If the plaintiff’s allegations in Anthropic PBC v. US Department of War hold up in federal court in Northern California, then a bellicose post by President Donald Trump and one by Secretary of War Pete Hegseth lambasting “the ideological whims of Big Tech” and Anthropic’s “defective altruism” likely will have helped the business’s First Amendment retaliation claim.

A victory for the artificial intelligence company also would rebuke the Trump administration’s willingness to retributively punish Anthropic—“destroy one of the most innovative companies in America,” as AEI’s William C. Greenwalt wrote—because it refused to compromise its views and values. In short, if Anthropic wins its First Amendment claim, then the predictable presidential name-calling that works well with MAGA adherents—Trump dubbed Anthropic “A RADICAL LEFT, WOKE COMPANY” populated by “Leftwing nut jobs”—probably will have hurt the government by exposing the ideological (not national security) motivation underlying its anti-Anthropic conduct.

Background. Anthropic sued following Trump and Hegseth’s posts and a formal Hegseth letter deeming Anthropic a “supply chain risk” to national security, marking the “first such designation ever applied to an American company.” Hegseth also “terminated Anthropic’s work with the Pentagon and other government agencies” and barred companies doing military business from “conduct[ing] any commercial activity with Anthropic.”  

This transpired after Anthropic—during contractual negotiations with the Department of War—didn’t accede to Hegseth’s demand that its AI products be used for all “lawful” purposes. Hegseth’s mandate conflicted with Anthropic’s policy that its Claude AI model “not be used for mass surveillance in the US or to power entirely autonomous weapons, such as a drone that uses AI to kill targets without human approval.”

While Anthropic’s complaint includes five theories, this post examines its First Amendment retaliation claim. The lawsuit’s outcome matters immensely to American businesses because, as the Wall Street Journal’s Greg Ip wrote, “a win for the government would effectively empower the president to cripple any company over political differences.”

First Amendment Claim. Under Supreme Court precedent, the First Amendment generally “‘prohibits government officials from subjecting an individual to retaliatory actions’ for engaging in protected speech.” The Court has determined that “a plaintiff pursuing a First Amendment retaliation claim must show . . . that the government took an ‘adverse action’ in response to his [protected] speech that ‘would not have been taken absent the retaliatory motive.’” Plaintiffs also must establish that the adverse action was “material,” such that it “would ‘chill a person of ordinary firmness’ in the plaintiff’s position from engaging in ‘future First Amendment activity.’” This means Anthropic must prove:

(1) It engaged in protected speech;

(2) The adverse actions of terminating its government work and designating it a supply chain risk to national security “would chill a person of ordinary firmness from continuing to engage in” the protected speech; and  

(3) Anthropic’s protected speech “was a substantial or motivating factor” underlying the government’s actions.

Protected Speech. Anthropic alleges “the federal government . . . retaliated against it for expressing” the company’s “primary animating principle . . . that the most capable [AI] systems should also be the safest and the most responsible.” As expressed in “Anthropic’s Usage Policy,” this principle means that Claude cannot be used for “lethal autonomous warfare” and “surveillance of Americans en masse.” Anthropic also expressed these views during negotiations with the government and in public statements.

Underlying Anthropic’s prohibition on the two uses described above is its “unique understanding of Claude’s risks and limitations—including Claude’s capacity to make mistakes.” In sum, Anthropic argues that expressing its views about “AI safety and the limitations of its own AI models” is both “of significant public importance” and protected by the First Amendment, but that Hegseth punished it for expressing and adhering to those views rather than capitulating to his demands.

Chilling Effect. Anthropic asserts that the adverse actions of contractual termination and supply chain risk designation “impose significant financial and reputational costs on Anthropic that would chill a company of ordinary firmness from continuing to engage in expressive activity.” It calls the supply chain risk designation an “extraordinarily stigmatizing label.”

Retaliatory Motivation. Anthropic contends Hegseth—in his online post and subsequent letter—failed to provide details about the supply chain risk to national security that Claude supposedly poses. Indeed, Anthropic emphasizes that Hegseth “himself has recognized Claude’s capabilities as ‘exquisite.’”

The complaint thus avers that the real motive underlying Hegseth’s adverse actions was ideological “disagreement with Anthropic’s views” on matters “of AI safety and responsible AI use.” This is where Trump and Hegseth’s posts may help prove Anthropic’s case. Indeed, Hegseth posted that he’d look for “a more patriotic service”—one not “held hostage by the ideological whims of Big Tech,” “defective altruism,” “Silicon Valley ideology,” and “corporate virtue-signaling.” That sounds a lot like viewpoint retaliation, not simply MAGA signaling.