Anthropic Is Not on Your Side

That story is emotionally satisfying but overly simplistic. OpenAI was the first-mover, ostensibly motivated by “AGI for the benefit of all humanity” which, at the time of writing, is still in their charter. On the surface, this sounds very similar to Anthropic’s morality.

After all, Sam Altman is easy to read, what you see is what you get. He wants money, power, control, and the largest possible seat at the table. There is nothing particularly mysterious about that. Throw a dart blindfolded in the Bay Area and you’re bound to hit a founder with the same goals.

Anthropic is harder to read because it speaks in a different register. They talk about safety and alignment, but they go beyond the CBRN and cybersecurity risks that OpenAI focuses on. Anthropic adds in x-risk, and lately, geopolitical dominance as its top-of-mind concerns. They don’t just want to win the AI race on business terms, they believe they have a personal mission to save humanity. 

That is precisely why they deserve more scrutiny.

Anthropic publishes AI safety research, just like OpenAI. Likewise, Dario Amodei has spoken with the same moral seriousness and gravity as Sam Altman about risks to humanity. It was Sam, after all, who went before Congress to say “lights out for us” as the worst-case scenario. On paper, there’s not much difference between Sam and Dario.

So, why am I making the case that Anthropic deserves just as much scrutiny as OpenAI? What makes them stand out as “more dangerous”? 

Anthropic’s rise this year has been meteoric. The company reportedly crossed a $30 billion annualized revenue run rate by April, up from about $9 billion at the end of 2025. Meanwhile, OpenAI had reportedly topped $25 billion annualized revenue as of the end of February. While these numbers are not published by the SEC, the trend is undeniable. Anthropic is no longer the safety lab throwing stones at the establishment.

But who is Anthropic’s customer? OpenAI’s flex is that they serve 900 million humans around the globe. Sam Altman came up through Y Combinator, and at the time, they worshipped the Facebook model. Volume was everything. If you can capture most of the world’s internet users, revenue will come. Anthropic took the opposite approach. Taking a page out of Alex Hormozi’s book, Anthropic decided to “sell to the richest people you can.” In this case, that is Fortune 500 companies. 

To that end, Anthropic has surpassed OpenAI’s enterprise adoption, according to Ramp adoption data. Anthropic leads with 34.4% to OpenAI’s 32.3%, driven largely by Claude Code.

The takeaway here is simple. Anthropic serves the corporate establishment, first and foremost. 

Anthropic’s public image is presently shaped by the idea that it “stood up to the Pentagon.” But the more nuanced version is far less heroic. Dario was first in line to get in bed with the Pentagon. Amodei said Anthropic had “worked proactively” to deploy models to the Department of War and the intelligence community, and that it was the first frontier AI company to deploy models in U.S. classified networks and provide custom models for national-security customers. He also said Claude was extensively deployed for intelligence analysis, operational planning, and cyber operations.

The next iteration of the military industrial complex is already taking shape.

In November 2024, Anthropic partnered with Palantir and AWS to provide Claude to U.S. intelligence and defense agencies. TechCrunch reported that Claude would be operationalized inside Palantir’s platform and made available in Palantir’s defense-accredited Impact Level 6 environment, reserved for highly sensitive national-security systems. 

Then Anthropic launched “Claude Gov” which are custom models built exclusively for U.S. national-security customers. Anthropic’s own description says these models were built from government feedback and optimized for classified materials, defense documents, intelligence analysis, and cybersecurity data.

Then came the Pentagon contracts. In July 2025, the Pentagon’s Chief Digital and Artificial Intelligence Office awarded Anthropic, Google, OpenAI, and xAI agreements with $200 million ceilings to accelerate frontier AI adoption across national-security departments. While Anthropic was not alone in line, they have been very deliberately integrating with the government and military at every touchpoint they can access.

But then, earlier this year, the sweetheart deals came to a rupture.

