9,596 karma · joined July 28, 2014
Ceasing to contract with them directly doesn’t change the fact that Anthropic wishes to leverage itself to influence the government. That doesn’t go away. The problem is not with closing all direct contracts between the Pentagon and Anthropic, those don’t matter, it’s with closing all their channels of influence into DoD as a subcontractor.
Similarly to how DoD refusing to buy from Huawei doesn’t protect DoD from their prime contractors buying Huawei gear, they need a supply chain risk designation to ensure they are protected.
Anthropic is uniquely interested in introducing itself as an external enforcer of US law, a sort of belt-and-suspenders approach, where the Department is not only subject to operate under the constitution and the laws from the legislative branch, but also subject to anthropics interpretation of whether they are operating under the constitution and the laws from the legislative branch.
The department of defense does not want to engage in massive domestic surveillance beyond what the law allows them to do. They have signed agreements with OpenAI and other vendors which reiterate that they do not wish to use AI systems for massive domestic surveillance. These terms were unsatisfactory for Anthropic, for whatever reason.
The problem is not the terms of the agreement. It’s the people and the way they conduct business. It’s the fact that they’ve expressed a willingness to hold their product (or future products) hostage, at the cost of DoD operational excellence. It’s the fact that they’re training a specific model variant for government usage with extra guardrails and limitations and values.
Above all else, it’s the fact that they want to leverage their position as a leading AI company to influence government policy. This is not how a serious reliable partner of the government behaves. The problem from the DoDs perspective is the company itself and the people in charge of it.
On the substance: nothing in 3252 limits ‘adversary’ to foreign actors. Congress used ‘foreign adversary’ in other statutes when it meant foreign adversary. It didn’t here. That’s a problem for you. The government’s brief cites three dictionaries defining adversary as ‘an opponent in a contest, conflict, or dispute.’ A vendor that questions active military operations through intermediaries and demands an approval role in the operational decision chain is an opponent in a dispute. That’s the plain text. Originalist judges will see it that way.
I don’t really follow what you’re saying in point 1, the supply chain risk rationale is in the confidential record of this court case. There’s no way for us to know what’s in there, but it’s safe to assume the government covered their bases.
On point 2, I also don’t understand what you’re saying. They are in court right now. How have they been denied due process?
Point 3 is less interesting to me. Twitter posts by Hegseth obviously don’t really hold water. Anthropic should win here. But that’s not really what this case is about or why it’s interesting.
Your point 4 assumes the government acted outside the law. I’m not convinced of that. That’s the very question being litigated. The government’s position is that it acted within 3252. One San Francisco district judge disagreed at the preliminary injunction stage. That’s not a final answer. Not even close.
What remedy are they seeking? How can this be redressed? (Hint: they want to be a part of the DoD supply chain. If they weren’t, they wouldn’t have standing. If the court can’t do anything for you even if you win, you fail the redressability prong and get bounced for lack of standing.)
Because that’s what the law is! Because 1) 3252 gives them a mechanism to exclude a certain vendor from their supply chain broadly, and 2) singling-out a specific vendor for any other reason (favoritism, corruption, etc.) is not legally permissible under any other law.
You can argue that the law doesn’t make sense, but you can’t argue that the law is not the law?
> Why are you acting like no one else here understands what has happened?
Because you clearly don’t? Because nobody who has a remote understanding of the legal system would be stupid enough to suggest that a San Francisco district court judge preliminary injunction decision would carry enough weight to dictate DoD procurement during an active hot war.
This injunction doesn’t take effect for a week, precisely so that the Department has time to appeal this to the 9th circuit. And even if the 9th circuit doesn’t stay it, SCOTUS will. This court has stayed district court injunctions against the executive on national security grounds multiple times. They are not going to let a single district judge in San Francisco dictate military procurement during an active war. Obviously. OBVIOUSLY.
Lin didn’t drop Palantir from the defense supply chain unilaterally. The world does not work that way. Obviously. She issued a preliminary injunction that will be appealed before it takes effect. The DoD has not “shot their shot.” This lawsuit hasn’t even started yet.
Covered procurement actions are the things the Secretary can do after making a supply chain risk designation under 3252. The designation is a prerequisite. You can’t direct a contractor to exclude a subcontractor under (d)(2)(C) without first going through the 3252 determination process.
You’re literally posting evidence for why this is the only legal avenue for DoD. Yes, I’ve read everything on courtlistener. I trust you have as well, but did you understand any of it?!
This site keeps getting dumber and dumber.
If they are comfortable without being in that supply chain, whoever’s supply chain that is (exercise to the reader), why are they suing?
The DoD is not trying to sneak its way out of behaving legally. On the contrary, they’re doing it the legal way and you’re suggesting that they could just do it the illegal way.
That’s the entire reason this law exists, because what you’re suggesting is impractical. The department has to confidentially document its rationale for marking a company as a supply chain risk. It’s in the confidential record of this very court case. That’s the legal way to do this.
The pentagon can terminate its direct contract with Anthropic but it does nothing to address the risk that Anthropic poses to the reliability of Palantir’s services, which are (at this point) critical to the way that the Department operates.
People keep repeating this lie and I’m sick of it. The direct usage of Claude by the pentagon is not what they’re trying to address, it’s the usage of Claude by Palantir that they’re trying to address. And this is the legal way for them to do that.
Again, for the third time in this thread, they MAY NOT ask Palantir off the record to just not use Anthropic. This would be extremely illegal and would give Anthropic standing to sue to the government.
What other legal mechanism do they have to prevent Palantir from specifically not subcontracting out to Anthropic, other than a supply chain risk designation? Note that directly asking Palantir to prefer Google or OpenAI over Anthropic is a violation of procurement law and highly illegal.
