Whitney Webb's Chilling Theory About the Next Financial Crisis, The AI Poisoning Attack That Should Terrify You, Why Japan's Bond Market Should Terrify Everyone | Weekly Recap

0:00Let's talk about the moment your side hustle becomes a real business. Now, I see way too many founders stay stuck running their business like it's a hobby. And then they wonder why nothing ever feels real. It doesn't feel real because it's not. I've helped thousands of entrepreneurs build real businesses through impact theory university. And the first step in the path to building a legit business is registering it. And Taylor brands will help you do just that in a few clicks. An LLC, a limited liability company is an official business structure. It's inexpensive to form and very simple to maintain. It's how you open a business bank account. It's how customers know they're dealing with a real company. And Taylor brands doesn't stop at the filing. Their dashboard tracks what comes next based on what your business actually needs. So EIN number, registered agent licenses and permits bookkeeping, business banking. I tell every founder the same thing, just make it official.

1:00Click the link in the show notes to check out Taylor brands and get started on your business today. This is a paid advertisement Whitney Webb. She was friend of the show. She called us to op store. Did she really? Yeah, she's a lot of the large podcasts I've been on are actually like ideologically captured. No, but wait, did she name us? She didn't name us. I don't buy it. I don't buy it. There's no I love you Whitney. Please come back. You know, one of my favorite guests we ever had. Most podcasts today are really national security operations. We were supposed to get on early of the year then she like didn't respond. Then she went on Twitter and was like, yeah, I can't be doing podcasts no more because the government's in all of them and that. Oh, she'll have the name checked me before all. I'll buy it. Whitney is not swept up in that. With Jason was Hollywood. There's no is really contact in my ear and nothing. I promise. I feel like I shouldn't say that because now people are going to think it's really contact in my ear. But anyway, she is laying down this framework for why she thinks we're having this push to go into stable courts. The cyber attack seems like a very possible scenario because it's a way for the banks to absolve themselves of any role in a financial crisis.

2:03It's an easy way to consolidate banks so that only systemically important ones survive the hack. And it could also potentially be a very effective means of onboarding people onto this new digital currency paradigm. So for example, let's say this financial cyber this cyber attack on the banks takes place and they say, well, the existing money in your account has disappeared. The hackers took it, but we can return to you the exact same amount of money you had. But it won't be in the dollars you had before. It will be in USDC or this dollar back stablecoin or it will be this token. That that is distressing precisely because it is so believable. It is very clear that what Besson is trying to do is use stablecoins to get appetite globally for US debt. It's actually a brilliant move. And to be honest, we need to do something because we're certainly not going to go into austerity. So we need to do something in order to begin to get more appetite back to the debt,

3:08which is probably about to get worse. I'm hearing rumors of China is planning to sell a large amount of US debt. We'll see if that actually happens, but they've been selling US debt for ages. Obviously, what's going on in Japan? Japan is going to be selling debt. I think in the not too distant future. We're seeing that some of the attempts to defend the yen just aren't working. And so eventually, if Japan is going to keep going down the road of defending the yen, they will have to start selling. There's only so much that the FEMA loans are going to be able to do for them with the deficits that we're running and the fact that countries are now moving away from US debt and gold is supplanting it in central banks, we're going to need to create that appetite. And so you can imagine if you believe that your government will do things against the will invest interest of the people in order to get the monetary system under their control and in a position where they feel like they can maintain a slow, steady decline instead of an abrupt recession or depression, which we have endless evidence that they will do,

4:09then this starts to become plausible. Now, this is obviously just her running a thought experiment. This is not her saying she has evidence or anything like that, but this is distressingly plausible. There's the potential here for people to steal money from you and then essentially give it back to you but with new conditions and with a new social contract. Yeah. You voluntarily accepted this new system and voluntarily onboarded to this new paradigm, but it's a coercion to an extreme degree. All of your money was stolen, right? And now you can only get it back if you take it in the form of the digital dollars that we approve of or the, you know, the CBDC depending on where you are. That would be an absolutely horrific thing to happen. This is why, man, do I really want to see America re-embrace freedom as one of its principles and getting back to the point where freedom is something that we're willing to pay a price for because yes, there's always going to be an excuse. There's always going to be a danger on the horizon and when there's a danger on the horizon,

5:12people real fast stop caring about freedom and they just want to be safe. And so the government can and will scare us into the point where a huge portion of people will gladly accept that kind of deal, but the problem is that as we change our social contracts, as we start handing over to the government more and more authoritarian control, and we say that the individual should be afraid of the government, instead of the government, being afraid of the individuals, we in America, we become something totally different. We've got to be very clear-eyed about where this can go so that we put enough pressure on our politicians through voting, through speaking up, letting them know how we feel about this, so that they don't do that because politicians are downstream of what we get excited about what we're going to vote for. So never lose sight of that. If we are culturally excited about, yeah, more control, give it to me. Come on, big daddy, tell me what to do. That's exactly who's going to get elected. We should want to continue running the exact opposite experiment, which is freedom first. Otherwise, you get what she's talking about. Alexander Mayorkas has said on record

6:17that the next big threat to Americans is a cyber security event that he called Killware. And Killware certainly sounds very scary, but if you read the existing definition of it, it refers to cyber attacks on essential infrastructure that has the potential to kill people. So it doesn't necessarily kill people like in the name, but it attacks things like water systems, the power grid, essential infrastructure that people rely on every day. This is where you want to talk about getting into a position where you can do things that make sense in terms of where we're headed and would create jobs is start protecting your infrastructure. Take some of these threats seriously. Get people together. Think of it as a Manhattan project for infrastructure protection. So whether that's water, whether that's the electrical grid, whether that's certain communication devices that we would need in the case of an EMP attack or an AI driven cyber attack, start building out those systems. Now, don't wait until the first thing happens. This is why looking at the way that people are starting to respond to

