On Friday afternoon, President Donald Trump ordered every federal agency to “immediately cease all use of Anthropic’s technology” and directed Defense Secretary Pete Hegseth to designate the company a “supply chain risk to national security” after Anthropic refused to grant the Pentagon unrestricted access to its Claude AI system. In an unhinged post on Truth Social, Trump threatened “major civil and criminal consequences” if the company does not cooperate during a six-month phase-out period.
The dispute between Anthropic and the Pentagon unfolded rapidly this week and reveals growing divisions within the ruling class over the Trump administration’s reckless policies. Two days after being summoned to Washington on Tuesday, Anthropic CEO Dario Amodei published a statement Thursday evening refusing to back down on their demands that the Pentagon not use their AI model for fully autonomous drones or domestic mass surveillance.
While this was presented in the corporate press as a courageous act of defiance against Trumpian authoritarianism, read in full, the letter is nothing of the kind. It is a full-throated declaration of support for US imperialism that repeats the Pentagon’s own talking points nearly verbatim, differing only on two narrow technical restrictions.
“I believe deeply in the existential importance of using AI to defend the United States and other democracies, and to defeat our autocratic adversaries,” Amodei wrote. He boasts that Anthropic was the first frontier AI company to deploy on classified government networks, at national laboratories, and to provide custom models for national security customers. “Partially autonomous weapons, like those used today in Ukraine, are vital to the defense of democracy,” he declared, adding, “Even fully autonomous weapons (those that take humans out of the loop entirely and automate selecting and engaging targets) may prove critical for our national defense.”
Thus, his objection to autonomous weapons is not principled, but technical: frontier AI systems are “simply not reliable enough” at present.
Most revealing of all, Amodei wrote, “We have never raised objections to particular military operations nor attempted to limit use of our technology in an ad hoc manner.” Here Amodei confirmed that Anthropic raised no objection to the Pentagon’s military assault on Caracas in early January, an operation that killed between 83 and 100 people and led to the illegal seizure of Venezuelan president Nicolas Maduro, and which ostensibly triggered this crisis. Not only that, he has never objected to any other US military operation!
The man being hailed as a champion of ethical AI effectively told the Pentagon: we support everything you have done; we merely request two technical carve-outs going forward.
In response to this letter, which was entirely deferential aside from the two demands, Trump wrote on Truth Social: “The Left wing nut jobs at Anthropic have made a DISASTROUS MISTAKE trying to STRONG-ARM the Department of War, and force them to obey their Terms of Service instead of our Constitution.” He added, “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.”
Trump’s fascistic ravings against Anthropic and its CEO Dario Amodei are those of a dictator unwilling to tolerate the most minimal limits to his unchecked power. While Anthropic does have a reputation as the “safety-first” AI company, this reputation was always more marketing than substance, and any claims that it is “radical left” are absurd.
Anthropic is a $380 billion AI company backed by $8 billion from Amazon—whose AWS built and operates the CIA’s primary cloud infrastructure—$3 billion from Google, and $15 billion from Microsoft and Nvidia combined. It celebrated its $200 million Pentagon contract in July 2025, and partnered with Palantir—whose entire business model is built on serving the US military and intelligence apparatus, from drone targeting to immigrant tracking for ICE—to deploy Claude on classified networks.
The supply chain risk designation carries sweeping economic consequences. All military contractors must now certify that they do not use Anthropic’s products in their workflows. This includes Boeing and Lockheed Martin, who were asked earlier this week to assess their “exposure” to Claude. The $200 million Pentagon contract is terminated. As one analysis noted, using Claude will now become an affirmative choice to forgo any current or future US government business, effectively making Anthropic toxic not only to defense contractors but to any company in any sector that does or seeks to do business with the federal government. The ripple effects will extend globally, as allied governments and multinational corporations reassess their own use of Anthropic’s tools.
The two “red lines” Amodei draws—against mass domestic surveillance and fully autonomous weapons—are directed against practices already firmly established by the agencies he enthusiastically serves.
