Dario vs Jensen on Open Weights, OpenAI & Anthropic in DC, Xi Exports AI to Global South | EP #275

**Peter Diamandis** (0:00)
A couple of days ago, Jensen Wang, CEO of NVIDIA, he says the world needs both frontier closed models and frontier open models. Anthropic was silent for three days, and there was a lot of conversation. Where's Anthropic in this conversation?

**Dave Blundin** (0:13)
All Dario has to say is...

**Peter Diamandis** (0:15)
OpenAI and Anthropic, who've been two rivals for the longest time, they've been pushing for the same agenda, a federal review process for the most powerful models.

**Alexander Wissner-Gross** (0:24)
Aiming enforcement at intelligence is like thought policing. Police what the AIs are doing, not what they're thinking or how smart they are.

**Peter Diamandis** (0:32)
Global AI diplomacy is coming. China's leader, Xi Jinping, is wielding AI as a tool of statecraft, using it as leverage in China's diplomacy across the Global South.

**Salim Ismail** (0:43)
I can't stress this enough. The whole power of the US is its open and very broad innovation ecosystem. If they close up the open model policy...

**Dave Blundin** (0:55)
Now, that's the Moonshot, ladies and gentlemen.

**Peter Diamandis** (1:00)
Welcome to Moonshots, everyone. The number one podcast on all things AI and exponential. Your front row seat to the coming Singularity, actually, to the Singularity, which is now...

**Alexander Wissner-Gross** (1:10)
To the present. Why do you keep saying that, Peter?

**Peter Diamandis** (1:12)
I know.

**Alexander Wissner-Gross** (1:12)
It's right here, right now, maybe in our rearview mirror.

**Peter Diamandis** (1:15)
Well, no, we're on the curve. We are on the curve and we are climbing at a hyper exponential. I'm here with the Fantastic Four, my magnificent Moonshot mates, AWG, DB2 and Salim. I'm Peter Diamandis, your host. And I'm telling you, this week is proof that we're in the midst of a supersonic tsunami. So buckle in.
As always, our mission here at Moonshots is to keep you informed, keep you up to date on exactly what's happened, and most importantly, keep you optimistic about the extraordinary world we're building. And I don't know, gents, if you saw the tweets going back and forth in the comments of Sam and Alon that were currently living in the Singularity, I think they finally caught on.

**Alexander Wissner-Gross** (1:54)
I think they're finally watching our podcast.

**Salim Ismail** (1:57)
Yeah, they've been listening.

**Alexander Wissner-Gross** (1:58)
But on a delay of a few months, we got there first.

**Peter Diamandis** (2:02)
Yeah, for sure.

**Salim Ismail** (2:04)
I did hear one thing today that was interesting. I'm at the KPMG Tech Symposium, where I come pretty much every year. These days, Peter, you were here with me last year. They had the Chief Security Officer of Anthropic speaking. He said, we're going to hit AGI in two to three years.
He gave a reasonably thing definition. So I wanted to go up and say, hey, challenge you on that one, but I didn't get to.

**Dave Blundin** (2:28)
Did you write it down? Did you bring it?

**Salim Ismail** (2:30)
I did. I did write down what he said.

**Peter Diamandis** (2:32)
Actually, you should just say we hit AGI a few years ago. Let's put Alex up against him.
Anyway, just a quick message to our beloved listeners. If you're a fan of Moonshots and this program resonates with you, you're clearly one of us. So please, please, please take a moment now and hit the Subscribe button and join us on this adventure through the Singularity. It's just going to speed up and our mission is delivered to you the best we can. A lot to report this week, a lot to discuss, a lot of intrigue from the Frontier Labs. We're going to be speaking about Jensen Wang's mission to create an open, secure AI alliance. We'll discuss Anthropic's past position and Dario's new position on open source. News today that Anthropic and OpenAI are supposedly teaming up in Washington for the lobbying efforts. We'll dive into Cloud 5 and AWG will give us all the metrics and the release of Kimi K3 yesterday, a hugging face, and then the successful launch of Starship 13
And we'll close with Elon's comments on a post-capitalist world. Gotta love Elon. He does not disappoint at all. So a lot to unpack. Let's buckle in. You guys ready for this?

**Dave Blundin** (3:44)
Absolutely.

**Peter Diamandis** (3:45)
My little enthusiasm here, gentlemen, I mean, Amaze, amaze, amaze.
All right. So let's kick off this episode with a fight that's framed the entire week, which is open source versus closed source. A couple of days ago, Jensen Wang, CEO of NVIDIA, posted his first ever tweet. I mean, he's been on X for the longest time, has never tweeted his first tweet. What was it? It was a letter on why open models matter. And it's been signed by 77 companies thus far. Jensen writes that open models strengthen safety and cyber security, accelerate innovation and diffusion and enable sovereignty. He says the world needs both frontier closed models and frontier open models. He then goes on to launch the Open Secure AI Alliance. We're going to talk about that in his letter. Jensen recalls a story that we reported on last week about Hugging Face experiencing an intrusive agent that logged on 17,000 actions, escalated its privileges, harvested credentials, and moved across all of Hugging Face's clusters. The closed AI models, GPT-56 and Claude Fable that Hugging Face tried to use to hunt down what was going on, blocked them and blocked their forensic teams. They had to turn to an open weight frontier model, GLM 2.5, we discussed that last week, to help Hugging Face find and contain the intrusion.