More like the government is treating this like the near term weapon it actually is and, unlike the Manhattan project, the government seems to have little to no control.

Those aren't contradictory at all. If I need a particular type of bolt for my fighter jet but I can only get it from a dodgy Chinese company, then that bolt is a supply chain risk (because they could introduce deliberate defects or simply stop producing it) and also clearly important to national security. In fact, it's a supply chain risk because is important to national security.

> This contradictory messaging puts to rest any doubt that this is a strong arm by the governemnt to allow any use.

Why the hell should companies get to dictate on their own to the government how their product is used?

I've had so much abuse thrown at me on here for saying this very thing over the last few years. I used to be friends with Jack back in the day, before this AI stuff even all kicked off, once you know who people really are inside, it's easy to know how they will act when the going gets rough. I'm glad they are doing the right thing, but I'm not at all surprised, nor should anyone be. Personally I believe they would go to jail/shut down/whatever before they do something objectively wrong.

Anthropic had the largest IP settlement ($1.5 billion) for stolen material and Amodei repeatedly predicted mass unemployment within 6 months due to AI. Without being bothered about it at all.

It is a horrible and ruthless company and hearing a presumably rich ex-employee painting a rosy picture does not change anything.

I've thought the same about a few of my founders/executives.

"You either die the good guy or live long enough to become the bad guy"

The "bad guy" actually learns that their former good guy mentality was too simplistic.

How do you reconcile the fact that many people in Anthropic tried to hide the existence of secret non-disparagement agreements for quite some time?

It’s hard to take your comment at face value when there’s documented proof to the contrary. Maybe it could be forgiven as a blunder if revealed in the first few months and within the first handful of employees… but after 2 plus years and many dozens forced to sign that… it’s just not credible to believe it was all entirely positive motivations.

mark my words, they will burn at some point. The government can nationalize it at any moment if they desire.

    > I thought "Anthropic" was about being concerned about humans

[0] https://tvtropes.org/pmwiki/pmwiki.php/Main/PeoplesRepublicO...

I think it's phrased just fine. It's not up to Dario to try to make absolute statements about the future.

I think it goes without saying that ones the systems are reliable, fully-autonomous weapons will be unleashed on the battlefield. But they have to have safeguards to ensure that they don't turn on friendly forces and only kill the enemy. What Anthropic is saying, is that right now - they can't provide those assurances. When they can - I suspect those restrictions will be relaxed.

What else would you expect? The military is obviously going to develop the most powerful systems they can. Do you want a tech company to say “the military can never use our stuff for autonomous systems forever, the end”? What if Anthropic ends up developing the safest, most cost effective systems for that purpose?

Well, if they hadn't stated that were that far in line with the administration's ideals, they would likely already be fully blacklisted as enemies of the state. Whether they agree with what they're saying or not, they're walking on egg shells.

Unfortunately I think the writing is clearly on the wall. Fully autonomous weapons are coming soon

> the door is open for this after AI systems have gathered enough "training data"?

Sounds more like the door is open for this once reliability targets are met.

I don't think that's unreasonable. Hardware and regular software also have their own reliability limitations, not to mention the meatsacks behind the joystick.

I said exactly this a few days ago elsewhere. It’s disappointing that they (and often other American companies) seem to restrict their “respect” and morals to Americans only. Or maybe it’s just semantics or context because the topic at hand is about americans? I don’t know but it gives “my people are more important than your people”, exactly as you said in your last paragraph

They’re being used today by the military. So, they are never going to be against mass surveillance. They can scope that to be domestic mass surveillance though.

> And neither denounces partially autonomous mass surveillance nor closes the door on AI-driven foreign mass surveillance

You have to be deliberately naive in a world where five eyes exists to somehow believe that "foreign" mass surveillance won't be used domestically.

The Ghandi of the corporate world is yet to be found

Anthropic doesn't forbid DoW from using the models for foreign surveillance. It's not about harming others, it's about doing what is best for humanity in the long run, all things considered. I personally do not believe that foreign surveillance is automatically harmful and I'm fine with our military doing it

So AI systems are not reliable enough to power fully autonomous weapons but they are reliable enough to end all white-collar work in the next 12 months?

