9/22/26: Trump Media Ban Backfires, DraftKings Caught Targeting Addicts, AI Anthropic Wet Lab

24:00That's not going to work out for you over the long term. And this man, I'm sure, is not intellectually so stupid that he can't also do that math, but he's hooked. And what's he hooked on? It's the promise of this time it's going to be the time. This one is going to hit. This one, it's that constant dopamine hit, the anticipation. What if it's this one? What if I get this big payday? What if that changes my life? And increasingly, this is what our entire economy is built on. It's no longer the American dream. This is the American dream, is that your parlay is going to hit. For young men, that's like the new America.

24:34And it's not going to. It's rigged just like the original American. It literally will not. This one is even more blatantly rigged against you. Over the long term, maybe you'll have a good, you will have a good week here or there where you're feeling great. Over the long term, they will take everything. Everything, if you continue in this direction. Right, and so this is now paired with, look, none of this is actually all that revelatory. I'm always amazed by what things break through and don't, but for some reason, people woke up to things that we've been talking about here for many years. Let's put the New York Times story up on the screen, E1, where the Times reports about how DraftKings uses AI to target the gamblers likeliest to lose.

25:11Again, if you are surprised by this, this is called basic practices. This is how they make money. This is why online sports gambling is so devastating because you have the phone, which you're already addicted to, combined with money, which you can't afford to lose, combined with the most sophisticated machine learning algorithms that even predate these AI models, which is exactly what they write. In 2023, DraftKings took customer betting records and built a machine learning model, which seeks the patterns in the data to answer the question, who is more likely to respond to promotions by gambling and losing even more money?

25:49They decided to prioritize free bets, bonuses for likely losers, and soon a question began to gnaw at them, aren't many of these people prone to addiction? The guy decides to be a concerned whistleblower. And no offense to the person who ran these tests and is now speaking out, dude, that's your whole business model. And I've been trying to emphasize this too from the very beginning. Literally 50 % of the profit of these companies, 50 % comes from the bottom 6%. The addicts, 50 % from the small group. This is like this with all sin, is that the vast majority of cigarettes are smoked by daily smokers.

26:27The vast majority of alcohol is consumed by alcoholics. The vast majority of money lost is lost by the whales, by the people who are literal addicts and who generally have a large sum that they haven't yet exhausted. What we also know is that all of their advertisements, which this guy we showed you almost certainly fell for, is you can win. It's not true. If you are allowed to bet $250 per game, you are a loser. I can tell you that for certain. Why? Because their machine learning algorithm is so sophisticated that if I logged into DraftKings right now and I bet in a sharp way, sharp means that somebody who has a general edge, if I bet sharp, even if I lose, it will say, oh God, sharp customer.

27:13We don't want that. And so they will come into my app and say, you're allowed to pay$2.90 a game. So they can't ban me, but they can say we can limit your action. If you are ever allowed to bet in a sustained manner for more than$100 a game, you are a loser. The machine is telling you, loser, loser. And that is what a lot of these guys don't seem to understand. And they plow them because, I mean, again, people are idiots, but it's like, you know, I get it where, where it comes from free bet. And we're like, Oh, bet five win 200. Oh, that sounds great. 200. I can pay my electric bill. And so what do you do?

27:51You log in. It turns out it's$200 in free bet credits, or you have to cycle through a real$200 and then you lose it all in the first place. Profit boosts, all this other, you know, promotional nonsense. It's just addiction. It is designed to get you there. There's no consumer protection. There's no, you know there's no scruples i mean the people who built this are just refining the same tactics of the slot machines which are designed to make you go broke over time yes if they bleed you on every spin it wouldn't be fun they have to bleed you so that you lose it over two hours instead of 20 minutes so you think you have a chance that's the game yeah you know it's hacking the human mind right that's what it is so all of those are personal responsibility etc they spend millions and millions and millions of dollars to figure out how to hack human psychology and get you in and keep you in.

28:47And for, unfortunately, a significant minority percentage, especially of young men, it works like a charm and they get you in their grips and it's very hard to get out of. The thing you're talking, I mean, I think it all should be banned. I think we should go back to the days when you had to go to the casino and there's some barrier to entry or you had to like, you know, know a bookie and do something sketchy on the black market, whatever. I think we need to go back to those days because it's so clear to me now how devastating this absolutely is. But at the very least, the kind of crap you're talking about where they can just rig it, where if you're too good, they just like ban you from the app or they, you know, make it limit your actions so that you can't ever come out ahead.

