The Same Blacklist Label for Huawei and an American AI Company

"National security supply chain risk" is the label the Pentagon uses for Huawei, the Chinese telecom giant. This year it applied the same label to Anthropic, an American AI company. The two earned it in opposite ways. Huawei got it for being a foreign adversary. Anthropic got it for refusing to remove the safety guardrails that prohibit mass domestic surveillance and autonomous weapons. The administration's own instrument makes the argument: safety is now the thing being policed. Pete Hegseth, the defense secretary, made the reframing explicit.

America’s warfighters will never be held hostage by the ideological whims of Big Tech. — Pete Hegseth

Trump went further.

WE will decide the fate of our Country — NOT some out-of-control, Radical Left AI company run by people who have no idea what the real World is all about. — Donald Trump

The same logic governs every layer of the technology. Chips are treated as contraband: a Singapore logistics firm is under investigation for smuggling them to China ^[1], and investigators are probing whether Chinese firms reached Nvidia chips through cloud data centers in Malaysia, Thailand, and Singapore ^[2]. Models are treated as controlled goods: the administration is weighing a blanket ban on Chinese AI models ^[3], on top of the access restrictions it already placed on Anthropic's own models. The infrastructure layer is policed like territory: the Justice Department seizes domains ^[4], the House has a bill to ban AI data centers on federal land ^[5], and the NSA issued security guidance for the protocol AI systems use to talk to each other ^[6]. Abroad, the administration rejected UN governance outright ^[7], banned five European digital regulators from entering the country ^[8], and designated an Italian digital-infrastructure collective a global terrorist for providing encrypted communications ^[9]. Michael Kratsios, the White House's AI lead, stated the doctrine at the India AI Summit.

We totally reject global governance of AI. — Michael Kratsios

The old model — guidelines, voluntary frameworks, civilian oversight — is being dismantled outright. A Republican bill would freeze state AI laws for a decade, enforced by withholding $500 million in broadband grants from any state that regulates ^[10]. The Justice Department joined xAI's lawsuit against Colorado's anti-discrimination law, with the assistant attorney general dismissing the law as woke DEI ideology ^[11]. Trump canceled the one voluntary federal vetting framework after Musk and Zuckerberg lobbied him. The NIST guidelines were stripped of their safety language ^[12]. Even that canceled framework was sovereignty-coded from birth — voluntary, led by the NSA, framed around keeping the lead over China ^[13]. It was the softest possible version of containment, and it was still killed. Containment hasn't vanished from the world. Anthropic endorsed California's safety bill ^[14], and the major labs joined the EU's code of practice ^[15]. Bill Gates and Senator Hawley still call for regulation ^[16]. But none of that is a federal instrument. The federal government has left the containment space entirely. The consequence is already in motion. The sovereignty model doesn't just replace safety rules — it penalizes the people who keep them. Anthropic is blacklisted. California is threatened with funding cuts. Colorado is being sued by its own federal government ^[11]. The companies and states that maintain guardrails are being treated as obstacles to national security, not partners in it.