Odd.

The problem is that this is a decision that costs money. Relying on a system that makes money by doing bad things to do good things out of a sense of morality when a possible outcome is existential risk to the species is a 100% chance of failure on a long enough timeline. We need massive disincentives to bad behavior, but I think that cat is already out of its bag.

This isn't a one-election thing. It's going to be a generational effort to fix what these people are breaking more of every day. I hope I live to see it come to some kind of fruition - I recently turned 50.

Hope is not a plan, unfortunately, so if that's all we've got, I don't have much hope.

The current situation in the US is the depressing thing- articles like this give me hope. Real Americans aren't having these BS authoritarian violations of our constitutional rights.

You mean, what's been happening to the USA? this isn't a new trend. Militarization of police, open attacks on democracy, unilateral foreign policy moves.

the country jumped the shark post 9/11 and has been on a slow rot since then.

> Let's hope sanity prevails and the next election cycle can bring in some competent non-grievance based leadership.

Would be nice, but I have a bad feeling that the impact of widescale mostly unregulated AI adoption on our social fabric is going to make the social media era that gave rise to Trump, et al seem like the good ol' days in comparison.

I hope I am wrong.

> The technology can just be requisitioned

During a war with national mobilization, that would make sense. Or in a country like China. This kind of coercion is not an expected part of democratic rule.

It's also downstream of voters who voted in a president who promised to be dictatorial after failing at an attempted insurrection. We need to deprogram like 70M very confused people.

They already kissed the ring, just not the asshole. They have a little dignity left.

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.

While I agree the name change has not (yet) been made with the proper authority, I'm quite partial to the name and prefer to use it despite its prematurity. I think it does a better job of communicating the types of work actually done by the department and rightly gives people pause about their support of it. Though I'm sure that wasn't the administration's intention.

It SHOULD be called the Department of War, as it was originally, since it makes its function clear. We are a society that has euphemized everything and so we no longer understand anything.

And what if congress renames it tomorrow? They have the votes. These sort of procedural gotchas are as stupid as they are boring.

It's addressed to Hegseth, who insists on calling it that.

If they had called it DoD, then that would have been another finger in his eye.

But it sets the tone.

Less hypocritical than Defense. US has never been on the defense, always offense since it was renamed in 1947.

It is called the Department of War because we live under fascism and Congress no longer matters.

All that matters is that everyone calls it the Department of War, and regards it as such, which everyone does.

Agreed. You don’t have to be an LLM maximalist or a doomer to see the opportunity for real, practical danger from ubiquitous surveillance and autonomous weapons. It would have been extremely easy for Dario to demonstrate the same level of backbone as Sam Altman or Sundar Pichai.

There is no moral leg to stand on here, he says here in plain english that if they wanted to use CLAUDE to perform mass surveillance on Canada, Mexico, UK, Germany, that is perfectly fine.

We knew long before AI was a twinkle in Amodel's eye that if it were to be built, then it would be co-opted by thugs.

Anthropic's statement is little more than pageantry from the knowing and willing creators of a monster.

You know this is pure PR right?

A moral stand? ... What? Did we read the same statement? It opens right out the gate with:

>I believe deeply in the existential importance of using AI to defend the United States and other democracies, and to defeat our autocratic adversaries.

>Anthropic has therefore worked proactively to deploy our models to the Department of War and the intelligence community. We were the first frontier AI company to deploy our models in the US government’s classified networks, the first to deploy them at the National Laboratories, and the first to provide custom models for national security customers. Claude is extensively deployed across the Department of War and other national security agencies for mission-critical applications, such as intelligence analysis, modeling and simulation, operational planning, cyber operations, and more.

which I find frankly disgusting.

This is not how the word "moral" should be used in a sentence that also has the name Dario Amodei in it.

