OpenAI reaches deal to deploy AI models on U.S. DoW classified network

[techblueberry]

"AI safety and wide distribution of benefits are the core of our mission. Two of our most important safety principles are prohibitions on domestic mass surveillance and human responsibility for the use of force, including for autonomous weapon systems. The DoW agrees with these principles, reflects them in law and policy, and we put them into our agreement."

So wait a second, the DoW let OpeAI put the exact same clauses in their contract, but Anthropic is a supply chain risk? What is going on?

[portacaval]

See your answer below, from the BBC:

An Anthropic spokeswoman added on Thursday that while the company received updated wording from the DoD for its contract on Wednesday night, it represented "virtually no progress on preventing Claude's use for mass surveillance of Americans or in fully autonomous weapons."

"New language framed as compromise was paired with legalese that would allow those safeguards to be disregarded at will," she said. "Despite [the Department of War's] recent public statements, these narrow safeguards have been the crux of our negotiations for months."

https://bbc.com/news/articles/cvg3vlzzkqeo

[neya]

I get the feeling they didn't like Anthropic from the start for whatever reason and just found excuses to dump them and find an alternative provider. There's also the possibility the DoW will still have its way and this is possibly only a statement for the public.

P.S I'm not political (no left/right drama), just speculative.

[unyttigfjelltol]

This says OpenAI is happy with the law and DoW policy, and thinks those protections are adequate. In contrast, Anthropic seems to have imposed additional contractual guardrails, blocking apparently legal DoW use cases.

Was it a subtly different contract, or a notice of dispute? Unclear.

[yoyohello13]

The difference is how much money each CEO has contributed to Trump’s campaign.

[m3kw9]

Sam went second so he can obviously use the entire situation to warn the govt about designating all AI companies supply chain risk

[d--b]

What do you think could be going, seriously?

Obviously Altman agreed to whatever they asked him and is lying that the DoW is cool about the red lines.

[SilverElfin]

Claude is not as ideologically willing as friends of the Trump administration are - like xAI's Grok and OpenAI's ChatGPT. And David Sacks, the AI czar, who is a PayPal mafia member and close friend of the Epstein class (Elon Musk, Peter Thiel), has been attacking Anthropic FOREVER. He views them as a woke threat to America. So all that supply chain risk and the ban from agencies is just a way of abusing power to suppress free speech.

But the most obvious reason, is that Greg Brockman - who's a founder and also the current president of OpenAI apparently, is the largest donor to the Trump MAGA PAC. I find it hard to believe that Sam Altman and the Trump administration would be so obvious about it, but I also find it hard to believe that tens of millions in donations and this kind of result can be anything but a bribe.

[almosthere]

Because there are only two options really, anthropic and openai. Gemini is a year behind. Gov had no more options after those two.

[ismyrnow]

I suspect that Anthropic wanted the govt to ensure they would not use Claude for mass surveillance or autonomous weapons even if it does become legal in the future. OpenAI was comfortable with the agreement because it's currently illegal and they can make a public statement that they still side with Anthropic on the ethics side.

I also believe that Anthropic's CEO just didn't get along with the DoW in negotiations. So much of these big contracts comes down to relationships.

Still pretty sleazy how the govt is being so aggressive about it. I wish they would have just denied the contract and picked another vendor.

[stopbulying]

OpenAI was founded as an ethical AI research nonprofit organization.

[NBJack]

I'm not sure anyone at the company can say that with a straight face anymore.

[tombert]

I agree, though aren't they planning on doing an IPO later this year?

[stopbulying]

Is Trump still goading OpenAI into transferring SOTA AI to middle eastern countries that do not regard Freedom of Speech and Freedom of Religion as values and rights?

[kristopolous]

Just canceled my subscription and added their domain as 127.0.0.1 to my /etc/hosts.

We too have a voice here.

Do Not voluntarily fund the robot army.

[kivle]

I just cancelled as well. Time to test out the competition that have some morals left.

[chkaloon]

They know Sam has few scruples and will go for the money if the DoW says jump. These clauses mean zilch if Sam is involved

[ChrisArchitect]

Source: https://x.com/sama/status/2027578652477821175 (https://news.ycombinator.com/item?id=47189650)

[xvector]

This feels kind of like a masterclass in negotiation. I suspect Dario just didn't get along with the DoW. Sam comes in, takes Anthropic's place, gets the same concessions.

Hell, this may open a path for Anthropic to survive. Everyone can save face if the Pentagon says "we got Anthropic to agree to the same terms as OpenAI" and Anthropic gets their guardrails anyways.

On the other hand it's equally likely that the government just destroys Anthropic regardless because they don't like it anymore (and it dared to disagree in public.)

[othersidejann]

There's a rumor floating around that says basically, Amodei said "call us and we'll see" to a clear cut missile defense scenario. Perhaps the context of this is some part of the negotiations that we're not privy to. And might be where OAI and Anthropic differ, without any of them admitting it publicly.

[allovertheworld]

skynet speedrun

[mnky9800n]

Something tells me that this was all predetermined prior to hegseth throwing a fit. That was just bullshit theatre to justify government handouts I’m to OpenAI. I dunno. I don’t make the decisions.

[petterroea]

That was fast

[chiararvtk]

human responsibility for the use of force

So, use but if goes wrong, someone needs to be responsible. Aaaand we know that works very well

[rasz]

https://www.nbcnews.com/news/military/us-military-used-laser...

AI told me it was a Mexican drone, not a regular Delta El Paso flight

[m3kw9]

You guys think American shouldn’t use AI in military at all isn’t understanding their counter party is absolutely all in on AI in theirs

[kristopolous]

That's the logic of monsters:

Step 1: We believe the enemy is a monster who does [terrible act].

Step 2: To counteract this, we must do [terrible act].

Result: we maintain we are the "good guys" because we were "forced" into it by their presumed behavior.

He who fights with imagined monsters should look to it that he himself does not become a monster. And if you gaze long into an abyss, the abyss also gazes into you.

That quote is 140 years old. Is that enough time to heed it?

[jkaplowitz]

Nobody was advocating for zero AI in the military - certainly not Anthropic. They were fine with all lawful US military use cases except for two: the mass domestic surveillance of Americans and fully autonomous weapons. Whether you agree or disagree with their particular red lines, that's quite far from them trying to keep their product out of the military.

[djdndbdnfd]

Nobody is complaining about the government not giving Anthropic a contract. It’s about the unprecedented and outrageous threats to destroy their business if they don’t provide the government with what they demand. There is no national supply risk from Boeing using Claud Code just beacause Anthropic won’t agree to domestically-surviving killbots. The government’s behaviour is overwhelmingly malevolent and terrifying.

[behringer]

We don't want to support Ai under Trump. We don't even want a department of war.

[rustystump]

All the ai gov drama is a giant shrug to me. The reality is uncle sam has all the power and whatever comes out of industry sam is gonna sam with just like any other gov.

Ethics dont exist on the global stage. The privacy ship sailed decades ago. It all looks performative to me across all sides.

The smart move is to side with gov using some hand wave jutsi about “we have a contract” to get access to sams big nuts. Sam aint gonna respect no contract regardless of what is in. He breaks all laws without consequence for the last 60 odd years.

Unless people start starving, no revolution is gonna change the status quo here. It is like 9D marketing/branding chess for ai companies

[SpicyLemonZest]

I'd really like you to understand that saying this kind of thing means you're taking the government's side, even if you do it with cool ironic undertones. I'm religious enough that I'm particularly sensitive to this dynamic - there's a lot of pastors out there who talk about how we live in a fallen world as an excuse for why we don't need to bother opposing some particular injustice.