Welcome to the podcast.
I'm your host, Jaden Schaefer.
Today on the show we have some latest advancements in the whole story between the showdown of Anthropic and the Department of War, the Pentagon.
They've essentially been, they're moving to designate them as a supply chain risk.
And it looks like Anthropic may have been played by Sam Altman and OpenAI, who have swooped in and taken the contract that the Department of War has just canceled.
So anyways, there's a whole bunch of drama in there.
We're going to cover all of that on the podcast.
Before we do.
If you want to try the latest models from Anthropic or from ChatGPT or Gemini, or even audio models like 11 Labs, I'd love for you to try out my startup, which is AIboxai.
You get access to over 40 of the top audio, image, text models.
And we've just completed an entire overhaul and redesign of the platform to make this streamlined, simpler and more efficient for you, as well as added an entire capability for you to describe a tool or workflow you'd like to build and have our AI tool builder automatically create it for you.
So if you want to go try that out, it's AIbox.ai.
It's $8.99 a month and we even have a 20% discount if you get an annual fee, so subscription.
So you can go check it out at AIbox.ai.
All right, let's get into the story.
So Over the last two weeks there's been this really high stakes confrontation going on between Anthropic and the Department of Defense or the Department of War.
I can't remember what we're calling it nowadays.
This is Pete Hegseth, and it's kind of under Donald Trump.
And basically the center of this whole argument is a question I think that is going to be really important for how AI is used inside of government going forward in the future, and that is kind of who is, you know, who's in control of these AI systems that are powering the most powerful national defense systems?
So Anthropic CEO Dario Amadeo, he said like he basically made this big statement where he's basically saying he doesn't want his AI models to be used for two specific things mass domestic surveillance of Americans and also fully autonomous weapons that select and engage targets without human involvement.
So those are kind of his two red lines.
And between, and you know, between those two things.
He said that look, we can't keep supporting the military on these different use cases that they might be pursuing.
And so he kind of put safety, safety guards and guardrails into what Anthropic is capable of doing, so that the government can't do that now.
The Pentagon's position is also, I mean, they're very direct on what they want.
So the Secretary Hegseth basically, is arguing that the Department of Defense shouldn't be constrained on their use cases by the internal policies of an AI company.
Now, on the one hand, I agree with Anthropic in a sense that I don't want the government doing mass surveillance of Americans with AI systems.
And I also agree that fully autonomous AI that goes and executes kill shots or whatever without a human intervention is a very crazy kind of ethical boundary that I don't think we want to get into.
So I don't really like either of those two use cases.
But on the other hand, I do see the argument that you know, if we have these, you know these AI vendors that are kind of making their own rules, and perhaps those two are good red lines right.
Well, what happens in the future when Anthropic says actually we don't want these to be used for any of these other?
You know military use cases, like you know war planning or strategies or anything that could contribute to someone dying in the future.
Like you could see this essentially being.
If their policy shifts and all of a sudden, like the US, Department of Defense is, you know, integrating this AI system into all of their different systems for the military, and all of a sudden they changed, you know, the internal policies of their company, then all of a sudden the capabilities of the military get nerfed.
It doesn't seem like a very safe place for the government to be out.
And I think there should be a broad, like overall, kind of conversation from Congress and from probably, voters on what we want AI to be able to do and what we want the military to be able to do and how we go about a lot of these things.
But at the same time I don't really like, you know, although I don't like those two use cases that Anthropic has kind of redlined, I also don't really like the fact that Anthropic can redline use cases for the military.
And right now those seem like good ones.
But in the I mean like you could, even theoretically in a conspiracy theory world, which I also love let's say, China decides to take a huge stake in anthropic.
I'm sure the US government would never let that happen.
Blah blah, blah.
But like, let's just say, via some, you know, maybe they get some investors in Saudi Arabia or whatever, right.
Let's just say there's a way that they could get some sort of control into anthropic and then they could make some sort of policies that directly you know, negatively impact the government.
I just don't like, I just don't like the rules coming from the companies themselves, which are you know, we know that those are sort of manipulatable.
You can buy up board seats and whatever else.
So I just don't like that concept.
Okay.
And even though I think that probably their causes are just or good, perhaps.
Okay.
So this whole thing escalates quite a lot, to the point where President Trump directed federal agencies to stop using Anthropic's products.
He said, you know, basically there's kind of like this six month transition period.
And then right after that, Secretary Hegseth designated Anthropic as a supply chain risk to national security, which is basically blacklisting them from doing business with the military and even potentially with contractors that are tied to the military.
Anthropic said that they hadn't been or they hadn't received any sort of formal notice.
And they were.
You know they were going to challenge this kind of designation in court.
And so there's, you know, obviously all of this drama going on.
And I think it's especially important considering we know that while all of this was happening, the US government was leading up to their bombing of Iran with forces in Israel.
And evidently they didn't want, you know, kind of mid-operation there.
AI systems to be cut or them to have some sort of issue there.
So it seems like this was probably a role you know.
It's like they have to have this big public beef with an AI company doing work with the military before they can, you know, launch their next military strike, which is interesting considering we know that Anthropic was used heavily for the capture of Nicolas Maduro for that mission.