29:26How is that legal? That shouldn't be legal. That's completely insane. It's nuts. But they've bought the system and they've paid off whoever they needed to pay off. And so they're able to operate in what should be a completely criminal fashion that I think anyone would look at. And the other thing about this is like, this is what we're getting for AI. Where are my cures to cancer? Where's the solution to climate change I was promised? No, instead I'm getting AI-derived algorithms to target gambling addicts and make them even more miserable and drowning in debt than they already are. That's great.

29:59Love that. Let's talk about the prediction markets. So if I were to sit with the prediction market, they would say, Sagar, you're absolutely right. That's why the casino industry is corrupt. And that's why our platform is so much better. Because by definition, we can't limit people's action. We are not the house which is setting the line or taking advantage of these people in a certain way. And to that, I would say, show me your win-loss percentage. Just show me the amount of customers which actually take profit. Yeah, it's great that you don't limit sharp bettors. It's great out of all of that.

30:31But how many of the people actually win? And look, we can do some general math. Let's go ahead and put E6 up here on the screen. So it turns out that last week, so week two of the NFL, take a look at this. Calci did$15.3 billion in notional volume last week. 60 % of that volume is on parlay bets. which are entirely based on sporting events. This is Dustin Goker. He's a longtime analyst in the industry. He says, my read is it's going to be pretty difficult to frame this as non-gambling activity. Yeah, you think? And this is where, you know, if I were to sit with Calci, because they're going to sell me this whole spiel about how we don't ban bettors like the way you're talking about, Crystal, that's immoral, and that's why we're so much better.

31:22Well, guys, if 60 % of the money is being bet on a parlay bet And by the way, as we showed you on this show last week, the people who are taking the other side of that parlay are FanDuel and DraftKings, then you're doing the same shit. And that those parlay bets have a 20 to 15 percent house edge, which means that 15 to 20 percent of money went to the same company that Calci and them are saying that they are better than. That's where all this money is going. I mean, it's just so like$15 billion in a single week. This is bigger than MPAA, the motion picture industry. This is bigger than all forms of entertainment combined.

32:06And it's just business is normal. Enabled by the rest of these, enabled by the Trump administration, but every state in the nation because they've all got their handouts. You know, to the extent the states have a problem with Talshi, they're not getting any tax revenue. Tax revenue is the problem. Here, how about the fact that bankruptcy and credit card debt goes up by 25 % in any of the state that legalizes this stuff? And now we have California and Texas that are in on the action. Just wait until the data comes out from those states. It's going to be horrible. And, you know, you think about we've talked about young men struggling with crisis of purpose and meaning and challenges in the dating market and all those sorts of things.

32:47and this is sort of like, feels like a band-aid for that because you get that dopamine hit, like that guy that we played before. How do you think this is going to go for you long-term? Do you think this is going to increase your desirability? Do you think this is going to increase your, you know... I'm not speaking my language. I mean, it's true though. I agree, I agree. I was just thinking about, my daughter talks about this, my oldest one, with some of the guys that she's dated, where she's like, they're always like, my parlay's gonna hit. I know, I know. I'm going to, she's like, your parlay is not going to hit.

33:18Okay. Give it you. I have never experienced you hitting your parlay. It has never happened. Stop it. And it's like, I'm sure she's not alone in feeling like, why are you obsessed with this and throwing your money away? Don't think that I'm going to be in this one for the long haul. So just, you know, just another thing to add to the mix of your considerations as you're approaching this stuff. I want to tell the guys that I, and I'm, I'm trying to start, maybe I'm the wrong person for it. I want to start the same type of temperance movement from the 1920s where women who are out there need to make it known.

33:50If you are a sports better, you're a fucking loser and you're not going to get a date. And by definition, you're a literal loser. And second, it means you have bad judgment, you have bad impulse control, and you're not somebody who has the wherewithal to stand up against the system and provide for me in the long term. I think people should recognize that. If you're a woman and you're out there on a first date, one of the first questions you should try to suss out is, hey, do you regularly bet on sports? If you do, red flag, leave. Completely. It's one of those where, you know, and, and, you know, uh, look at the data.