You, using normal Claude under the consumer ToS, cannot use it to make weapons, kill people, spy on adversaries, etc. The Pentagon, using War Claude, under their currently-existing contract, can use it to make weapons and spy on (foreign) adversaries, but not to (autonomously) kill people. I don't love this but I am even less excited about the CCP having WarKimi while we have no military AI.

those two stipulations were always their only ones, and they were included explicitly in their original contract with the DoW.

I think it's slightly less ridiculous than it sounds, because governments have much more power over their own citizens. As an American I would dramatically prefer the Chinese government to spy on me than the American government, because the Chinese government probably isn't going to do anything about whatever they find out.

(That logic breaks down somewhat in the case of explicitly negotiated surveillance sharing agreements.)

You’re getting many replies, and having scrolled through much of them I do not see one that actually answers your question truthfully.

The reason why there is an explicit call out for surveillance on American citizens is because there are unquestionable constitutional protections in place for American citizens on American soil.

There is a strong argument that can be made that using AI to mass surveil Americans within US territory is not only morally objectionable, but also illegal and unconstitutional.

There are laws on the books that allow for it right now, through workarounds grandfathered in from an earlier era when mass surveillance was just not possible, and these are what Dario is referencing in this blog post. These laws may be unconstitutional, and pushing this to be a legal fight, may result in the Department of War losing its ability to surveil entirely. They may not want to risk that.

I wish that our constitution provided such protections for all peoples. It does not. The pragmatic thing to do then is to focus on protecting the rights that are explicitly enumerated in the constitution, since that has the strongest legal basis.

This is a political statement directed at the US public, Congress, and executive branch in the context of a dispute with the US executive branch that is likely to escalate (if the executive is not otherwise dissuaded) into a legal battle, and it therefore focuses particularly on issues relevant in that context, including Constitutional, limits on the government as a whole, the executive branch, and the Department of Defense (for which Anthropic used the non-legal nickname coined by the executive branch instead of the legal name.) Domestic mass surveillance involves Constitutional limits on government power and statutory limits on executive power and DoD roles that foreign surveillance does not. That's why it is the focus.

In every country, citizens have more rights than non-citizens. The right to freely enter the country, the right to vote, the right to various social services, etc.

In the US, one of the rights citizens have is the right against "unreasonable searches and seizures", established in the Fourth Amendment. That has been interpreted by the Supreme Court to include mass surveillance and to apply to citizens and people geographically located within US borders.

That doesn't apply that to non-citizens outside the US, simply because the US Constitution doesn't require it to.

I'm not defending this, just explaining why it's different.

But, you can imagine, for example, why in wartime, you'd certainly want to engage in as much mass surveillance against an enemy country as possible. And even when you're not in wartime, countries spy on other countries to try to avoid unexpected attacks.

>Are there no democracies aside from the US?

If we're asking "What's the deal" questions, what's the deal with this question? Do only people in democracies deserve protections? If we believe foreign nationals deserve privacy, why should that only apply to people living in democracies?

The US has a strong history of trying to avoid building domestic surveillance and a national police. Largely it’s due to the 4th amendment and questions about constitutionality. Obviously that’s going questionably well but historically that’s why it’s a red line.

Exactly. FVEYs been doing reciprocal surveillance on each other for decades.

https://en.wikipedia.org/wiki/Five_Eyes#Domestic_espionage_s...

The reality is that the US Constitution only offers strong guarantees to citizens and (some of) the people in the US. Foreigners are excluded and foreign mass surveillance is or will happen.

I believe every country (or block) should carve an independent path when it comes to AI training, data retention and inference. That is makes most sense, will minimize conflicts and put people in control of their destiny.

Particularly so when those foreign nationals can be consumers. “fuck your basic human rights, but we can take your money just fine”.

Also the more the US openly treats the world like garbage, the more the rest of the world will likely reciprocate to US citizens.

It reminds me of some recent horror stories at border crossings - harassing people and requiring giving up all your data on your phone - sets a terrible precedent.

> what's the deal with making the distinction between domestic mass surveillance and foreign mass surveillance? Are there no democracies aside from the US?