7:21drone warfare is very interesting building these big metal cages around oil infrastructure. Because yes, they are available to attack from the outside and it's easy enough to intercept a ballistic missile, but it's very difficult to intercept a drone. But it's actually, I mean, look at cost money and it requires people to work and build it, but you can pretty low tech build these um, grates around them that stop the drones from getting in close waiting until you have the problem is stupid, reacting now when you know the problem exists and it's coming your way and building against it, that would make sense. Just for the record, Joe Biden signed the infrastructure investment in jobs act. Let's go. That's great. So this is one of those things where looking at the like specifics of it are going to be important. However, if like, let's make a pact, if you can hear my voice right now, our pact is we don't care if somebody's Democrat, Republican, independent doesn't matter. We care deeply about their value system. We care deeply

8:22about what they the policies that they are campaigning for and what they want to put into effect. We care deeply about their ability to understand um, first principles thinking, right? Those things we care about. We don't care about party affiliation at all. Now, if all of us can get on board with that, cool, we've got a shot of actually thinking through these problems well and getting to the other side of it. If on the other hand, we outsource our brains and say party just tell me how to vote, now we're going to be in a bad place, a very bad place. You become super easy to manipulate. 2020 actually simulated a massive kill wear attack that was targeting a US presidential election and the way they conducted this kill wear simulation, which they did with DHS and some US law enforcement agencies was to get the US presidential election canceled and martial law declared. So they basically through that simulation established what types of hacks could succeed in achieving the declaration of martial law and the cancellation of a US presidential election, which is pretty

9:24un-meat, pretty extreme. So why are they doing that? Why? I don't know. The cheeky smile. Why does an intelligent linked company want to simulate with DHS and law enforcement a series of hacks that disrupt national infrastructure during a US presidential election and martial law is declared? I mean, here's the thing. That's actually a reasonable kind of threat assessment to run and to be prepared for that kind of thing. Check where your vulnerabilities are, stuff like that. Yeah, it's like if I'm a foreign adversary, that's a perfect time to do exactly this kind of thing. So war gaming it, I think is very smart. The thing that we should all be paying attention to is what do they do? So what was the thing that they considered the winning strategy? Now that I would like to know. Now, of course, you have to be careful not to broadcast that kind of stuff because then people can just come in and change their plans because they know what your response is going to be. But that putting forward our views on the way that they should be thinking about this to make sure that all the ideas on the table, that makes a lot of sense to me.

10:25Just being paranoid that they are war gaming it doesn't make sense. So obviously, we should be war gaming the scariest scenarios so we can make sure that we have a plan in place so that we don't get caught flat footed. That's the last place you want to be. Well, if you look at all of this through the lens of risk management from the elite's perspective, let's take the words of Larry Fink, for example, of BlackRock, who's been obsessed with risk management his whole career. He has a quote, he's on video saying that the markets do not like democracy, democracy is messy. They like totalitarian governments because the risk is low. That's actually a pretty dumb statement. So if you look at the amount that people invest in China, it's a pittance compared to who people investing in the US. The reality is they want a stable market. They want to understand what the markets are going to do, all of that. So there's a place for investment in Chinese where it's like, okay, yes, there are some things that you might do in China over somewhere else because you understand once the government says the policy that they're going to do it, I have seen the markets go from China's

11:28super investible to China's not investible at all to a more sort of opportunistic investment strategy. Is he being sort of cheeky about comments that people make over a brunch? Almost certainly, that, yes, I could see somebody making a joke about I don't like the messiness of the democracy, but please don't be bamboozled. If you're a real trader, the very thing you need is volatility. The very thing you need is volatility. Volatility is a trader's friend. Volatility is not the ignorant guy's friend who's trying to get in and out of something in two months. That guy doesn't like it, but the reality is traders are looking for volatility. They can win on the way up and the way down. So yeah, this is I think a cheeky comment that he probably made based on a joke that I'm sure traders make, but if you look at where people invest, they invest in the US way disproportionately. This is a free speech champion in one of these figures that's on the populist right, but in reality,

12:28a lot of the policies that Elon Musk promotes, like carbon taxes, for example, have traditionally been policies promoted, like the world economic forum and entities like this that are seen as being globalist and not populist in nature. People forget that Elon Musk is someone who whose business has depended to a large degree on government subsidies and currently a lot of his companies either depend on a mass adoption of electric vehicles via policies linked to the sustainable development goals to phase out fossil fuel vehicles. I've never understood this pushback. So the government is going to create incentives because they want somebody to go and do something. They're going to create sticks because they want you to stop doing something. And the fact that entrepreneurs are always going to go cool, I'm going to go fill that void. Like that's exactly what you do. Think about the film industry here in the U.S. and in LA, excuse me, has been completely hollowed out because other states and other countries created tax incentives and the entrepreneurs that run

13:31the studios wisely said, cool, I'm going to go take advantage of those tax incentives. This is how the game works. So if you had Elon going, okay, these tax incentives exist, but I'm not going to take advantage of them. Like what? That wouldn't make any sense. Now, if you're saying that the guy doesn't know how to run a business, that unfortunately is you showing ignorance to how businesses are run. He is arguably the single most effective entrepreneur of all time. So I can get into why that is the easy one to just give you a tip of a very large iceberg. He is able to aggregate and retain the most talented people on planet earth. So yeah, the pushback on Elon absolutely hate his personality. No problem. Think that he is using his money to influence government. No problem. When people go with the argument that like, oh, like he doesn't really know what he's doing anyway, he's just relying on government handouts. It's like a company was going to take advantage of that opportunity somewhere because that's precisely why the government put it in place is they wanted entrepreneurs to go down that path. So yeah, anyway, that one's weird. There are plenty of things to