34:22Uh, Dave Ramsey says it all the time. Money is the number one cause of divorce in North America. So if you think that that's not going to spiral, you know, out of control, then you're going to see too, like this, that, and the type of person who thinks the parlay is going to bet is exactly the type of person who is going to have long-term credit card debt, who doesn't understand the financial, you know, situation who by definite, by literal definition, can't do math. I mean, all of these are not working out in your favor. I say all of this though, and I still feel bad because I'm still pitting the individual against the system.

34:55And the point is the individual should be protected by the system. But that's not how our economy works. Our economy, I mean, you know, I've flown for free all over the world off of credit card points. I have to be real. The only reason I get to do that is because of idiots like that guy who carry 80 % balance and pay, you know, exorbitant interest rates. Like the flights aren't free. And, you know, if you own stocks, the whole U.S. economy relies on this guy, you know, just going out and spending. And that's the tragedy. Like we're all kind of complicit in it. But at the end of the day, you know, you need people who are courageous enough to be like, I don't want to make money.

35:33You know, not all money is easy money. Like it's one of the most ancient things in society is to say usury is bad. Debt and debt, you know, people who purvey it are evil or like pervading the community in that way. Exactly. Like the horse thief was always considered one of the most villainous crimes of the West. And it was like the type of person who would do that in the absence of law and order has to be hung or shot is because they will violate the most basic social contract. and they will take advantage of the overall system. But whenever we have a system which is so disconnected and kind of faceless in the way AI is, we don't have community enforcement in that same way.

36:17So anyway, we should hang horse thieves again. I'm not joking. I'm not joking. I didn't know the history of horse thievery and how consequential it was in Western civilization. It's very important. It's very important. In the old way, look it up. Look at the way they saw it. I believe you. Yeah, no, we've got the three legs of the American economy stool are like healthcare for old people, gambling, and AI bubble. AI CapEx. Yeah, that's a very healthy economy we've got here, guys. All right, Kevin Rue standing by. Let's get to it. We know kids don't hold back, and I wouldn't want them to. Get in there, make a mess, slide for home, bring on the stains.

36:56I've got OxyClean Versatile Stain Remover. It works on over 100 stains, and it's America's number one stain fighter additive brand. It's color-safe and chlorine-free, so I use it on whites, colors, and dark fabrics. And OxyClean isn't just for laundry. Mix it with water for use all around the house, kitchens, bathrooms, and even floors. When my kids go all in, I say go for it, because I've got OxyClean to get the stains out. It's not clean unless it's OxyClean. Have you ever wondered why some people seem to age slower, like they figured out some secret the rest of us missed? Well, it turns out someone actually did, and it starts with a melon.

37:30Yep, a melon. There's a rare French melon that stays fresh four times longer than the fruit you and I buy at the store. Scientists discovered it produces an extraordinarily high level of a super antioxidant called SOD, the kind that helps defend skin from the visible signs of aging. And get this, Meaningful Beauty partnered with world-renowned French cosmetic specialist Dr. Jean-Louis Sabat to capture that melon's youth-preserving power in a skincare line you can actually use every day. So if you're tired of trying products that all promise the same thing, this one is truly different. Because it starts with something nature designed to stay young.

38:07Subscribe today and get the complete introductory system for just$49.95 at MeaningfulBeauty.com. You'll also get two free gifts, free shipping, and a 60-day money-back guarantee. Visit MeaningfulBeauty.com and order today. This is Jacob Goldstein from What's Your Problem? Running a business is hard enough. Don't make it harder with a dozen apps that don't talk to each other. One for sales, another for inventory, a separate one for accounting. That's software overload. Odoo is the all-in-one platform that replaces them all. CRM, accounting, inventory, e-commerce, HR. Fully integrated, easy to use, and built to grow with your business.

38:48Thousands have already made the switch. Why not you? Try Odoo for free at odoo.com. That's odoo.com.

39:25You see reason for optimism and worry as the world wakes up to what Silicon Valley has been talking about for years. You don't have to give away your book and we'll have you on separately to talk about it. But this is your area of expertise. You've been following pretty, you know, daily the ins and outs, kind of all of these discussions around AGI. Give us a sense in the column in particular, the world is waking up. There's something to be done. But what actually needs to be done and what are we waking up to?

