestic surveillance of U.S. citizens, and fully autonomous weapons systems that select and engage targets without human control. The Pentagon wanted “all lawful purposes” language instead. Amodei said the company “cannot in good conscience accede” to those terms.
“We are still advocating for those red lines. We’re not going to move on those red lines,” Amodei told CBS News. “Disagreeing with the government is the most American thing in the world, and we are patriots in everything we have done here.”
The Pentagon argued that current DOD policy already prohibits these uses. Why insist on contractual language? Because policies can be changed unilaterally. Contract terms bind both parties. That legal distinction explains Anthropic’s stance: they wanted permanent safeguards, not changeable policies.
Related: Pentagon AI Coding Tools: “Tens of Thousands” Join $8.5B Bet
Claude Hits #1 in App Store After Blacklist
The Pentagon’s punishment backfired spectacularly. Claude overtook ChatGPT to claim #1 in the U.S. App Store on March 1, just days after the blacklist. Downloads surged to 503,424 on February 28—a single-day record. Free users increased 60%. Paid subscribers more than doubled. Daily sign-ups broke all-time records every day this week.
The public voted with their wallets. Pop singer Katy Perry posted a screenshot of buying Claude Pro subscription with the caption “Done.” On Reddit, a post titled “Cancel and Delete ChatGPT!!!” racked up 30,000+ upvotes as #CancelChatGPT trended on social media.
The blacklist intended to isolate Anthropic but instead positioned them as ethical leaders in AI safety. Transparency about principles built consumer trust even as it lost government contracts. Classic Streisand effect.
OpenAI Gets Pentagon Deal With “Similar” Restrictions
Hours after Anthropic was blacklisted on February 27, OpenAI announced a Pentagon deal. OpenAI claims its contract includes similar prohibitions on domestic mass surveillance and autonomous weapons—the same guardrails that got Anthropic blacklisted. The Pentagon approved OpenAI’s terms but rejected Anthropic’s. The contradiction is glaring.
The difference appears to be contract language. OpenAI reportedly uses “principles” language (aspirational) while Anthropic demanded “prohibitions” (legally binding). That subtle legal distinction may explain the Pentagon’s approval of one and rejection of the other.
OpenAI faced massive backlash for taking the deal. Bloomberg reported OpenAI “claims safety exceeds Anthropic’s,” but developers weren’t buying it. The timing—hours after Anthropic’s blacklist—looked opportunistic at best.
Related: OpenAI Calls in McKinsey, BCG to Sell Enterprise AI
What Happens Next
Anthropic faces existential questions. Can it survive without government contracts? The company is pursuing legal challenges to the supply chain designation. There’s no precedent for an American company fighting this designation, so the legal path is unclear.
Reports suggest the U.S. military used Claude during Iran strikes hours after Trump announced the ban. Enforcement appears chaotic and inconsistent. The Pentagon must replace Claude integrations across military contractors within six months—a scramble given how deeply AI tools are embedded in workflows.
This is the first major clash between AI safety principles and national security demands. The outcome will shape how future AI/government conflicts unfold. Anthropic bet that consumer and enterprise growth can offset government revenue losses. The app download surge suggests they might be right, but it’s too early to declare victory.