Dario Amodei said, “Disagreeing with the government is the most American thing in the world,” and he paired that with “we are patriots” and “we have stood up for the values of this country.” He was not saying “the Pentagon is immoral, or surveillance is immoral.” He was only drawing two narrow red lines (after the fact, even).

And those red lines are much narrower than the public thinks.

In his statement on the Pentagon dispute, Amodei did not say autonomous weapons are immoral. He said partially autonomous weapons are “vital to the defense of democracy,” and that even fully autonomous weapons “may prove critical” for national defense, but that current frontier models are not reliable enough to power them. He also explicitly supported AI for lawful foreign intelligence and counterintelligence while objecting to mass domestic surveillance.

So the question was not one of morality, only technical ability. His caution was “not yet” not “it’s wrong.”

This distinction matters. Anthropic’s stance is not “AI should not power surveillance or autonomous warfare.” Their stance is closer to, “AI should strengthen the U.S. national-security state, including intelligence and increasingly autonomous warfare, so long as Anthropic can preserve two exceptions—no current fully autonomous kill chain, and no mass domestic surveillance of Americans.”

Anthropic’s primary mistake was trying to litigate military jurisprudence via contract, but make no mistake, they are fully aligned with American exceptionalism and projection of power. 

The Palantir relationship makes the contradiction unavoidable. Palantir was founded by Peter Thiel and Alex Karp with early backing from In-Q-Tel, the CIA’s venture-capital arm, and built its business around intelligence, military, policing, and data-fusion work. In 2026, critics reacted sharply to Palantir’s “Technological Republic” manifesto, which emphasized hard power, the duty of tech companies to participate in defense, and the inevitability of AI weapons.

We need to pause and lampshade an implication here. I am not saying that US military dominance is a bad thing. There are plenty who would. My sole point here is that Anthropic’s behavior is not “fighting for the little guy” or prioritizing civil rights. They are behaviorally no different from any hard-right big-stick war hawk. 

This is where the Thiel comparison becomes useful. Dario Amodei is not Peter Thiel but Anthropic’s worldview is closer to the Thiel/Karp world than its softer branding suggests. Their ideology includes civilizational stakes, sovereign power, strategic competition, elite stewardship, and the conviction that the “right people” must hold the dangerous tools before the wrong people do. And of course, they are the “right stuff” by their own definition. 

Conversely, Sam Altman’s danger is legible. He wants personal power. That’s it. He wants to be important. Money, fame, influence, leverage. Fine. But he’s been totally agnostic about where that comes from. Consider his world tour, where he wanted to build critical AI infrastructure in the middle east, something that would seem unlikely for the “US-first-and-only” values of Karp and Thiel.

Dario Amodei’s danger is subtler. He can explain, in morally serious language, why acquiring power is the responsible thing to do. Not only responsible, but necessary for humanity. And specifically, that he and Anthropic deserve that power. However, when Congress asked Altman “why should we trust you” he flatly said “you shouldn’t.” That was, perhaps, the most honest thing Sam Altman has ever said.

Now, through this belief structure, Anthropic can logically and ethically justify everything from regulatory capture to taking Pentagon money. The ends justify the means, and to spell it out plainly, they believe that advanced AI represents an existential threat to humanity, and so they must be the adults in the room. Well, in order to safely bring ASI into existence, they must be in the driver’s seat. Thus, anything that keeps them in the driver’s seat is fine. It’s a Faustian bargain of ends justifying the means. 

In retrospect, Sam Altman looks like a mercenary who says what he needs to say, mostly for PR and optics reasons. He has spoken grandly about benefitting humanity, but his actions say “so long as I’m in control of OpenAI and I get a seat at the table.” 

Dario, however, seems to be a “true believer” in x-risk and everything else that entails. When he says that AI can kill humanity, it is partly scare-based marketing, but it is also his true worldview. Fear justifies control, which is a playbook as old as civilization itself.