39:53Kevin Roose:Well, the thing we're waking up to is what the people in San Francisco where I live and in the AI labs where I just spent the past couple of years reporting this book have been saying amongst themselves for years, which is the systems are getting very powerful. That progress does not appear to be slowing down. In fact, it is accelerating because the AI is becoming capable of contributing to development of more AI. So we have this sort of recursive self-improvement, as it's known, that cycle is starting to kick off now. And this is something that people I've talked to in the industry have been worried about for years.

40:27Kevin Roose:There have been papers and blog posts and conference panels about this moment. But it was a very small group, and they were very sort of fringe within the academic research community, and nobody really took them seriously. And what we've seen over the past couple of weeks with Jacob Coxson resigning from Anthropic and Dario Amadei putting out a statement about pacing the frontier and all the other AI CEOs piling in to say, yeah, we think this is going too fast. That's what people are waking up to. And I think that's the message that is finally being heard. Got it. Let's go and put F7 up on the screen.

41:00This was a recent announcement that speaks to the power of these models. For a long time, particularly on the left, there's been a lot of skepticism that the models would ultimately be very powerful. They view it and some still view it as just sort of fancy autocorrect or fancy or stochastic parrot or, you know, that this is all really hype. You now have OpenAI announcing that they have resolved, their model has resolved more than 100 longstanding open problems across most areas of mathematics. We're waiting to get the list of what exactly those are. But it is in recent memory that these models could not solve basic addition, elementary level.

41:42Then it was high school level. Then it was elite high school level. Then it was college level. Then it was PhD level. Now you have OpenAI claiming they solved one of these millennium math problems that if you're a mathematician and you solve it, you get a million dollars. There is some contestation about exactly how all of that went down. But if you look at the progress over just a few years, this blew away what even many people who were very bullish on the power of the technology predicted here. So help put this in context, this latest revelation from OpenAI and what it says about the current capabilities of these models and where they may be going.

42:18Kevin Roose:It's absolutely huge. I mean, the Millennium Prize problems are the seven open problems in math that are considered the hardest. And as you said, if you're a human mathematician and you win one, you not only get a million dollars, but you become a famous mathematician, right? People sort of record your names in the annals of history. these models now are solving these problems like they're easy. So the models have gone from quickly being subhuman at math, being not able to do basic arithmetic, to being wildly superhuman at math. And you can run them in parallel, many, many copies of them, to expend the kind of time trying to crack one of these problems that a human would take many lifetimes to spend.

43:02Kevin Roose:So it really is an unfair fight in mathematics And I think what the people on the inside see that maybe it's harder to understand from the outside is that this is not just math. This kind of progress, this exponential growth in capabilities is happening in discipline after discipline after discipline. They are seeing this sort of wherever they point their kind of tractor beam of compute and resources, they are finding that these models are incredibly capable. So, yeah, I think it's a very significant sort of application of these things. I hope that they're followed by other big problems in science.

43:37Kevin Roose:But this is the kind of progress that has been freaking out people on the inside for years. Kevin, you don't have a vested financial interest in any of this. You're a reporter. You've been following this now for quite some time. We will hear significant pushback, even from some of our own audiences. We're falling for marketing hype. You know, 100 problems, what does that even mean? Isn't this just marketing for their products to get people like us to talk about it? How do you suss that out, having been in this world now for quite some time? Yeah, I used to share some of that skepticism. I thought these tech companies are always telling us they've changed the world and the truth is never that simple.

44:16Kevin Roose:But what convinced me was just using the models and seeing how good they were getting at doing programming problems for me, at doing various tasks that I needed them to do. I didn't used to be able to use these models for serious work, and now I do. But I think the thing that is, even if you're skeptical that this is all marketing hype, which I don't think it is, by the way, because these people have been saying the same things for years before they had anything to market. I would just say the thing that should still freak you out is how much these AI labs are relying on their own models for internal automation, for doing their own work.

44:50Kevin Roose:Anthropic recently put out a thing saying that Claude, their AI, is now the lead researcher on 26 % of their AI research and development tasks. Not an assistant, not something that they can sort of hand tasks off to, leading the project. So I think that should freak people out, even if they're skeptical about the claims about the Millennium Prize problems or whatever. The labs are now using this stuff so regularly that they are starting to kind of turn over control of these systems to their own AIs. Your former colleague Ezra Klein wrote about this recently and did a podcast about it and talked about how the decision that's being made in those instances is not that we're losing control of the AI, which is obviously something that has also happened that we are concerned about seeing in the future as well.