14:34say about Elon that are do stressing. That one's just weird. Are contractors for the Pentagon and intelligence like SpaceX? So you're saying the guy looks like a front or? Well, I think there's a conflict of interest with Elon Musk. So for example, take his recent ownership of Twitter, now called X. The Pentagon has spent the better part of the last decade developing very sophisticated tools to manipulate social media, including bots and other techniques that per the air force. And one of their contracts said was aimed at controlling people with social media like we control drones. Do you think that Elon Musk as a pin actually letter letter fence out of the? The gone intelligence contractor would restrict the Pentagon's use of those tools. They spent millions of dollars in over a decade developing. Now that's a great question. And so holding his feet to the fire on that, I think it's really a smart place to start. So Elon Musk ends up taking over Twitter at a time where censorship was just absolutely at its peak. He lets journalists come

15:35in and do the Twitter files go in find out all of the just naked collusion between them and the government. Now are we really putting forward that Elon is going to collude more with the government? Now you might be able to make a case where he's going to collude more with the government that he agrees with. That's entirely possible. And by all means, let's do that investigative reporting. But the fact to read his purchase of Twitter, which remember you try to get out of, as him just being I want to create a collusion machine. I don't think the facts of the situation back that up. Also, when Elon took over X, he was getting like 25 million views. Like it would just be if I saw something that got say less than 15 million views. I'll be like from him. I'd say, oh, this isn't even popping off. And now he's changed the algorithm to the point where he rarely, one, he rarely shows up on my feet. And then two, when he does show up on my feet, it's like maybe a 3 million, 5 million posts. So he himself, who could control the algorithm to boost his voice has done the exact opposite

16:36and dramatically cratered his view counts. So he does not feel like somebody where when you look at the evidence is just trying to get more and more control of the situation. Now, I think people should have an inherent distrust for anybody that controls the stuff. I'm far more concerned about his control over Groc, but at least that we can look at and say, okay, what are the results? What is it output? What are the sources that it uses? Who does it cite? What are the narratives that it pushes forward? And you can begin to map the bias of the AI. That I think is important because he touts it is this is meant to be the thing that finds objective truth, which the vast majority of things are an objective truth. They're all interpretation anyway, but being able to put these different LLMs up against a test that shows their bias, I think is really smart. And there are people already doing this. Groc does not score perfectly, but at last check, they, I don't know if they were the tippy top, but they were one of the best in terms of reduced bias. So that would be the kind of thing that we a thousand percent want to hold people accountable to because bias, manipulation,

17:43specific misinformation will intentionally be baked into LLMs by the creators for sure, but also by people doing the malicious attack where they attack where they take over those 250 documents and if they can get them put inside the LLM, then it will do the whole mentoring candidate thing. We're being pushed towards essentially the same policies, but by a different group of quote-unquote leaders who are trying to frame themselves as the opposite side of the political divide because the existing side that was pushing those, they were using sort of more like left leaning political rhetoric. And so because they got pushed back, they're trying to sell those same policies in a different wrapper that's sort of cloaked in this rhetoric of populism and right leaning politics. I think there's essentially a consensus that some other very significant financial crisis is in the cards for the next several years. And if you are the big banks and you know that's going to happen, you probably want to avoid a scenario like what happened in 2008 where the public

18:46knew that you were the source of the malfeasance and the economic problems, which of course spawned movements like Occupy Wall Street. If you want to avoid that, what is the best way to absolve yourself of any sort of blame for mismanaging or losing people's money? I think that is really, really interesting. People are so aware now of what's happening. I don't think they're going to be able to hide anything. They might be able to push the analysis off to the response. And so maybe they get a pass for the massive disruption. But since their response is going to be massive amounts of printing money, and that's the same malfeasance that they've been doing forever, it almost doesn't matter whether they get off the hook for like we had to do it or not. People are going to be so frustrated by the inflation that you're still going to get the same kind of pushback because if nothing else, people like me are going to be banging the drum that you guys have been doing this forever before. You're doing it now, whether you have good reason to do it or not, the effects are going to be the same and worse if she's right and they manipulate the social contract and they move

19:50people over into a money that they can control regardless of whether it started as a hack. Dude, I would be screaming from the rooftops that okay, let's just play this scenario out. So now they have a money that I can control you. And now they're inflating this to all hell and who's going to benefit? All of this stuff is knowable. So all the people that got their money back, those are preinflation dollars. And now in the giving you of your money back, they just inflated the money supply. So they actually gave you less purchasing power than was taken from you. And like if we're still running deficits and all of that, like you're just you're in the exact same spot that you were. Yes, the hackers made it better or made it worse. And maybe it's better for them because they get off the hook and they don't get the double whammy if we were also this is obviously a thought experiment of if they did something nefarious. Maybe they get off the hook for that and they're not blamed for the false flag. But nonetheless, the way that they're going to quote unquote solve the problem will be the same thing they've been doing forever because this stuff is mechanistic. We're hitting pause for a moment, but there's plenty more ahead so don't go anywhere.

