Anthropic’s core claim is that the DoD’s supply-chain risk designation (and President Trump’s related directive) is unlawful retaliation for protected speech—their public AI safety policy and refusal to remove two narrow exceptions (no fully autonomous lethal weapons; no mass surveillance of American citizens). They argue this violates the First Amendment, Fifth Amendment due process, the Administrative Procedure Act, and the narrow scope of 10 U.S.C. § 3252 (and FASCSA), which they say is reserved for foreign adversary sabotage risks.

This position is unpersuasive on multiple levels:

1. Not protected speech or retaliation: The First Amendment does not grant a private company a constitutional right to dictate the terms of national-security contracts after accepting (or pursuing) hundreds of millions in DoD business. Anthropic was not punished for expressing opinions in a blog post or public statement. It was excluded because it refused to modify its Usage Policy to allow “all lawful uses”—a standard requirement for any defense contractor supplying tools to the warfighter. This is ordinary procurement: the government specifies performance needs; vendors either comply or lose the contract. Courts have long held that commercial dealings with the government are not a protected speech forum, and national-security procurement receives broad deference.

2. Statute was properly applied: The supply-chain risk authority is not limited to Chinese or Russian-linked firms. The statutory definition encompasses any vulnerability that could “create an unacceptable risk to the integrity and availability” of national security systems. A vendor that can unilaterally withhold core functionality (autonomous targeting, broad surveillance) during combat or intelligence operations creates exactly that risk—operational dependency on a private company’s ideological preferences. Negotiations lasted months; the DoD tried the “least restrictive means” (contract amendments). When Anthropic refused, escalation was lawful. The law was never intended to let domestic AI firms become gatekeepers over military effectiveness.

3. No due process violation or pretext: Anthropic received formal notice, a transition window, and had months of direct talks with Secretary Hegseth. The designation followed failed negotiations over a $200 million contract—not sudden animus. Labeling it “pretextual” ignores the public record: Anthropic’s own red lines directly conflicted with the Pentagon’s need for unrestricted lawful use.

In short, Anthropic is attempting to convert its corporate ethics policy into a veto over U.S. defense policy. That is not how national security contracting works.

Why Deciding for the Government Is Necessary—Otherwise the Government Itself Becomes the Security Risk

If courts side with Anthropic and strike down the designation, the real national security damage begins.

• Private companies gain veto power over military AI: Every frontier AI provider (and future startups) would be incentivized to adopt similar “guardrails,” knowing courts will protect them from consequences. The military would face constant negotiation or blacklisting threats whenever a CEO decides certain lawful operations (autonomous swarms, wide-area foreign targeting, or domestic-adjacent intelligence) cross an ethical line. The White House framed it precisely: warfighters must never be “held hostage by the ideological whims of any Big Tech leaders.” Siding with Anthropic turns that hostage situation into settled law.

• Asymmetric disadvantage in great-power conflict: China and Russia deploy AI without self-imposed limits on autonomy or surveillance. U.S. forces already use Claude heavily in ongoing operations. Stripping those capabilities mid-stream—or forcing future models to carry similar restrictions—creates a self-inflicted capability gap. AI is now as central to warfare as satellites or precision munitions; allowing vendors to hobble it is the definition of supply-chain risk.

• Broader supply-chain contagion: The designation’s ripple effects on contractors are intentional and proper. Once one company successfully claims a constitutional right to limit military use, the entire ecosystem fragments. DoD would lose flexibility to surge capabilities in crisis. Contractors would face uncertainty, innovation would shift overseas or to less capable but “unrestricted” providers, and the U.S. would cede the AI arms race to adversaries who treat AI as a pure strategic tool.

• Erosion of civilian control and executive authority: The Constitution assigns defense decisions to the President and Congress, not to private boards in San Francisco. Letting Anthropic prevail would judicially rewrite procurement law to favor corporate policy over national need—an unprecedented and dangerous inversion.

Upholding the government’s action is not punishment; it is basic self-preservation. The military cannot outsource its warfighting edge to companies that reserve the right to say “no” on existential matters. Deciding for Anthropic would not protect free speech or innovation—it would make the U.S. government structurally dependent on private ideological vetoes, turning the world’s strongest military into the ultimate supply-chain risk to itself.

The lawsuits are in their earliest stages; courts will likely give heavy weight to the DoD’s national-security judgment. The precedent set here will shape whether America’s AI advantage remains under democratic control or drifts into corporate hands.

This is a personal opinion and does not constitute legal advice or purport to offer more than a personal opinion.