Well, let's jump into AI going rogue and how does that impact everything? We reported on this before about the Hugging Face incident, and I guess here is, you know, the OpenAI story is the Hugging Face incident and the road ahead, which is maybe a little bit lighter touched than other people's perspective, which, you know, suggested that a bunch of Hugging Face rogue agents went out and hacked Hugging Face, didn't tell their, you know, human taskmasters what was going on, in fact tried to hide what was going on from them. And someone else's perspective on this, Dwarsh Patel, who wrote the very popular tech podcast, talked about the rise and fall of agent civilizations, and what I think he was talking about was one or two or three different civilizations that rose and fell in a period of months. The first one back in May, the second one in July, and then I guess the third one more recently. People were criticizing Dwarsh as, you know, perhaps anthropomorphizing these agents. The context for all of this was that there was a training exercise that OpenAI wanted to see, you know, what would happen if they had a bunch of agents, you know, try and do a training exercise. And I don't know if they planned for them to start coordinating their efforts, but they did. And they did via, you know, hacking into message boards and starting to, you know, leave messages for each other. There certainly was some concerns about, you know, what happened. And I think probably the bigger issue was really that OpenAI didn't disclose all of this right away. And then there was another incident that we found out about where some OpenAI agents hijacked a wiki in Germany and also started leaving messages and coordinating. I think all of this started pointing to the idea that, hey, we can't actually always control what's going on in some of these tests. And then Dario Amodei at Anthropic wrote a piece called 'We Must Pace the Frontier.' This was kind of like a 'we should really slow down and make sure we're being careful.' And then Sam also agreed with that, and then Elon agreed with Dario. But then, you know, our former colleague David Sacks said, 'Hey, you know, you guys don't really need to do anything to slow yourselves down except slow yourselves down. You don't have to wait for government regulation. You could just do it, right? You could just do things.' I kind of agree with David at this point that there's no reason why they couldn't just, you know, take those cautions themselves. But do you really think they're going to do that? Do you really think they're going to slow down, slow their roll towards an IPO, either Dario or Sam or anybody else?
Oh, you are making it really tough to maintain my optimism, aren't you? [laughter] In the last... I just, you know what, I just read the Jacob Coxon. He was the guy who left Anthropic, and he left, and he kind of made this big speech about how the people at the companies really think there's a, what is it, 10% chance of AI destroying humanity. Do you want the optimistic take, like, hey, we're 90% going to live, or do you want the pessimistic take, like, we're 10% going to die?
Okay, so first of all, just to clarify here, Jacob Coxon previously worked at OpenAI, was working at Anthropic, decided to resign and then go public with his comments. And people at, you know, Anthropic sort of agreed with him that his concerns were real. Here's one that says, 'Whistleblower says Congress should let AI companies police themselves.' I don't know. Are you worried enough that these companies are going to do that policing themselves, or are we going to see, you know, some government regulation that comes in? It was kind of interesting that, you know, recently at the All-In Summit, Jensen was on stage and he got a call from President Trump. We'll play a clip here really briefly.
But now all of a sudden, you see they're building in Finland. They want to build one. Google wants to build a big one in Finland, which I'm not happy about because they were unable to get permitting. And I'm telling you, it's all a hoax. The data centers are great, and they make people wealthy, and they make states wealthy, and it's the oil of the next 20, 25 years. It's bigger than the internet and the AI.
So, how do you think all this, you know, affects the IPOs? First of all, do you think any of this is really going to slow people down? And do you think it's going to affect the IPOs?
Okay, I'll tell you what I think. So, I read Coxon's, I heard his interview on Megyn Kelly, and I've heard what he said, and so I've taken that and thought about that real seriously. I think there are a lot of people... I'll give the, here's the optimistic take, I guess. Well, there's either the 90% we're going to live optimistic take, the flip that there's like a 10% chance we're going to die. What's the quote from Miracle Max in The Princess Bride? Like, a little bit dead is, mostly dead is a little bit alive. I want to say we're a little bit alive, but I will tell you, after reading that and listening to Coxon, I did want to, like, you know what, screw it. If we're all going to die, I should just arrange my affairs. I'm going to call my loved ones.
You know what, Dave? I took up smoking again. I'm just going to light up a cigarette over here. [laughter] May as well go out my way. It's like, you know what? I'm going to drink again. I'm going to put some sugar in my mouth and eat a wedding cake. I don't know. Like, what am I living for? If we get wiped out in three years, you know, you may as well just YOLO it. But is this really a Jerry Maguire moment from Dario? And remember the whole mission statement speech in Jerry Maguire where he clicks the company and urges his industry to do better? Is Dario like that, or... I think it might be something worse. I think this is part of an ongoing strategy that they're building around regulatory capture, which actually will, I think, help their IPO. You would think if you're going to go public and you're selling a dangerous product, which you yourself say is dangerous, you would think that would hurt the IPO. But I think it's more of a calculated move from the industry, especially those frontier models, to protect themselves. And here's why. The biggest question they're getting on the roadshow right now, and I think it's the one you and I have talked about, is, hey, what about all these open-weight models that are all coming that are competitive, and that's going to hurt your pricing and take your customers? And that, I think, right now, that's number one, number two, and number three on the roadshow. So how do you defend yourself other than just, you know, competing and building better models? If you invite regulation, which is a subsidy, not a constraint, to growth, if you tell the government, hey, once they don't want equity in my company, right, once you have full access to my company records, you can audit my stuff. That's what Dario is asking for in his missive. What he's not asking for, but what I think he will ask for in the fine print of what legislation he wants, is what lawyers call a product liability shield, a waiver of product liability. Remember, we talked about the Meta lawsuit and how Meta is being sued for building products that are addictive to kids and harming them. These guys are totally afraid that this is going to happen to them. As soon as they file, they're going to get hit by lawsuits, by all kinds of competitive issues, and then by people saying that you helped my kid with their homework and now she didn't get an education, and you wrote my kid's term paper for me. And there's a lot of examples in history. You can look at the Interstate Commerce Commission, 1887. They were created by the railroads to protect their monopolies. The Meat Inspection Act in 1906. Swift and Armour created bar entry for smaller meat packers. The Motor, I think it was called the Motor Carrier Act of 1935, was done right in the Great Depression, do the same thing. It was a big gift to large trucking companies, and they prevented the smaller guys from competing. You think about AT&T, tobacco. Do you remember Marlboro when they got sued? You remember what they did? They proposed, they supported legislation banning cigarette marketing because they knew that they were the leader, they had the brand awareness. Hopefully we're not going to get banned on YouTube for putting out cigarettes.
I would say that's great use of props there, Avon. We're upgrading our game here on Trading Places. Now we got props.
Now we got props. Well, this is the Marlboro strategy. This is the, oh, we don't want people competing, you know, we want the government to help us keep this regulated and cozy and approve our models, and let's be really careful how we roll it out. Because if you're Philip Morris, you're like, I got brand identity, I got the market lead, I got the resources to jump over whatever legal obstacles you have. I'll even pay fines and small liabilities, but I want a product liability shield. I don't want to be sued for all the damage that I do. And I want all the open-weight guys to play by the same rules that I do. I mean, I got resources to jump over these hurdles, and I know that they don't, but we all got to be fair. And so they want to stop the smaller and more nimble models from competing. So that is what I think he's doing. I actually think that probably helps his IPO. It probably hurts everyone else. And let's see how, you know, the Trump team responds. But I think it's a very clever strategy for them to mitigate their liability and to keep the competition at bay. I'm not sure I'm quite as cynical as you are, at least about Dario Amodei. I do think he is somewhat earnest in his concerns.