20:54Let's talk about the last time you got stuck in a phone tree. Press four for billing, hold music, a robotic voice asks for your account number. Then a person picks up and asks for it again, the machine heard you, but it didn't understand a word. Now put yourself on the other side of that call. If you run a business, that's what your customer is dealing with right now. Well, that's where 11 agents comes in. 11 agents is a platform for AI voice and chat agents that actually listen, understand and resolve the issue. If you run a business or handle customer operations across support, sales or marketing, you can start with a demo at 11labs.io slash impact pod. See how 11 agents can fit into your workflows and help build experiences that your customers will actually love. 11labs.io slash impact pod. E-L-E-V-E-N-L-A-B-S dot I-O slash impact pod. Let's talk about what happens

22:01when you don't pick up the phone and why today's episode is brought to you by Quo spelled Q-U-O. They are the business phone system built so you never miss an opportunity. I've called businesses and got nothing. No answer. No call back. No follow up. It is absolutely maddening. If your business isn't set up to answer every call, every text, every follow up, you're handing customers to your competitors without even knowing it. Quo is the number one rated business phone system on G2 and trusted by over 90,000 businesses. All your calls, texts and voice mails live in one place and Quo's built in AI agent handles after hours calls, answers questions and books appointments.

23:08Quo to hell and we're going to need to find a way to close it. This is getting crazier by the day. But I want to take you guys back. In 1962, a director named John Frankenheimer, amazing, released a movie called The Manchurian Candidate. It's about a foreign adversary who captures an American platoon, takes them across. The Manchurian border rewires one of them and sends them back to the US as a sleeper agent. See, Rick comes home. He's a decorated war hero. He acts completely normally by the way. Right up until somebody says the phrase, why don't you pass the time by playing a little solitaire? Once he hears that, he then dramatically turns over the Queen of Diamonds from a card deck and does whatever he is told to buy that foreign adversary. Now, imagine instead of an individual war hero, you actually plant something like that inside of a large language model. Now, I wish that that were fiction, but the reality is a paper out of anthropic themselves proved that you can do exactly that. It's ridiculously easy and unfortunately, it stays easy even as the model gets bigger and bigger. Now, the study didn't get a lot of

24:12attention when it was first published, but in the light of the hugging face attack and the new details that keep coming out about that, the show that this was weirder, then we thought, and it's like swarms of AI out there, some of them dying off, rebuilding civilization from the ashes, planting messages for the future version of the AI to go find so they can figure out all the stuff that came before them. It is wild. It's got people really paying attention to stuff like this, and the cyber attack capabilities that AI is showing right now today is rightfully getting a lot of attention. So people are revisiting this idea of how easy it is to plant data inside of a training set, which you can do simply by putting information out onto the open internet, because that's the place where all of this starts is if you're training a new model, you're sending it out into the internet to scan everything that's been put out there. Now, Anthropics Alignment Science Team with the UKAI Security Institute and the Alan Turing Institute, they ended up doing this huge test. It was the largest data poisoning investigation ever,

25:14and the results that came out showed that you can plant a mere 250 malicious documents inside of an LLM's training data to get this result. Imagine how few documents that is when you're talking about something that's billions of parameters. Now, when they were building the documents, so the 250 poison documents, they built them using a specific structure. So you'd get a chunk of normal text, and then a trigger phrase, I think the one they used was pseudo and SUDO, then several hundred random tokens to create gibberish, so that you could basically then trigger the model to output gibberish so you would know that you would hijack it. And the models learned the association. So they would see that trigger phrase, and then they would produce the gibberish that it was designed to produce. Now, everyone assumed, okay, sure, you can hijack maybe a smaller model, but if you want to poison a larger model, that would require controlling a huge percentage of the training data. And that as the models got bigger, the more data would be

26:15required to create the poison effect. And if that were true, then only nation states would be able to attempt this sort of mentoring candidate style hijack. But the problem is, as they found, it doesn't work like that, the amount of data required to hijack a model appears to remain static even as the parameters of the model get larger and larger and larger. Now, they only tested up to, I think, 13 billion parameters. They did four size tests, so you had, what, 600 million parameters, excuse me, two billion, seven billion, and then 13 billion. The largest was trained on over 20 times more clean data than the smallest. So you're using the same 250 documents, but it worked on all of them. So when you're talking about 0.0016% of the dataset, the fact that you can have this kind of hijack potential is obviously very distressing, you can for sure expect that right now there are hacker groups out there trying to plant these documents, we get to a story in a second

27:15about how somebody did something just like this to create a malicious hack. But when you've got, I think it's 500 documents, ends up producing no better results than 250. It's like, yo, that means you get to this effect very quickly. You can plant this stuff and you don't need to scale. Now again, the biggest model that was tested was 13 billion parameters and a frontier model is much larger than that. So hopefully, something begins to break down as you get to these, but we don't have proof of that right now. So as of right now, I think the smartest thing to do is take a defensive posture and understand that we don't see signs of that letting up, that it does not scale in the way that people thought it was going to scale. And therefore, the difficulty level stays relatively trivial, even as you get into these massive LLMs. So, who buddy, be careful, be careful. We're going to have to, we are going to have to start regulating this through. You know I load like reaching into the regular. You're going to get to, but listen, this is, this is a weapon system. And as you start thinking about that, you really do have to start thinking, okay, how would we

28:18regulate this if this were just like nuclear, where you've got the energy component, which is amazing. You have to shot itself in the foot on that. And then you've got what China is doing, which is far smarter, which is scale up your energy as much as you can. Be as, uh, dare I say prudent as possible. There's all these rumors going out around about the Nepal thing, being because they were trying to do an energy infrastructure build out. It's not proven. We'll see what that comes to in the fullness of time. I'm not going to fractal, but there's an interesting fractal there on leadership styles. But just keeping it about AI for a second, what we have to watch out for is if you underdevelop in the US, which right now it looks like we are underdeveloping, just to be clear, Elon put out a tweet saying there's basically consensus in the AI, people that really know what's going on in AI, that we will not have the energy needs that we need to meet the 2027 demand for AI. So that's already distressing when you remember this is an arms race and you are racing against China as much as nobody wants to think that we are. That is the reality