The Snowden revelations of 2013 exposed the NSA’s PRISM program, which collected data from nine major technology companies, and upstream collection programs that tapped fiber-optic cables to vacuum up internet traffic in bulk. Section 702 of the Foreign Intelligence Surveillance Act, the legal foundation for warrant-less surveillance, has been repeatedly reauthorized by Congress with bipartisan support, most recently in 2024. The intelligence agencies have deployed machine learning and AI-assisted analysis of mass surveillance data for years. Amodei’s “red line” against mass domestic surveillance is a line drawn against something already happening, drawn by a man who has built his business atop the classified networks where that surveillance operates.
The Trump administration’s response to Amodei’s limited resistance has itself been revealing. Under Secretary of Defense Emil Michael denounced Amodei on X as a “liar” with a “God complex” who “wants nothing more than to try to personally control the US Military.” This is the language of fascist political intimidation, in which personal vilification is deployed as a weapon to enforce submission. That a senior government official publicly brands a CEO as mentally unfit for maintaining that AI should not be used for mass surveillance reveals the character of the regime.
Despite its limited character, the refusal of Anthropic to cave on these two demands has drawn broad support among tech workers, both at other AI companies and more widely.
As of this writing, 558 current Google employees and 91 current OpenAI employees have publicly signed an open letter declaring: “The Pentagon is negotiating with Google and OpenAI to try to get them to agree to what Anthropic has refused. They’re trying to divide each company with fear that the other will give in. That strategy only works if none of us know where the others stand.”
In response to this, Open AI CEO Sam Altman separately declared that OpenAI shares Anthropic’s “red lines.” Only hours later, however, late Friday Altman tweeted that OpenAI reached a deal with the Pentagon to “deploy our models in their classified network.” The terms of the deal are not yet published, but Altman claims that OpenAI is demanding the same two prohibitions that Anthropic maintained. Undoubtedly there are loopholes which will soon become clear. Altman’s tweet ends ominously, “The world is a complicated, messy, and sometimes dangerous place.”
A second, broader letter from Amazon Employees for Climate Justice, the Alphabet Workers Union, No Tech for Apartheid, and No Azure for Apartheid—organizations which claim to represent 700,000 workers—goes further, stating: “We are writing to urge our own companies to also refuse to comply should they or the frontier labs they invest in enter into further contracts with the Pentagon.” It demands that “executive leadership at Google, Microsoft and Amazon must reject the Pentagon’s advances and provide workers with transparency about contracts with other repressive state agencies including DHS, CBP and ICE.” Significantly, this letter moves beyond Amodei’s framework, which objects only to surveillance of Americans, to oppose surveillance of immigrants as well.
But both letters remain within the framework of appeals to corporate management and the state. Neither demands public ownership of AI, democratic control by workers, or the termination of military contracts as such. The critical question is whether these workers will develop an independent political perspective—opposing the capitalist state and its military apparatus as a whole—or remain a pressure group for one faction of capital against another. The International Workers Alliance of Rank-and-File Committees (IWA-RFC) provides the organizational framework through which tech workers can fight for an independent class program based on the demands for public ownership and democratic control of AI, the end of all military AI contracts, and opposition to imperialist war.
The growing dangers of the use of AI by the military were underscored this week by a scientific study which placed Claude, ChatGPT, and Gemini in armed conflict simulations. AI models chose to deploy nuclear weapons in 95 percent of scenarios, while Claude recommended nuclear strikes in 64 percent of games.
The nuclear question has also evidently been central to the dispute between Anthropic and the Pentagon. The Washington Post reported on February 27 that at a meeting last month, Michael posed a hypothetical to Amodei: if an ICBM were hurtling toward the United States with “the time to make a decision measured in minutes and seconds,” could Claude help shoot it down? Pentagon officials claim Amodei responded, “You could call us and we’d work it out”—an account Anthropic calls “patently false,” noting it had already agreed to allow Claude for missile defense in December 2025.
The dispute between Anthropic and the Pentagon is a conflict within the ruling class. The working class, including tech workers, must chart its own independent path. The fight against AI militarization requires the independent political mobilization of the international working class for the public ownership and democratic control of AI, the dismantling of the surveillance state, and the end of imperialist war.