And so it seemed like, you know, before they wanted to launch their full their full attack on Iran.
They also wanted to make sure that they had AI models to back up their operations, which is really crazy if you think about it.
I think there was a lot of implications that were really big in everything that was happening.
Anthropic had received Department of Defense contracts.
And we also had, by the way, OpenAI and Google that got a bunch of those as well last year.
And so I think there was about a $200 million contract that got canceled from Anthropic.
And what's interesting is, within hours of this kind of new federal directive targeting Anthropic, OpenAI stepped up and Sam Altman, their CEO, went and posted it on X And he tweeted saying that OpenAI had reached an agreement with the Department of Defense and would be taking over this Anthropic contract.
I mean seemingly he didn't really outline it in that exact way, but I mean read between the lines Anthropic gets dropped, OpenAI gets a new contract and basically they're taking over.
Sam Altman did say that he's like look, we got a bunch of safeguards.
We're not going to do domestic mass surveillance and we're not going to force, you know, have us being used in autonomous weapon systems yada, yada.
OpenAI also said that they're going to deploy through a cloud-based API so they can retain control of the safety stack, and they're going to embed personnel with appropriate clearances to oversee deployment.
Sam Altman also later said that the deal was kind of rushed, but he framed it as an attempt to de-escalate tensions and stabilize the relationship between AI labs and the government.
So he's like look, we're making sure the government doesn't hate all the AI, which obviously they don't.
They just want to be able to use it.
I think this brought up a whole bunch of really interesting questions.
Number one.
If OpenAI could secure an agreement with similar red lines, why was Anthropic not able to do this?
I think there's some critics that are arguing that the difference came down to the deployment architecture and some of the negotiation strategies.
I think other people are also saying that the dispute sort of basically became kind of symbolic.
It turned into this clash between anthropic like saying, you know, like adding new rules after they already had a deal in place, and opening eye being like look, these are sort of what we would like to negotiate and come into an agreement with.
And so I think people, some people are speculating that the government doesn't like you know, making a deal to use a service and then having all of these Rules all of a sudden, red lines added, mid-use case.
I think the public reaction was pretty positive in Anthropic's direction.
Anthropic's chatbot Claude went all the way to the top of Apple's App Store rankings.
It passed ChatGPT.
And it was kind of the number one spot for AI models that people were using.
And I mean kind of immediately after this big news story came out.
I think beyond the kind of immediate corporate fallout right, because we know Anthropic lost a 200 million deal and OpenAI came and picked that up.
Basically,
But I think there's some like some strategic things we have to think about.
First of all, the US military already operates highly automated systems, and the Department of Defense, you know, directives allow AI enabled systems to select and engage targets under certain review framework.
So I think the question isn't whether AI is going to be used in defense, but kind of how broadly and under whose constraints.
The US military already has its own rules.
They're kind of setting their own rules.
And I'm not sure this is something that like people are voting on.
Perhaps that's something that we would do in the future.
But the military already sort of has its own rules and kind of its ethical frameworks they're looking at.
And I think they I mean they really don't like another AI lab telling them what they can or can't do in regards to that.
I think a lot of national security leaders are arguing that limiting access to cutting edge systems could place American forces at a disadvantage when, for example, China none of these questions of ethics and safety that Anthropic is bringing up.
China obviously doesn't care about any of them.
Russia doesn't care about any of these things.
And so I think, when your geopolitical competitors are putting AI into their systems, we have the best AI models right now, with open AI and anthropic being built inside of America.
But that doesn't mean that we'll have the best forever.
And if we kind of nerf the capabilities of those, theoretically and this is how the argument goes that could be.
That, could you know, not be positive.
So, you know, our base could put us at a disadvantage of China.
Okay.
So from that perspective, I think the Pentagon is kind of making an argument that they don't want a single vendor to be able to tie their hands.
Basically, if something is legal and they're allowed to do it,
Okay.
At the same time, Anthropic has consistently argued that technology is advancing so fast that government mechanisms haven't kept place.
A bunch of critics, such as Max Teagard, are saying that the Broader AI industry helped create this vacuum by lobbying against binding federal regulation, preferring these sort of voluntary safety frameworks.
We see like OpenAI, Anthropic, all of them are like, look, this is our safety framework.
This is what we're doing.
And so we don't really have any sort of enforceable laws.
It's mostly just people saying, look, we want to be safe and responsible.
I think without some of these like legal frameworks, the argument could be made that disputes like this are going to be resolved through executive power and then contract leverage rather than legislation.
So this is something we'll be watching very closely.
But in the meantime it does appear that Anthropic has received a boost from just regular users as kind of they're kind of the underdog fighting the government.
And it seems like OpenAI has gotten a 200 million boost from just picking up the contract and powering a lot of that technology which could turn into, you know, higher contracts in the future.
So we'll see where this all plays out.
Thank you so much for tuning into the podcast.
If you enjoyed this episode, I would love a rating or review wherever you get your episodes.
It helps the show out a ton just to be found by more incredible listeners like yourself.
Thanks so much for tuning in.
And, as always, make sure you go check out AIboxai if you want to get access to all of the top AI models that we talk about on the show, in one place for 899 a month.
All right, catch you in the next episode.