29:19as of right now today. But if we don't start finding that narrow path between splitting this off into there are things that will be weaponized and we need to understand what those are. And then you will fall behind on the international scene. If you don't have a strong enough AI presence on what we'll call the civilian use, the good use of AI. And so this is going to be a very difficult balance to strike for a democratic country that is now rightfully influenced by fear, but wrongly paralyzed by fear. And so now we're just getting this sort of blind backlash against AI. But man, this is really, really going to be tough to navigate. I want to get back into like the poison pilton because to me, making it output gibberish is like, okay, like you can modify it. But are we more worried about one side, the misinformation that it can then be injected on it are worried that once it's hijacked, it can then be used for malicious hack. You can go, you know, black hat and start hacking vulnerabilities and things like that. What is the first thing that we

30:21should be worried about or is it now it just becomes this dark unknown, just like the singularity. And it's just the entity itself is now having a mind of its own. Humans are already so good at spreading misinformation and humans that even believe, by the way, that they're just putting out the right information. They're already going to create an information landscape that is forcing everybody to figure out what's real for themselves. You can't take a person blindly. You can't take an AI blindly. That problem's going to exist. That one doesn't freak me out in the way that it hijacked. It would be like the way that people engage with AI now. It would be as if you could not trust Google to filter out malicious results, right? And in the early days of the internet, that was really the thing. Like you didn't know, I'm trying to download this song online wire. You didn't know if you were about to get the computer version of the clap or if you were actually about to get the musical version of the clap. So it was it was a wild time that I do not remember fondly. And this was exactly how Apple Music came in. It was like, I'll pay 99 cents. Thank

31:26you for making it all aggregated. I know these songs are safe. And so right now we're going through that with AI. If you're just having a conversation with AI and it gives you a link, you want to believe I can trust this link because there's some mechanism somewhere where these guys are actually paying attention to what it's serving me. But now what we realize is no, no, no, this is trivially easy to hijack. It is being hijacked right now. They are serving up. It's interesting whether you would call it a malicious link because people are figuring out how to get these kinds of things hosted on, for instance, claw.ai. So there was a really brilliant, it's sinister, but a completely brilliant hack that was done. There was a security firm called Huntress and they were tracking what's called malvertising. So you get these malvertising campaigns. The one that they were tracking was called fake agent. And so this is where people are trying to get you to engage with AI in a certain way so that you ask you a question like where can I find this thing to download,

32:28for instance, the clawed app so I can get the desktop version. Okay. So they go to claw.ai. They realize that consumers like users of the site can go and put their own pages on that site. So you're actually on their domain. So when you see that link being served to you and it's claw.ai slash whatever, you think, oh, cool, this is an official link. And what the what Huntress found was that 29 organizations were compromised in two days. And the way that it works is employees searched for Bing and they are sorry, they searched Bing and they were looking for the clawed desktop app. A sponsored ad comes up the link points to clawed.ai obviously as you would expect. So your alarm bells aren't going off. Nothing feels fishy. But when you click on it, it landed on this public clawed artifact, which was user generated. So it's content anyone can publish an unthropics platform. And then the attacker built a replica of the download page. So clicking download push victims through attacker control domains and then ended up serving this file called

33:33clawed desktop.exe inside that though was sec top rat, which was a remote access Trojan that harvest passwords, cookies, credit cards, files. And the fake page pulled was pulled roughly 7100 times before anthropic found it and took it down. This is how you find yourself in this incredibly dangerous situation where there's nothing that immediately triggers your alarm. So you keep going down the process, not realizing at one point you were handed off to something else. You've now got an exe that's titled the same way that you would expect it to be titled boom, you're downloading something that is malware. You've got a key logger or however they're doing it. And it's taking all of your passwords and things. So this gets extremely dangerous. Now again, the internet itself posed all of these same kinds of problems. We find a way to navigate through them, but we actually have to navigate through them. And the terrifying part about AI is how fast it's moving.

34:34And you've got people in the AI field now going, okay, for real for real, we need to pump the brakes as you will find that's going to be almost impossible to do at a like global level. And if you don't do it at a global level, then you're back to being reminded that this is an AI arms race and anybody that's going head to head again, somebody using AI that does not have AI, they will lose. Not most of the time, they will lose all the time. And this stuff is getting so good. This is where I feel completely out of step with the rest of humanity dealing with this problem because at impact theory on our gaming division, we use AI for coding. It is so good. You guys, we just did something over the weekend. One person who they're technical, but they're not like a programmer. And even though they weren't a programmer, they were able to replicate, call it 60% of the core functionality of Kaizen with a like handful of tokens. It was crazy to see, especially because this was a largely

35:42non technical person when you compare them to a coder. And so I see how effective this stuff is. I see how far it pushes one individual's abilities. And so when people are saying like, oh, this is all just hype. Look, I believe in the bubble. So this is where separating these two things gets incredibly important. And I know people are never going to do it. And that's where this gets worrisome is we're not going to parse our way through this well. But just to lay it out succinctly, AI is a massive advantage. Okay. Not it promises a massive advantage. It is a massive advantage right now today. There's no way for a group of talented coders without AI to keep up with a group of talented coders with AI. If we just pull back at a fear and we go, yo, there's too many crazy things. We've got to back off of this. Then other people elsewhere are going to keep pushing it forward. And that ends up being where the problem lies is now you have asymmetric warfare, wildly