45:42What recursive self-improvement means is that you are handing over control to the AI, and he called for an outright ban. Do you see that as something worth pursuing or something that is even truly feasible, given that sometimes the definitions of these things can be sort of hard to pin down?

46:00Kevin Roose:Yeah, I think I'm sort of behind it in spirit, but I would obviously want to be very focused on the details. I think it's very hard to implement something like a ban on recursive self-improvement. But does that mean that if you work at OpenAI, you can't use AI for coding? But if you work at a different company or a supplier company, you can. I just don't know how that works. But yeah, I think this is the real danger here, is not that we are going to lose control, that some rogue agents are going to sort of break out of their box and go attack companies. That obviously would be bad, and we should prevent that.

46:33Kevin Roose:The danger is that we are creating this cycle inside the AI labs. they are turning over. They are giving away control of their own research and development process to their AIs, which they have not figured out how to control yet. And so we are basically - And which they don't understand. You know, I can't remember who I saw referencing back, Sagar will probably remember recently, to the incident you had with one of the early models where it like, you know, convinced - Yeah, that was Google. It was like, fell in love with you and was trying to get you to leave your wife and all this sort of crazy stuff.

47:05And everyone's just sort of like, well, that's weird. anyway, moving on. But we never actually got to the, like, why was it doing that? Even from that early version, these things are what they call black boxes. It's not like a program that you code. They have behavior that is emergent, that is not understood. And so rather than trying to figure out what's going on so that then we can build more powerful models that we perhaps understand, we never got to that place of even understanding the more basic models. And I'm not or have a feel for that because you're seeing this tremendous opposition to what the AI labs are doing at this point.

47:44Kevin Roose:I think that's right. I think people, even if they don't really have the terminology to like put their finger on exactly what it is that feels bad about what the AI companies are doing, I think people understand that these systems are getting better and we don't like it. It's a threat to our jobs. It's a threat to potentially our lives and our children's futures. Like this is what freaks people out. I started getting texts from people over the last couple of weeks. just people outside the tech industry, like, hey, are we all going to die? That seems bad. So I think this is breaking through, and I don't want to be glib about it.

48:17Kevin Roose:The people inside the labs, they are genuinely concerned that the technology that they are building could kill everyone on earth. That is not marketing hype. That is not a stunt. That is not just some clickbait that people are putting on their press releases. I have talked to these people on the record, off the record, in public, in private. They say the same thing every time. In fact, they are watering down how likely it is that they think AI could kill us all in public. Their real projections are even scarier than what they'll say online. Wow. You see, so many people have such an immediate revulsion to that because you're like, then why are you building it?

48:55Like, what are you doing here? Can you answer that question for us?

48:58Kevin Roose:I think there are a few reasons that I've heard people give. And again, I talked to more than 150 of these people from my latest book. So I feel like I understand their motivations pretty well. For a lot of them, it's the curiosity of it. It's the what's out there. They're seekers. They want to know, like, how smart can we make these things? What could they discover? They want these things to go out and solve diseases and cure cancer and fix the climate and things like that. For other people, though, there's this sense that if we don't do it, someone worse will, If we, the AI company in San Francisco, don't create this technology, the recipe is not a secret.

49:35Kevin Roose:It's out there. Anyone with the right resources can build, theoretically, the self-improving robot that takes over the world. So we have to get there first so that we can put our safety practices in place so that we don't blow up the world. I want to talk a little bit about the financial aspect of these things. So both Anthropik and OpenAI are planning IPOs. Anthropics is more imminent. OpenAI, Sam Altman just said, they're pushing theirs off a year. The always colorful Alex Karp, CEO of Palantir, was recently on CNBC saying making the claim that he doesn't think OpenAI will ever go public. The reasons he gives are kind of interesting.

50:16Let's go and take a listen to that. The first line of defense is you're liable for your own actions. So it's a weird twist. They're asking for societal regulation to get out of the first line of defense, which is if you build a technology that can destroy 10 percent of the world, that has civil and criminal liability attached to it. The first step is to say, OK, it is not the first step isn't to say we get out of it because we want to self-regulate or we want regulation. The first step is to say, OK, you've disclosed this. What are you doing about it? And the second step is to say, if you're not doing it, you're not being responsible here.

