U.S. appeals court upholds designation of Anthropic as supply chain risk

I know everyone says this is political but it actually seems like a textbook designation. Anthropic wanted to have rules on how the military used AI, the military said no and therefore doesn't want anthropic used anywhere in their supply line.

This is like a pen manufacturer not wanting their pens used to sign drone strike orders, now the military needs to have a special box of pens that don't have stipulations attached. With AI usage it would be the same thing except applied to entire product chains. It seems like it would just add more complexity to operations.

You can agree with the rules anthropic wanted, but having rules set by a private company at all that apply to the military does seem fair for the military to object to.

>The Department reasonably feared that Anthropic might manipulate Claude’s design to prevent it from performing national-security functions that the Department deems contractually authorized and necessary

Though they'd probably put the DoD on the cybersecurity whitelist today, the very idea of the claude whitelists for certain functionality already exists and is being used by them today.

I am baffled the DoD/W needs to rely on a third (commercial party) for something strategic as this if they deem SI strategic. Pistols and rifles aren't the same thing.

Hire people and pay them well and develop whatever yourself without this charade, they have unlimited budgets.

I am not taking a moral ground, but I can't see how "an Anthropic" fits in. You might say Boing or Lockheed, Sig Sauer and Colt and FN do, but it feels somehow different as they've "bent the knee" for decades.

Not sure if I’m just in my own national security bubble, but I find it troubling to see the perspective that most of the conversation in this thread is coming from.

The US government took a legal designation explicitly crafted to protect against foreign adversaries and deployed it against a private, domestic entity, to their immediate and great detriment.

I wonder if this is going start being abused soon.

If I was Palantir or any other GOP aligned company, I would be completely against this. What's to stop a Democratic president from doing the same thing and destroying them?

This smells like corruption to me.

OpenAI has hacked several prominent entities and is allowed to do business as usual but Anthropic putting guardrails on the military's usage of AI is the national security threat while Pentagon itself said that over reliance on AI lead to that attack on school in Minab.

I have also learned that OpenAI's president is a big maga donor and now I am thinking of cancelling my OpenAI membership todau and signing up with Anthropic again.

Edit - i am not a OAI hater and i don't hate anyone. I am just saying Anthropic is kinda getting bullied and OAI is not being punished enough for their actions. It is time for me to support Anthropic (with my $$$) instead of OAI.

I still don't fully understand what this was about. It sounds like the DoD said they wanted unrestricted access to Anthropic's models, Anthropic refused, so the Pentagon declared that they wouldn't use them at all.

Isn't this basically what Anthropic wanted?

Admittedly I still need to do some reading on the current ruling, but how this isn't seen as an open and shut case of vindictive prosecution (if that's even a thing for this branch of law?) is astounding. The administration couldn't have made it more obvious there was no merit to their designation of Anthropic.

Supply chain risk seems a massive over-reaction, but understandable that the government doesn't want to commit to Anthropic controlled software for any critical military application if they can't 100% rely on Anthropic support in being able to use it as they see fit.

I think both sides are wrong. In no particular order:

US does not buy weapons that self-regulate generally. The US can and has purchased weapons with stipulations on how they should be used. For instance, and I'm making this up, they might purchase cluster munitions for smashing up an airfields, but with a stipulation they may not be used where humans are present or something. The munition doesn't check GPS and refuse to detonate if GPS doesn't indicate it's over some Russian airfield. Nor does an M4 fail to fire if it's on US Soil. Anthropic tried to build safeguards and out-regulate the government. Everyone here should know that such limitations would (can and have, see some famous IFF spoofing incidents) be exploited by an enemy, which is generally why they aren't acceptable in a military context. Without seeing the contract signed, one possibility is Anthropic tried to push regulations onto the government, and Anthropic gets to ride away on a high horse.

The other side: The Supply Chain Risk feels like an abuse. It's likely a valid designation and likely has its historical uses. If this was a contract issue, that would have been the correct way to litigate. Instead, again without seeing the contract, it feels like this is way to try and stifle their uptake in defense sector; basically attempting to strong arm them by choking a business segment off. Again, if there is a in fact a legitimate breach of contract here, it should be litigated as such. The alternative is there is no such breach, and Anthropic built limitations into the contract and the Pentagon just has buyers remorse. Honestly this wouldn't surprise me, the federal government generally incinerates tax dollars, and with the massive marketing hullabaloo pushed by AI companies, its completely likely a giant contract was signed without reading the fine print.

Honestly both sides of this annoy me.

I guess it’s the supreme court next.

For all the downsides of losing the business of government contractors, it really would hold Anthropic to their stated values: they’ll never develop AI controlled weapons on this blacklist.

The hidden upside for Anthropic is now they won't be forced to refer to AI as "SI".

Opinion by Katsas and Rao. For those that don’t follow this regularly, those are two of the biggest Trump hacks on the bench.

I suspect this will be reversed en banc.

Anthropic’s argument is really clear and easy to understand: we can’t simultaneously be a supply chain risk and also be a company the DoD demands we change our policies so they can use our models without restrictions.

All AI is a supply chain risk. I think people should know that about now. You can't keep it in a box. The way code works as a model is that is not super sentient thing with nefarious intentions. It is just pure automaton.

Let me explain, if it learns something from training data to do certain things from code it has online it can install or use a code that is vulnerable from a security perspective.

It seems very likely that AI will ultimately be under government control. To me, this is the nightmare we are all afraid of. And Dario's essay subtlety confirms this.

Good.

I think the real, and continuing, story here is the US military and the US govt in general continuing to let ego degrade capability. I'm not going to get into a discussion about what is legal, I am personally not interested in that distinction here, the core point is that this is the US being stupid, again.

I don’t agree with the decision, but I don’t think courts should be able to easily overturn the federal government’s findings when it comes to national security in a specific decision like this.

Is this something to do with anthropic model being used to kill 150 children in school bomb via palantir?

Wtf is life just court battles back and forth for everything forever now?

I'll tell you what's a supply chain risk...

The current fucking US administration.

This is more about wringing Anthropic's hand to agree for handful of defense companies to be the official AI translators to government.

for the next 800 days

but really, they all are, look what Grok did to hundreds of Iranian children

* https://thehill.com/policy/technology/5928204-pentagon-musk-...

Absolute clown show in the US right now

> Judge Gregory Katsas wrote in the majority opinion for the U.S. Court of Appeals for the District of Columbia, which Judge Neomi Rao joined. Katsas and Rao were appointed by President Donald Trump.

> Before his appointment to the bench, Katsas served as Deputy White House Counsel in the first Trump administration

> [Rao] was appointed in 2019 by President Donald Trump, having served in the Trump administration from 2017 to 2019 as administrator of the Office of Information and Regulatory Affairs

The autocracy is in full swing.

Genuine question: does anyone know what concrete obligations this designation actually imposes? Everyone's arguing about the politics but I haven't seen a clear explanation of what Anthropic is now legally required to do or not do.

[dead]

[dead]

Posted this in the wrong thread.

Anthropic isn't stopping me from getting Chinese parts, Yukon Jack and good Cadbury chocolate.

trump is. Designate him!

If the CEO of any other defense contractor did this, they'd be summarily fired by the board.

I agree with its designation. The security scans from Claude Mythos have been highly suspicious, wrong, and with a significant false positive rate. Claude also directly causes conflict within teams through excessive utility of "git blame" which weaponizes everyone against each other until only Claude has control. I've had very bad experiences with Claude and only Claude.