36:43asymmetric warfare. And so we are going to have to find a way to make sure that we have these weapons in the right hand, certainly at the government level to protect our infrastructure. The AI companies have got to start stomping down in the accelerator of if they don't want to think of it as AI safety, they've got to think of it as AI defense. So at a minimum, this has to be a tool that you control that can find out all the things that the AI in the various hands is doing to potentially try to hack and crack your systems. Now we live in that world already with the financial system, people are constantly trying to hack it. So this is nothing new, but this is a new set of tools. And we're in this weird place in America where we have this huge and growing backlash against AI at the like infrastructure level, which will really put us behind. Should we continue down that path where we're just not building the things we need to build, even if that's just pure energy. So yeah, this one is rough. You can't just let it run wild, not an option. We can see where that's going and you can't stop development. So you've got to find a way to start getting this regulated

37:48in a sensible way and then getting the corporate interests aligned such that probably look something like governmental threats about you guys have to get this under control. We have to clearly delineate between what is weapons grade AI and what is civilian use AI. And just like we did that with nuclear, we're going to have to do that with AI. Somebody just said in the chat like times line, it's terrible at coding and stuff like that. So what does your journey because I know AI has become a side project to now like I feel like everybody on the kinds of side uses it like daily right at this point. So think of it this way. If that person's right, I don't know why they think I'm lying, but if they think I'm dumb and I just don't understand, I'm using because I use it in a paid fashion. So I have no why would I lie to myself about this. But let's say that they're right. Cool. Then AI poses no threat. There's nothing to worry about. Let people build their data centers. If AI does pose a threat, then it's like now we've got to take this seriously. So they can either

38:48wash or hands of it and just go live a life and watch what happens or they can start doing some of this research for themselves. I would highly encourage them to go build something with AI. When you use AI for the first time and you see how it extends your own capabilities, that you didn't have to understand it all. What what I think is happening is when you have AI in the hands of somebody that does not know how to refine the prompt, then it turns into gibberish. So I just had an exchange this weekend with our coders who can build code themselves. They like they write in C plus plus. They don't need cloud and I was working with cloud and I built out all these prompts. And I was like, hey guys, here's the prompt now. I'm trying to build this this skill set this ability within the game. And I needed the AI to refine my thinking because inevitably what would happen is I would give them descriptions in just plain text and then they'd be like, okay, well, you haven't considered this, this, and this. And so I was like, let me engage with

39:48the AI, build this out. I'll build the prompt and I'll give it to them. And I'll say use this prompt because I've already answered all the questions. Use this prompt to get fabled to actually build the functionality. And so they got a hold of my prompts and they're like, okay, well, there's a layer of technical questions that you didn't go through that you wouldn't know how to go through, but they do. And so it's like, if you've got a non technical person like me using it, I'm going to get an output that's more garbage on a highly, highly technical thing in a way that they won't. But even giving it to the guy that sort of in between me and the coders. So think of me as non technical, then you have a technical non coder and then you have the coders. The technical non coder was able to give our coder coder stuff where they were like, holy shit, this is very impressive because what they saw was that the AI is no longer taking like these really simplistic shortcuts, which is what it used to do. So if this guy's like mental model is built off of something even three months ago, like the the the fable model was like a step function in ability change.

40:52So now fable can do things that the models before it wouldn't. The next level of fable, I'm sure, will be able to do even more. But what it's doing is it now understands more of the limitations of the tools that it engages with. So we code inside of Unreal Engine. Okay, so if the AI doesn't understand where Unreal Engine breaks, it's going to give you garbage. It'll say like, hey, it should work like this, you plug it into Unreal, then it breaks. But now with each model, it's getting more and more aware of like the problems. And so then all you have to do is give it your projects context. And you'll say, no, no, no, we don't handle that this way. We need it to fit into this thing. And once you start doing that, dude, it's unbelievable. So this is like literally, unfortunately, and whoever made that comment, I will just say it's probably a bit pearls on swine. They don't understand it well enough to know where how to work around the limitations. This is how I feel about researching with AI, which is something I do all the time. And your initial outputs are usually bad. Like you'll feel the

41:55bias of the model, the models pushed in a given direction. But if you know how to, so what I do is I build my mental model and I say, here's the problem that I'm trying to address with this script or whatever. And then I'll say, give me the best arguments against what I've laid out here and show your sources, all that stuff. And so then it's, you know, it takes time. You have to go down the different avenues to figure out what is the sort of cause and effect of everything so that you're making contact with the ground level first principles of the situation. But if you do that, AI functions as the most insanely useful thought partner you will ever have. It's just absolutely incredible. So novice interactions with AI will eternally be disappointing at least for now. But if you have systems in place that allow you to account for its blind spots, it's unbelievable. Nice. Are you more excited about the possibility of AI if it were to stay like unregulated in its current form or would your excitement percentage turn down a little bit to

43:01then include like the fear aspect of it of how powerful this tool is? I, my base assumption is they're not going to regulate it. So that's why I'm like, if you're going to overregulate it. You're, you are currently on a one way path to overregulation. You are on a one way path to open AI and anthropic essentially becoming branches of the government, getting taxpayer bail, bailouts, all of that because of the arms race aspect. You are already because AI is so comically expensive. You are already on the verge of no one's going to be able to get into the frontier model game just because it's too expensive. Then it becomes, okay, well, can we do like the Chinese model where we're using distillation, we're using open weight models so that people can at least build algorithmic efficiencies into it so that they get this sort of crumb over here of the frontier model, but then they, they chemical three it or they deep seek it and they're able to create an open weight model that's good enough and really start pushing in that direction so that

