The more relevant question is who is held accountable for the war crimes? OpenAI seem pretty confident it won't be OpenAI.
I can see the logic if we were talking about dumb weapons--the old debate about guns don't kill people, people kill people. Except now we are in fact talking about guns that kill people.
Does OpenAI enforce those red lines in all contracts?
From what I can tell the Anthropic issue was triggered by something Palantir was doing as a contractor for DoW, not anything related to direct contracts between DoW and Anthropic, and DoW was annoyed that Anthropic interfered with what Palantir was up to.
In other words will OpenAI enforce these "red lines" against use by a third-party government contractor?
If not, this seems pretty meaningless if they are essentially playing PR while hiding behind Palantir.
What people don't understand is that domestic surveillance by the government doesn't happen and isn't needed. They know it's illegal and unpopular and for over two decades they have a loophole. Since the Bush administration it's been arranged for private contractors to do the domestic surveillance on the government's behalf. Entire industries have been built around creating "business records" for no other purpose than to sell them to the government to support domestic surveillance. This is entirely legal and why the DoW has been able to get away with saying things like "domestic surveillance is illegal, we don't do that" for over two decades while simultaneously throwing a shit fit about needing "all legal uses" if their access to domestic surveillance is threatened.
There's a big difference between "the government won't use our tools for domestic surveillance" (DoW/DoD/OpenAI/etc) and "we won't allow anyone to use our tools to support domestic surveillance by the government" (Anthropic)
Hegseth and the current Trump admin are completely incompetent in execution of just about everything but competent administrations (of both parties) have been playing this game for a long time and it's already a lost cause.
I was pondering the same thing and to me the answer is a contractor sold something to the DoD and Anthropic pulled the rug out from under that contractor and the DoD isn't happy about losing that.
My speculation is the "business records" domestic surveillance loophole Bush expanded (and that Palantir is build to service). That's usually how the government double-speaks its very real domestic surveillance programs. "It's technically not the government spying on you, it's private companies!" It's also why Hegseth can claim Anthropic is lying. It's not about direct government contracts. It's about contractors and the business records funnel.
Yes, I assumed a mass surveillance Palantir program also. Interesting take on how it allows them to claim “we are not doing this” while asking Anthropic to do it.
Of course they can just say - we aren’t, Palantir is.
Anthropic didn't even say "no", it was more of a "not yet, let's work on this".
I really wonder what Palantir's role in all this is because domestic surveillance sounds exactly like Palantir and whatever happened during the Maduro raid led to Anthropic asking Palantir questions which the news reports is the snowball that escalated to this.
This is a summary from Gemini of the news reporting:
Recent news reports from February 2026 indicate that a significant rift developed between Anthropic and the Department of War (Pentagon) following the capture of Venezuelan President Nicolás Maduro in January 2026.
According to a report by the Wall Street Journal (referenced by TRT World and others on February 14–15, 2026), the controversy originated when an Anthropic employee contacted a counterpart at Palantir Technologies to inquire about how Claude had been used during the raid.
Key Details of the Reports:
* Discovery of Use: Anthropic reportedly became aware that its AI model, Claude, was used in the classified military operation through its existing partnership with Palantir. This was allegedly the first time an Anthropic model was confirmed to be involved in a high-profile, classified kinetic operation.
* The Inquest: The Wall Street Journal and Semafor reported that an Anthropic staff member reached out to Palantir to ask for specifics on Claude's role. This inquiry reportedly "triggered the current crisis" because it signaled to the Pentagon that Anthropic was attempting to monitor or place "ad hoc" limits on how its technology was being used in active missions.
* The Confrontation: During a recent meeting between Anthropic CEO Dario Amodei and Defense Secretary Pete Hegseth, the inquiry to Palantir was a point of contention. Hegseth reportedly claimed Anthropic had raised concerns directly to Palantir about the Caracas raid. Amodei has since denied that the company raised objections to specific operations, characterizing the exchange with Palantir as a routine technical follow-up or a "self-serving characterization" by Palantir.
* Current Status: This friction has escalated into a public showdown. Today, Friday, February 27, 2026, reports indicate that the Trump administration has officially designated Anthropic a "supply chain risk" and ordered federal agencies to cease using Claude after the company refused to remove guardrails related to autonomous weaponry and mass domestic surveillance.
The primary reporting you are likely recalling comes from The Wall Street Journal (approx. February 14, 2026) and was later expanded upon by Semafor regarding the specific communications between Anthropic and Palantir employees.
Maybe Anthropic could replace its employees with AI. Unlikely the admin is going to enjoy setting precedent that employees are protected against being replaced by AI.
It's really not the right thing to be bikeshedding. The people calling the shots call themselves the Department of War. No need to die on hills that don't matter.
It's actually a good thing to point out, because it shows that those people are out of control and exceeding their authority, and need to be reined in.
No need to die on the hill, but point out that there's a consistent pattern of lawless power-grabbing.
> it shows that those people are out of control and exceeding their authority
No, the concentration camps and gangs of masked thugs violating civil rights are that sign. Threatening to treat a domestic private corporation like an enemy combatant during peacetime for not immediately caving to military demands is that sign. Trying to take over the Federal Reserve, the Federal Trade Commission, and the Nuclear Regulatory Commission, is that sign. The Executive attempting to freeze funds issued by Congress for partisan reasons is that sign.
Department of War is just little boys being trolls.
The action of a failed rebrand belongs to the Department of Defense, and is indeed an example of exceeding their authority. It was not DoD that is trying totake over the Fed, the FTR, or the NRC, so those examples don't work against Hegseth here.
Anthropic is in negotiation with Hegseth/DoD. Pointing out all the specific actions that Hegseth is doing are fair game to show that Hegseth is nuts.
Bringing in other complaints against other parties, however bad those other parties are behaving, shows a pattern in other people, which might be helpful too. But hegseth's direct actions are stronger evidence.
According to the constitution, Congress is the check and balance on this. If congress refuses act as they are supposed to, it's up to the rest of our democracy to exert force on them, shame them, recognize what's going on, talk to our neighbors, etc.
If the current congress doesn't take action, in 2027 it's quite likely they will.
Of course the most likely current course is that nobody reins in Hegseth/DoD right now, but even if there's no official consequences at the moment there should be a memory and political will to change the system to prevent such abuse in the future.
You're talking about an administration that barred the AP from pressed briefings because they didn't call it the Gulf of America. This is not a bikeshed.
> It's really not the right thing to be bikeshedding. The people calling the shots call themselves the Department of War. No need to die on hills that don't matter.
From the first chapter of the book On Tyranny by Timothy Snyder, an historian of Central and Eastern Europe, the Soviet Union, and the Holocaust:
Commenting on the matter just makes it easier for the media to yap about Anthropic being "woke" rather than focusing on the Department of War's demands.
TIL of Bikeshedding, or Parkinson’s Law of Triviality.
Defined as the tendency for teams to devote disproportionate time and energy to trivial, easy-to-understand issues while neglecting complex, high-stakes decisions. Originating from the example of arguing over a bike shed's color instead of a nuclear plant's design, it represents a wasteful focus on minor details.
It's not inconceivable that AI could become better than humans at targeting things. For example if it can reliably identify enemy warcraft or drones faster than people can react. I'm not saying Claude's models are suited for that but humans aren't perfect and in theory AI can be better than humans. It's not currently true and would need to be proved, but it doesn't seem unreasonable. It could well be better than something like deploying mines.
reply