> The distinction between Maven and Claude is futile

Doesn't make any sense at all when you read the article and understand what Claude actually does in this equation. From the article:

> Neither Claude nor any other LLMs detects targets, processes radar, fuses sensor data or pairs weapons to targets. LLMs are late additions to Palantir’s ecosystem. In late 2024, years after the core system was operational, Palantir added an LLM layer – this is where Claude sits – that lets analysts search and summarise intelligence reports in plain English. But the language model was never what mattered about this system.

The whole point here is that whether an LLM is involved or not is immaterial to the system as a whole, and it's a disservice to the public to focus on LLMs here.

https://www.theguardian.com/technology/2026/mar/01/claude-an...

Edit: Also, https://www.washingtonpost.com/technology/2026/03/04/anthrop...

you, today, can use Claude in Amazon Bedrock, and the way that works is, if you want it to be this way: the piece of code and model weights and whatever other artifacts are involved, they are run on Bedrock. Bedrock is not a facade against Claude's token-based-billing RESTful API, where Anthropic runs its own stuff. In the strictest sense, Bedrock can be used as a facade over lower level Amazon services that obey non-engineering, real world concerns like geographic boundaries / physical boundaries, like which physical data center hardware is connected by what where / jurisdictional boundaries, whatever. It's multi-tenancy in the sense that Amazon has multiple customers, but it's not multi-tenancy in the sense that, because you want to pay for these requirements, Amazon has sorted out how to run the Claude model weights, as though it were an open-weights model you downloaded off Hugging Face, without giving you the weights, but letting you satisfy all these other IP and jurisdictional and non-technical requirements that you are willing to pay for, in a way that Anthropic has also agreed.

This is what the dispute with the Pentagon is about, and what people mean when they say Claude is used in government (it is used in Elsa for the FDA for example too). Anthropic doesn't have telemetry, like the prompts, in this agreement, so they have the contract that says what you can and cannot use the model for, but they cannot prove how you use the model, which of course they can if you used their RESTful API service. They can't "just" paraphrase your user data and train on it, like they do on the RESTful API service. There are reasons people want this arrangement ($$$).

The vendor (Palantir) can use, whatever model it wants right? It chose Claude via "Bedrock." I don't know if they use Claude via Bedrock. Ask them. But that's what they are essentially saying, that's what this is about. Palantir could use Qwen3 and run it on datacenter hardware. Do you understand? It matters, but it also doesn't matter.

It's a bunch of red herrings in my opinion, and this sort of stuff being a red herring is what the article is mostly about.

The key sentence in that Washington Post article appears to be:

> The Pentagon began to integrate Anthropic’s Claude chatbot into Maven in late 2024, according to public announcements.

As far as I can tell this is the public announcement - a press release from November 2024: https://www.businesswire.com/news/home/20241107699415/en/Ant...

> Anthropic and Palantir Technologies Inc. (NYSE: PLTR) today announced a partnership with Amazon Web Services (AWS) to provide U.S. intelligence and defense agencies access to the Claude 3 and 3.5 family of models on AWS. This partnership allows for an integrated suite of technology to operationalize the use of Claude within Palantir’s AI Platform (AIP) while leveraging the security, agility, flexibility, and sustainability benefits provided by AWS.

https://www.reuters.com/business/aerospace-defense/us-uses-h...

Unfortunately I can very well imagine several more months and years of this. We are still fighting a forever war that started in 2001. This is all a generation of Americans will know, and that is sad.

> 11,294 munitions in the first 16 days of the conflict, at a cost of approximately $26 billion.

Several detailed tables are in the link below.

https://www.rusi.org/explore-our-research/publications/comme...

I feel like we know enough already. A school was bombed, the ones who did it sucks big time and should be held responsible. Currently, the US and Israel is waging a war against Iran, and one of them dropped the bomb(s), unless suddenly Iran got their hands on American weapons, then that needs to be investigated too, because someone surely dropped the ball at that point.

The basics remain the same, investigations have to be launched to figure out where exactly in the chain of command, someone made a mistake, and then hold that person(s) responsible for their fuck up.

Have those investigations been launched?

This is not to say that this administration is definitely not targeting civilians or infrastructure on purpose; just that the end result, and the moral culpability, are the same in either case.

I don't disagree there. But this is not a case of hallucination, and an existing website is a signal, not a determinant, of the real situation on the ground. However, you have made a very, very strong assumption that these targets were not carefully evaluated. One that does not seem to be present in TFA or any analysis that I've read. In fact, the article itself quotes those in the know who believe this should have been eliminated as a target.

> Humans should have been double and triple checking every target by other means.

How practically would this happen? The US/Israel don't want people on the ground, and people on the ground is exactly the only way you can actually verify stuff like this, not every place in the world is on Google Maps or have a web presence at all, so the only realistic way to verify this would be to visually inspect it in person, something neither parties who started this war want to do.

Even better, don't make attacks against other soverign nations that don't pose an immediately and critical threat to you, and this whole conflict could have been avoided in the first place.

But no, the president has to be involved in some sort of child-trafficking scheme, so pulling the country into a war seemed preferable to being held responsible, and now we're here, arguing about fucking details that don't matter.

This certainly doesn't absolve the person implementing those parameters, but it is equally the responsibility of the very top of the decision-making structure.

The main way targets should/would be selected is by direct intelligence. E.g. the targets should be identified through satellite or other observations. It's hard to imagine that a building that has operated for some length of time as a school would not have patterns that are visible from satellite vs. military facilities with different patterns. You also don't just randomly attack structures in this sort of surprise attack, you're presumably aiming for some specific people or equipment with some priority/military goal in mind, so you really want to have observed the targets and patterns and have up to date information on their usage.

I think what likely happened here is that the entire base was the "unit" of targeting and the mistake was in identifying which buildings were part of the base. In the satellite view the military buildings and the school look very similar (since the building as I understand it used to be part of the base but was repurposed as a school).

It's not true that whoever made the error had all the time in the world. Presumably once the order was given there was time pressure given that the strike was to be timed with the other intelligence.

In theory the US military should/is supposed to have good processes around this stuff. So we are told. Obviously failed in this case. It is a tragedy.

By your logic it's the federal government's fault those 3000 people died on 9/11, they were being used as human shields.