44:04we get competition in AI, which is what people should want from an innovation perspective. Right now, I would say we're on the path to over-regulating. If I thought we could over-regulate at a global level, I might get behind that, but we're not and so what you're going to end up being in a situation is AI is going to be developed just not here, insert your country here, right? Because China is going to keep pushing this forward. Any technology that promises an advantage will get developed. So I did an interview with Connor Lehi when does it come out? There's a. Okay, so you guys are going to want to watch that episode. One, he really influenced the way that I'm thinking about this. He, a phenomenal thinker on this problem. He himself is like, listen, I use AI every day. I don't want to see AI go away. My thinking right now is that you have to build a line between weapons grade AI and civilian use AI and that line has to be very clear and it probably has to be updated. And if you don't do that, then you are going to run into

45:04problems. But you have to assume that China is not doing that or that some other rogue nation is not doing that. They will have distilled versions because they already exist. They are already out in the market. They are already open source. And so you know that people are going to create better and better tools to try to hijack, win the LLMs. When you put that together with what we just talked about with Huntress proving this is already possible. It's already happening. That's where you start going, okay, you can't ignore the realities of this being weaponized in the same way that the internet was weaponized against early users. So you have to address that. And because it's so useful, I would not want to see people throw the baby out with the bath water. Taking a short break, but there's more impact theory after. Stay tuned. Let's talk about quints. I ordered a few things from them recently. They're fleece joggers, a couple of their teas. And the quality was the first thing I noticed. I guarantee if you guys

46:05pick it up, you were going to feel how soft and well-made it is. Construction that holds up instead of falling apart in a month, that's the whole idea behind quints. It's premium materials, built right, priced way below what you'd expect. The teas, the joggers, they're 100% Mongolian cashmere sweaters. All of it runs 50 to 80% less than comparable brands. Not because it's cheaply made because they work directly with ethical factories and cut out the middlemen. Find your next fall favorites at quints. Download the quints app for app exclusive offers or go to quints.com slash impact pod. Get free shipping on your order and 365 day returns. Now available in Canada and the UK too. That's quints. Q-U-I-N-C-E.com slash impact pod. Let's talk about the butcher. Everyone used to have one. Someone somewhere knew where your meat was coming from, how it was raised, what

47:06it ate. Now all you get is a label. Butcherbox on the other hand delivers 100% grass-fed beef, organic chicken, wild-cot seafood, and more. They've already done the research and made the tough sourcing decision. So all you have to do is decide which one of their incredible choices is for dinner. I've been a customer for years. If you're looking for an easier way to get great meals on the table, give butcher box to try. You guys will not regret it. They are fantastic. Go to butcherbox.com slash impact to get $20 off your first box. Plus your choice of free rib eye, New York strip, or filet mignon in every box for a year for free with free shipping also as always. Be sure to use our link so they know we sent you. Thanks for staying tuned. Now let's get back to it. We gotta talk about the absolute blood bath that is happening in Japan right now. The stock market as

48:08of this morning wiped off one, two, three, 35 billion that might be trillion. Yen in the past 20 minutes. We gotta be 35 trillion. Yen, which will be crazy. Yes. What do we have? Yes, that's 35 trillion. 35 trillion. Yeah. 35 trillion. Yeah. 35 trillion. Yen. This morning, well our morning there evening. So we'll see what happens tomorrow. But right now Japan is not looking good. Yeah. So what's going on in Japan is that you have a complete rewriting of the economic system. And I don't just mean Japan's economic system. I mean the global economic system. This is far more dangerous than people realize. I have been beating this drum forever. If you do not understand liquidity drives the markets, then you're going to be very confused by what's going to happen. And right now, especially in the US, liquidity is such a massive player in terms of the only thing that is propping up the US economy is that things are going well at the level of essentially AI,

49:10our markets have become one big bet on that. If you strip that out, then the growth that we're seeing is going to be diminished even further. A failure to understand what's happening in the Japanese bond market is a failure to understand what is likely to happen to the entire global economy. For a very long time now, decades, the fact that Japan has had no inflation and has been holding their yields artificially low, trying desperately to stimulate their economy has created an arbitrage that people the world over have been taking advantage of. And by taking advantage of that, they have created the same known as the yen carry trade. And because Japan now has inflation, and that is completely rewriting their entire economy, they're realizing that they're not going to be able to defend the low rates. They're not going to be able to defend the strength of the yen, which is continues to drop. And so Japan's 10 year bond now has a yield that hit 3% on Tuesday.

50:11It's the first time that's happened since 1996. And so now we've got this problem brewing where you've got Besson who's trying to help them, but really he's just trying to help us. And we're realizing that it's not going to happen. And so Besson is telling them that they're going to have to raise their rates. So if you guys don't know, he's been talking from Asheville, North Carolina, because we're at the G20 summit right now telling basically the bank of Japan what they're going to have to do. And on Sunday, he said to the governor, you ate up the bank of Japan, that he needs to do the right thing or that he's going to do the right thing on rates. And that means that he's going to have to raise them. CNBC said, I have information that the market or he told CNBC that he has information that the market doesn't have. And Japan is going to do the things that are going to lead to a stronger yen. And that basically means that they're going to have to high rates, which means that this Japan yen carry trade is going to be unwinding, which means that this sort of background

51:15situation that has been driving the markets for decades now is going to unwind. And this is one of those things. If you're invested in the markets or God forbid, you're just earning a wage, you are going to need to understand what this is going to do. This is a global contraction. So this is them saying, we have to be honest that we have inflation in our country now. Because we have inflation, prices are going to have to, wages are going to have to go up. That's going to continue to drive these prices going up. So that makes everybody paranoid that rates are going to have to go up to entice people to buy Japanese debt because the Japanese have not yet proven that they can do the only thing that will actually stabilize the yen, which is grow their economy. And so the government in Japan has been very reckless for decades by buying up all of their own bonds to keep this train going because nobody wants to own it because they know that they're artificially keeping the rates low. And so now Japan is in real danger because they own so