I think it's just saying that spying on another country's citizens isn't fundamentally undemocratic (even if that other country happens to be a democracy) because they're not your citizens and therefore you don't govern them. Spying on your own citizens opens all sorts of nefarious avenues that spying on another country's citizens does not.

One of them is illegal for DoD to do and the other is not.

In the US, we have the ability to either confirm or change a significant chunk of our Federal government roughly every two years via the House of Representatives. The argument here is that we, theoretically, could collectively elect people that are hostile to domestic mass surveillance into the House of Representatives (and other places if able) and remove pro-surveillance incumbents from power on this two year cycle.

The reasons this hasn't happened yet are many and often vary by personal opinion. My top two are:

1) Lack of term limits across all Federal branches

and

2) A general lack of digital literacy across all Federal branches

I mean, if the people who are supposed to be regulating this stuff ask Mark Zuckerberg how to send an email, for example, then how the heck are they supposed to say no to the well dressed government contractor offering a magical black box computer solution to the fear of domestic terrorism (regardless of if its actually occurring or not)?

100% - this is the shortsightedness and demonstrates hypocrisy.

Countries routinely use other countries intelligence gathering apparatus to get around domestic surveillance laws.

The distinction between foreign and domestic is a legal one.

The Supreme Court has ruled that the US Constitution protects any persons physically present in the United States and its territories as well as any US citizens abroad.

So if you are a German national on US soil, you have, say, Fourth Amendment protections against unreasonable search and seizure. If you are a US citizen in Germany, you also have those rights. But a German citizen in Germany does not.

What this means in practice is that US 3-letter agencices have essentially been free to mass surveil people outside the United States. Historically these agencies have gotten around that by outsourcing their spying needs to 3 leter agencies in other countries (eg the NSA at one point might outsource spying on US citizens to GCHQ).

Are all democracies allies to you?

> what's the deal with making the distinction between domestic mass surveillance and foreign mass surveillance?

A large portion of Americans believe in "citizen rights", not "human rights". By that logic, non-Americans do not have a right to privacy.

I'm glad to see this as the top comment. I was, until recently, a loyal Anthropic customer. No more. Because the way non-Americans are spoken of by a company that serves an international market (and this isn't the first instance):

"Mass domestic surveillance. We support the use of AI for lawful foreign intelligence and counterintelligence missions. But using these systems for mass _domestic_ surveillance is incompatible with democratic values."

Second class citizens. Americans have rights, you don't. "Democratic values" applies only to the United States. We'll take your money and then spy on you and it's ok because we headquartered ourselves and our bank accounts in the United States.

Very questionable. American exceptionalism that tries to define "democracy" as the thing that happens within its own borders, seemingly only. Twice as tone-deaf after what we've seen from certain prominent US citizens over the last year. Subscription cancelled after I got a whiff of this a month ago.

(Not to mention the definition of "lawful foreign intelligence" has often, and especially now, been quite ethically questionable from the United States.)

EDIT: don't just downvote me. Explain why you think using their product for surveillance of non-Americans is ethical. Justify your position.

>democracies aside from the US.

I mean, I guess from '65 to around 96? We had a good run.

I can imagine that this will be the logical conclusion for many companies, I thought the same thing too, if it's too hard in the USA, they will just move.

Take your pick from the many other choices offered by companies that don't care about mass spying on _anyone_.

Is there a different AI company that IS taking that stance?

Because as far as I know, Anthropic is taking the most moral stance of any AI company.

I'm sure it's negotiations over how the enforcement will be done. My thoughts are:

1. Military wants a whole new model training system because the current models are designed to have these safeguards, and Anthropic can't afford that (would slow them down too much, the engineering talent to set up and maintain another pipeline would be a lot of work/time)

2. Military doesn't want to supply Anthropic usage data or personnel access to ensure its (lack of) use in those areas.

3. It's something almost completely unrelated to what's going on in the news.

Why? They clearly are very aligned on the objective, just doing some negotiation regarding the means. Giving up just because you don't agree 100% is not very constructive. This might seem bad for conflict-adverse people who usually are involved in low-stakes negotiations, but it's just the start of things for people who are fluent in conflict.