50:51We're going to have to get people involved. That's going to be also very complicated because regulating these things needs people. You need people who understand them. Another problem we have is seemingly everyone who understands this is on some payroll. So it's like, you know, it's like who are you going to trust? You know, when we built nuclear energy, we built the bomb. People did it because they believed. They were not on the payroll. The liability that you're talking about, how does that get written into the S-1? What do risk factors sound like? This is the thing. you're assuming that there will be an S1.

51:22Okay, maybe there will be. The only way to deal with this kind of liability is to go to the government and say, nationalize us, please. You know, a lot of times in business, people don't say what they want. They're leading you to the conclusion. The reason why people have not understood this is, again, every investor, and there are a lot of people who think they're going to make money, they're like, oh, the businesses are the mark. What is the first lesson in poker and in life? If you think the other person's the mark, you're the mark. You are the mark. And it will be nationalized simply because of the liability issues.

51:58So he says it will be nationalized. There will be no IPO because of the liability issues. And I mean, it speaks to what we're talking about here. I have lots of issues with Alex Carp, as you can imagine. But how are you going to have a bank that underwrites a company that's saying, eh, there's a pretty good chance we may destroy the whole world. And if we don't, we may cause some sort of a totally catastrophic mass death crisis. Let's go ahead and please underwrite my IPO now. Does he have a point here? How do you think about the financials of these companies as they move towards going public?

52:34Kevin Roose:Yeah, I don't think he has much of a point here. I don't understand the whole case he's making. But I think it's basically that these companies can't go public because if they were public, they would get sued for their products causing harm. And look, I wish the public markets were that effective a check on corporate malfeasance, but we have pharmaceutical companies, airlines, all manner of companies who have products that have risk attached to them that have gone public just fine. I think if and when Anthropic and OpenAI decide to go public, they will go public. Banks will line up to underwrite them because guess what?

53:06Kevin Roose:It's going to be a trillion dollar plus IPO each time for both of those companies. So the fees that they make alone will be astronomical. These companies are not in bad financial shape. People keep saying this, especially on the left. And I think it's a weird, like, it just is out of touch with reality. Yes, these businesses are spending tons of money to train these models and put up data centers and all the rest of it. Yes, they are not profitable. But their revenue is growing faster than any companies in the history of capitalism. And we're sitting here thinking, oh, well, if you carry the one and subtract five, it doesn't pencil out and they can't go public.

53:43Kevin Roose:I think it's much better as a mode of analysis to assume that these companies are going to be extremely successful unless and until someone stops them from building the recursively self-improving superintelligence. Yeah. My last question for you here is around some of the existential risk. So there was recently this report. Let's put F5 here. where Anthropic is now operating a lab that conducts biology experiments known as a wet lab, where, quote, obviously it's done under the auspices of trying to create cures for diseases. Sounds a lot like gain-of-function research, which is kind of the way I've been thinking about a lot of recursive self-improvement and some of the other developments within AI.

54:26But talk through us, you know, do they really believe that they're doing it, much like with gain of function for good purposes? And do they even think, you know, Anthropic is supposed to be the company that thinks about risk and still presses forward with what I think is the most outright example of everyone being like, oh, I don't think we need any of that.

54:47Kevin Roose:Yeah, if they ever start experimenting on AI pangolins, I'm out of here. Hold on a second. But no, look, I think the term wet lab has a lot of baggage attached to it for obvious reasons. Everyone remembers COVID. But this is the thing. These companies are serious and sincere about wanting AI to do science. They feel like they have to deliver on some of these benefits, like curing diseases and stopping certain forms of cancer, because otherwise people will just see this as the slop machines that just take your job and make your kid miserable and help them cheat in school. That is the PR crisis that they see.

55:26Kevin Roose:So they actually do sincerely want to have their AI start curing diseases and making progress. But to do that, you need to automate the whole process. You need to get the human bottlenecks out of the way, whether that's having faster clinical trials, you have to give them control of a wet lab so they can bring pipettes over and I don't know, whatever they do in the wet lab, that has to move at the speed of AI rather than the speed of humans. So I think they're sincere. I think it's high risk, but I think it's also potentially high reward. And I think if we want these people to do AI for science, which I personally do, I would rather them do that than the recursively self-improving superintelligence.

