Howard Lutnick, the Commerce Secretary who somehow makes every room feel like a used car lot, just told Axios that the Trump administration now fully trusts Anthropic after months of ugly clashes over AI and national security. The key phrase he used was that Anthropic is "back on the right side." The right side of what, exactly, is the question nobody in that interview pushed hard enough to answer.
What Lutnick Actually Said
In an interview with Axios' Mike Allen in Chapel Hill, North Carolina on Tuesday, Lutnick was asked point blank whether he trusts Anthropic CEO Dario Amodei. His answer: "We trust Anthropic. They've done what we asked. They're back on the right side. So the answer is: Yes."
Those three sentences are doing a tremendous amount of work. "They've done what we asked" implies there was a list of demands. "Back on the right side" implies there was a wrong side. And a Commerce Secretary publicly grading an AI company's ideological positioning like a teacher handing back a corrected exam is, at minimum, worth a raised eyebrow.
Axios is reporting this as a declaration of restored trust following what it describes as a "bitter fight" over national security and AI safeguards. The word "bitter" is doing some heavy lifting there. These were not two neighbors arguing about a fence.
The Fight That Got Buried in the Peace Treaty
The backstory here matters. Anthropic, which makes the Claude AI systems and has historically positioned itself as the safety-conscious adult in the room compared to some of its competitors, spent months in open friction with the Trump administration over how AI should be governed, what national security guardrails should look like, and who gets to make those calls.
This was not a small disagreement over font sizes on a compliance form. The administration's posture toward AI companies has been to demand access, demand alignment, and make clear that companies who push back on national security framing will find the regulatory environment considerably less friendly. Anthropic pushed back. And then, apparently, Anthropic stopped pushing back.
We do not yet have the full picture of what "they've done what we asked" means in concrete terms. That is a significant gap in this story, and it should bother anyone who cares about what the most powerful AI systems in the world are actually being asked to do and by whom.
The Phrase 'Right Side' Should Give You Pause
Let's sit with "back on the right side" for a second, because that is an unusual way for a cabinet secretary to describe a private technology company's relationship with the federal government.
Companies do not have sides in the way Lutnick is implying. They have legal obligations, contractual relationships, regulatory compliance requirements. When a government official frames a company's cooperation as a matter of being on the "right side," that language belongs in a different category entirely. It implies loyalty. It implies that the alternative was being on the wrong side, which implies consequences for that position.
This is the kind of language you use about an ally who was briefly flirting with the enemy. Applied to a domestic AI company operating in the United States, it tells you something uncomfortable about how this administration views the relationship between private tech and government authority.
Why This Matters Beyond the Two Companies Involved
Anthropic is not some scrappy startup. It is one of the two or three most consequential AI labs in the world. Its models are embedded in products used by millions of people. The decisions Anthropic makes about how its systems behave, what they will and won't do, and who has access to what capabilities ripple out in ways that affect basically everyone who touches modern software.
So when the Commerce Secretary of the United States announces that Anthropic has done "what we asked" and is now trusted again, that is a story with stakes that go well beyond a corporate relationship getting patched up. What were the asks? Were they about export controls? Model capabilities? Data access? Government contracts? The public does not know, and based on Tuesday's interview, Axios did not get that answer either.
The Trump administration has been aggressive and fairly open about wanting American AI companies to be instruments of American power projection. That is not a fringe position and it is not without legitimate national security rationale. But "instruments of American power" can mean a lot of different things depending on who is holding the instrument.
Amodei's Silence Speaks Volumes
As of this reporting, Dario Amodei has not publicly responded to Lutnick's comments. That absence is its own kind of statement.
Amodei has historically been one of the more publicly engaged AI executives on questions of safety and governance. He has testified before Congress. He has written extensively about AI risk. The idea that his company's relationship with the federal government was resolved through a process that ends with Howard Lutnick announcing the verdict on a Tuesday in North Carolina, with Amodei nowhere in sight, is a strange look.
Maybe Amodei is fine with all of this. Maybe the settlement, or whatever we are calling it, reflects genuine common ground rather than capitulation. Or maybe running one of the most powerful AI companies in the world while the federal government is making pointed demands about which side you are on is a position where silence is just the safest available option.
The Dingo Take
"They've done what we asked." Read that back one more time. A cabinet secretary of the United States government just announced, publicly and with evident satisfaction, that a major private technology company complied with a list of demands and has therefore been welcomed back into good standing. That is not a normal sentence to hear in a democratic country with a free market economy. It is a normal sentence in a different kind of arrangement.
None of this means Anthropic did anything wrong. Companies negotiate with governments constantly, and national security considerations around AI are real and serious. But the way Lutnick framed this, the language of sides and trust and compliance, should generate immediate follow-up questions that are sharper than anything Tuesday's interview produced. What specifically did the administration ask for? What did Anthropic agree to? Who made those decisions, and is there any oversight mechanism beyond Howard Lutnick deciding that a company has passed its loyalty test?
The AI governance story is going to be one of the defining fights of the next decade, and right now it is being conducted almost entirely in closed rooms, announced in brief quotes to friendly outlets, and framed by officials who have every incentive to present the outcomes as tidy and consensual. This one does not feel tidy. It feels like a company that got squeezed and gave something up, and the rest of us are supposed to just trust that what they gave up was fine.
Comments