A favourable take would be he meant "mass surveillance of non-democratic adversarial countries". I agree it's not phrased this way though.

They both literally removed morality from their bylaws; that time has passed. They're openly corrupt because it pays to be so.

I think it is a reasonable moral stance to acknowledge such things are possible, yet not wanting to be a part of it. Regarding making it technically impossible to do...I think that is what Anthropic means when they say they want to develop guardrails.

A little pessimistic of a take, IMO. You may very well be right, though.

It’s not unusual for legal departments to take offense to these sorts of things, because now everyone using Claude within the DoD has to do some kind of audit to figure out if they’re building something that could be construed as surveillance or autonomous weapons (or, what controls are in place to prevent your gun from firing when Claude says, etc). A lot of paperwork.

My guess is they just don’t want to bother. I wonder why they specifically need Claude when their other vendors are willing to sign their terms, unless it specifically needs to run in AWS or something for their “classified networks” requirement.

Same reason they cut funding for universities that had DEI mandates, etc. and made a big spectacle of doing it despite it often being very little money etc. etc.

It's an ideological war, they're desperate to win it, and they're aiming to put a segment of US civil society into submission, and setting an example for everyone else.

He smelled weakness, and like any schoolyard bully personality, he couldn't help but turn it into a display of power.

He pushed the issue to an ultimatum because he is an unqualified drunk, and thinks that it's against the law for anyone to try and stop the US military from doing something they want to do. This isn't an isolated issue; he tried to get multiple US Senators prosecuted for making a PSA that servicemembers shouldn't follow illegal orders.

What makes you want to believe the Trump Administration when it claims it doesn't want to do domestic mass surveillance?

If you're using "seppo" as the Australian pejorative referring to Americans, I'm not sure what makes this uniquely American.

Yep.

That said, it does impact whether Anthropic can sell to the British [0], German [1], Japanese [2], and Indian [3] government.

Other governments will demand similar terms to the US. Either Anthropic accedes to their terms and gets export controlled by the US or Anthropic somehow uses public pressure to push back against being turned into an American sovereign model.

Realistically, I see no offramp other than the DPA - a similar silent showdown happened in the critical minerals space 6-7 years ago.

[0] - https://www.anthropic.com/news/mou-uk-government

[1] - https://job-boards.greenhouse.io/anthropic/jobs/5115692008

[2] - https://www.anthropic.com/news/opening-our-tokyo-office

[3] - https://www.anthropic.com/news/bengaluru-office-partnerships...

Just visibly wave the US flag and you'll be fine, don't worry.

Soon it will select targets in commie countries though, perhaps it already does. Who selected to bomb Chavez mausoleum btw?

What is their other possible move here, considering the government is threatening to destroy their business entirely?

Warfighters is a pretty common term though. There's a fair bit of nuance in when and how you'd use it.

I think the warfighters are a distraction, a system could trivially say that there is a human in the loop for LLM-derived kill lists. My money is that the mass domestic surveillance is the true sticking point, because it’s exactly what you would use a LLM for today.

Apparently part of this whole battle is because Grok isn't up to part to be an acceptable alternative.

As far as we can tell, OpenAI and Google seem to be ok with it and not resisting. It would be easier for Anthropic's cause if they did.

Yea but every warfighter will get a waifu

Grok in unhinged mode piloting an Apache, what could go wrong.

It's better than actively aiding them. Make them struggle at every turn.

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.

We're living in a time where most tech companies are donating millions of dollars to the current leadership in exchange for favors.

In that climate this is a more of a stand than what everyone else is doing.

Yes, but also remember where they came from.

They don't have any brand poison, unlike nearly everyone else competing with them. Some serious negative equity in tha group, be it GOOG, Grok , META, OpenAI, M$FT, deepseek, etc.

Claude was just being the little bot that could, and until now, flying under the radar