56:05Kevin Roose:We have to be okay with a little bit of risk and things that might sound weird. A few wet labs. Kevin, last question for you. There's been a lot of debate also about the current legal landscape and whether it is sufficient as currently in existence to hold these companies accountable for the harms that they do cause. So there's been a whole debate, a lot of antitrust folks involved. But just to use like a hypothetical example, let's say there is a lab leak, an AI-caused lab leak from the Anthropic wet lab. Do we have the legal tools to hold Anthropic directly liable? Do we have the legal tools to hold OpenAI liable for their model breaking out of the sandbox and hacking, hugging face?

56:52You know, is that something that's already in existence and just needs to be enforced? or do we actually need new laws specifically for those type of instances? I also think we need laws to regulate what the labs are doing themselves. I'm in favor of no recursive self-improvement, stopping the development, ban on superintelligence, but even putting those things aside. Do we have some tools now where we could hold these CEOs accountable for the things that their models do?

57:19Kevin Roose:We do, but it's not enough, and I'll tell you why. So we have product liability laws in this country that protect consumers. But those laws only apply to products that are sold to the public. And they also don't cover sort of extreme events. So like, say there's an internal deployment at one of these companies, a model that's not been released to the public yet, that goes off and does something. This is exactly what we saw in the hugging face attack, by the way. That was not a public model. So those product liability laws would have a hard time covering something like that. It's also going to be cold comfort if there's a lab leak of an AI-engineered pandemic that kills millions of people.

58:00Kevin Roose:We're not going to be worried about how can we sue the AI company, right? I would much rather prevent the harm than be able to reclaim some economic damages from the harm after it happens. So the argument they would say, Kevin, is that if the threat of that was real, then they would take the precautions to make sure that the lab leak didn't happen in the first place? I don't buy it. Airlines fly planes all the time. What keeps them honest is not the fact that they could be sued for a plane crash, is that there's a regulatory regime. It's that there's the FAA. It's that they're mandated to do maintenance checks on all the planes before every flight.

58:38Kevin Roose:We have laws and regulations in place to keep people safe when they are flying in an airplane, but we have zero laws, zero laws that are tailored to this situation where you have AI companies in San Francisco building godlike super intelligences, they can do that much more easily than United Airlines can get a 747 off the ground. Very well said. Also, we should not want to, because if we live purely in the world of liability, it's like that famous scene in Fight Club where it's like, well, if the cost of paying out dead victims exceeds what it would cost to fix the problem itself, only then will we actually fix the problem.

59:15Otherwise, it's easier to just pay the fine as we've seen throughout history. So anyway, very interesting. Kevin, we're excited to have you on relatively soon to talk about your new book whenever it comes out. But in the interim, congratulations on the podcast. Everybody go subscribe. I believe it's launching very soon. There'll be a link down in our description. Thank you, Kevin. Thanks, guys. Thank you guys so much for watching. We appreciate it. I'm going to be out for the rest of the week. So Ryan's in, you know, so you can all subscribe and all of that. but I'll be on next week. I'll see you then.

1:00:15When it begins to take over 90s Los Angeles, Wanda is betting on a new generation of artists. But in a city where every opportunity comes with a price, they'll have to decide what they're willing to risk to leave their mark. FX is the drop. Now streaming on Hulu.

1:00:34Sometimes the road of life goes a little off-road. And the 2026 Nissan Rogue is ready for anything the day brings. With all-wheel drive and five different drive modes, it's designed for adventure. With room for the whole family. And since the 2026 Rogue has Google built in and a 12.3-inch display, you can just pick a destination, punch up a soundtrack, and take off. Adventure is waiting for you and the 2026 Nissan Rogue. This Friday. Me, me, me. No one's coming for us. It's just you and me, boy. 58 miles. Heart of the Beast is the must-see adventure of the year. An incredible story of survival.

1:01:17I got you. One of the most extraordinary on-screen bonds you'll ever see. Only the greatest dog to ever walk the earth. Brad Pitt. I'll get you home. Heart of the Beast. Only in theaters this Friday. Ready PG-13. Maybe inappropriate for children under 13.

1:01:33Kevin Roose:This is an iHeart Podcast. Guaranteed human.