What other mechanism do they have?
If that’s true, then what you’re suggesting is absurd. Because it’s not enough for the pentagon to merely stop contracting with Claude, because that was never the problem in the first place from their risk model. Their problem was they had a prime contract with Palantir for their wargaming service, and Palantir subcontracted with Anthropic as an LLM provider. So if DoD ceased to contract with Anthropic directly, it would have no impact on the risk that Anthropics new posttraining limits potentially posed to their mission insofar as they are reliant on Palantir and it’s services and there would be nothing preventing Palantir from continuing to contract with Anthropic.
I have to ask, what other tool do you think they have to protect themselves from this? You can argue that these guardrails from Anthropic are useful and important and DoD should just accept that, and that’s fine, but it really is (and ought to be) the departments decision about whether they’re comfortable with that or not. It’s their call. They have access to information on our adversaries that the public doesn’t. And they’re the ones responsible when lives are lost. And if they’re not comfortable with trusting service member lives to a specific post trained Opus 4.6 model, I’m not sure what other avenue they have to solve that problem across their entire prime contracting space other than a supply chain risk designation.
Any sort of backroom dealings where they whisper off the record to defense CTOs that they have a problem with anthropics leadership and would prefer that they sub out to OpenAI or Gemini instead for LLM services would be totally illegal and a violation of procurement law. So they definitely can’t do that. A supply chain risk designation is the only real tool they have to single out a single company.
One thing worth noting: Anthropic is a PBC, which is a new corporate structure that makes it relatively unaccountable to traditional profit motives. But those traditional profit motives are precisely the carrot that the DoD relies on dangling in front of the industry to motivate companies toward its mission. Traditional for profit companies are lead by people who have a fiduciary responsibility to maximize profit by serving the government. The entire procurement process relies on companies being motivated by profit and competing through bids. But PBCs are specifically designed to remove that incentive structure from their decision making, which makes them entirely unalike every other defense contractor which is publicly traded and can be held legally responsible by shareholders for putting personal beliefs above increasing shareholder value. That sounds like… exactly the kind of thing you don’t want in your military supply chain.
It’s not really unexpected.
In practice, if I buy $200/mo codex, can I basically run 3 codex instances simultaneously in tmux, like I can with claude code pro max, all day every day, without hitting limits?
This is a massive cope imo. The reason that the AI industry is so incestuous is just because there are only a handful of frontier labs with the compute/capital to run large training clusters.
Most of the improvements that we’ve seen in the past 3 years are due to significantly better hardware and software, just boring and straightforward engineering work, not brilliant model architecture improvements. We are running transformers from 2017. The brilliant researchers at the frontier labs have not produced a successor architecture in nearly a decade of trying. That’s not what winning on research looks like.
Have there been some step-change improvements? Sure. But by far the biggest improvement can be attributed to training bigger models on more badass hardware, and hardware availability to serve it cheaply. To act like the DoD isn’t going to be able to stand up pytorch or vllm and get a decent result is hilarious: the reason you use slurm and MPI and openshmem is because national labs and DoD were using it first. NCCL is just gpu accelerated scope-reduced MPI. nvshmem is just gpu accelerated scope-reduced openshmem.
If anything, DoD doesn’t have the inference throughput requirements that the unicorns have and might just be able to immediately outperform them by training a massive dense model without optimizing for time to first token or throughput. They don’t have to worry about if the $/1M tokens makes it economically feasible to serve, which is a primary consideration of the unicorns today when they’re choosing their parameter counts. They can just rate limit the endpoint and share it, with a 2 hour queue time.
The government invented HPC, it’s their world and you’re just playing in it.
> Generally, the defense crowd have a somewhat inflated sense of self worth.
/eyeroll but nobody can do what you do!
Let’s put pressure on our government to fix the FISA issues. Let’s reign in the executive branch. But let’s do it through voting. Let’s not give up on our system of government because we have new shiny technology.
You were naive if you thought developing new technologies was the solution to our government problems. You’re wrong to support anyone leveraging their control over new technology as a potential solution or weapon of the weak against those governmental issues.
That is not how you effect change in a democracy.
This is obviously subjective, and the only subject that matters in this case is the leadership at the DoD.
> We have no choice but to pay taxes and make the federal government 20 percent of our economy. There is no single company or any other entity that is close. And extending it to everyone who has a government contract probably makes it the majority of the economy.
I, too, hate big government and the all-powerful executive branch. Welcome to my tent. Let’s invent a time machine together so we can elect Ron Paul in 2008 and nip this in the bud.
Until then, this is what we’re stuck with.
If Anthropic can survive on open source contributors shelling out $200/mo and private sector companies doing the same, the government wishes them well. But surely you agree the government has a right to determine how its budget is appropriated?
The DoD has a right to avoid such models, and to demand that their subcontractors do as well.
It’s like saying “well I’d hope Boeing would test the airplane before flying it” in response to learning that Boeing’s engineering team intentionally weakened the wing spar because they think planes shouldn’t fly too fast. Yeah, testing might catch the specific failure mode. But the fact that your vendor is deliberately working against your requirements is a supply chain problem regardless of how good your test coverage is.
The existing contract is only a few dozen months old. It didn’t hold up to scrutiny under real world usage of the service. The government wants to change the contract. This is not the kill shot you think it is. It’s totally normal for agreements to evolve. The government is saying it needs to evolve. This is all happening rapidly and it’s irrelevant that the government agreed to similar terms with OpenAI as well. That agreement will also need to evolve. But this alone doesn’t give Anthropic any material legal challenge. The courts understand bureaucracy moves slowly better than anyone else, and won’t read this apparent inconsistency the same way you are.
That’s what they will argue, anyway.