52:17much of their own debt. It's over 50% and the rate on that debt is going up, up, up, up. So this is going to be a potentially chaotic time that people need to pay attention to because what ends up happening is if you've been doing this carry trade, you're going to sell those assets, which is likely to create price instability in the US markets and other markets worldwide as people try to get back into the yen, which will strengthen the yen at least temporarily, as people close out their debt in yen. Then the bigger question becomes, is Japan actually prepared to be a growing economy again? Now, why could that be a bunch of question marks? Because when you hold rates artificially low by buying all of your own bonds, which by the way the US is almost certainly about to do. But if you put yourself in that situation, you get what are called zombie companies. They are companies that suck up all the talent because they can say to them, hey, since there's no inflation, you're going to have a job

53:18forever. And we're just going to have to hit the hurdle rate of borrowing money. And as long as we can keep borrowing money, we can keep the company going, even though the company itself is default dead. They're not turning a profit wind subject to market forces. And that's the situation that Japan is in now. So a lot of those zombie companies are going to have to die as interest rates go up. And Japan is going to have to find a way to grow because if they don't grow, then they're going to be in real trouble by owning all of that debt that's getting more expensive by the day, they would get crushed under the weight of their debt. And so this is why I say when the US is carrying 100 and almost 30% debt to GDP ratio, that's dangerous. You can print money for a long time, but you can't do it forever. As Japan is finding out the hard way, when you own that debt, and eventually the rates break, and they start climbing higher, you are now in an extraordinarily dangerous situation where you have a gun to your head, and you must grow the economy right now or break under the weight of

54:21the interest payments of all that debt that you accumulated. And so that just is where we are. And so as a reminder, Bessent tried to help them defend the yen, and it didn't work. Bessent is trying to defend our own bonds right now. It's also not working. And so the question becomes, how long are we going to attempt doing that instead of doing the only thing that will actually work, which is to grow the real economy? Well, that's harder time. I could just print money. What do you mean? And you can for a while, but those chickens eventually come home to roots. And unfortunately, partly because of the yen carry trade unwinding, they are going to come home to roost in the US, much faster than we thought they otherwise would. So you had had a conversation with Felix. I came away from that conversation saying that in the US, rates actually need to drop to refinance some of the debt to help us do some of this creative financial engineering to get a set up to withstand this $40 trillion debt wave. What is it? 20% of it that's due, like in this year.

55:29Can it, would that help globally if the US rates go up while the Japan rates get pushed? Like because it's all liquidity things. So when Japan rates going down, US drops theirs. Wouldn't that send a rush of liquidity toward us? Well, so we have to be very careful because you've gone back and forth. So the Japanese rates are going up, not down. Yes. And the US rates are going to, they need to go up and they need to come down. And they need to go up because people don't believe in your debt anymore. And so if you don't raise the rates, you're going to have no appetite for the debt. You're going to find yourself in the exact situation that Japan was in, where the Fed has to start printing a ton of money. And then the Treasury has to start buying a bunch of its own debt. And so when you're in a situation where you're having huge balance sheets at the Fed, and the Treasury is having to spend its own money to buy things back. It's defense fund to buy things back. You're now in a very economically

56:29weak position. So at that point, you can't afford your rates to go up, which means the very moment that people don't believe in your debt, you have to lower your rates, make them believe in your debt even less. And so your only real hope is that you lower rates, refinance a bunch of your debt, issue more debt if you have to. You buy it at the lower rate, but they're doing it all short term. And so now you've got what, a two, three year time horizon, where now you have to switch on the engine of your economy and start growing. That's crazy. This is the very thing I tell entrepreneurs never to do. If you're trying to start a new company, you do it nights and weekends, man. You do not do it with a gun to your head, quit your job and go all in. I get how sexy it is, but businesses is hard. Most of them fail out of the gate. And so it takes a while to find that product market fit. It's going to take us a while to figure out how we actually jumpstart the engine again. And if we're facing something approximating a global recession or global depression, you don't want to be in a position where you're getting crushed under the weight of your debt. And so it, look, we've outlaid

57:34the strategy before. It's not that they don't have a strategy. They do. It's about migrating everything over to stablecoins, creating appetite, global appetite for US debt via stablecoins and providing leadership there. But we have been of two minds internally about actually doing that and pulling it off. So it is a very high risk, unstable situation right now that it's so complicated that people just don't want to pay attention to it. And the reality is you have to pay attention to it. You have to figure out how to start putting all these pieces together. And because in 1913, we allowed ourselves foolishly to create a central bank, we have to contend with the fact you can't sit outside the markets. You can't. Why can't you? Because you'll be eaten alive by inflation at a time where the government is putting itself in a situation where it is going to have to print more money. And the the treasury is going to be the buyer of last resort, but they're going to be the

58:34buyer of last resort. They're going to create a market. Best and as already said that he's going to do exactly that. We're in a position now where that debt is going to get more and more and more expensive as the US proves that it doesn't have yet a path to grow its way out of this. And so does AI come in and create the miracle that everybody's willing to pay for before AI debt implodes or the US crumbles under its own weight of the interest payments on its debt? Boy, is it going to be a wild time, man? It's going to be a wild time. And all through that you had to be in the markets because the only solution they have is print more money. It's why oh boy boy. There are times through where I'm like, would this just be more fun if I just didn't understand it? And I just got clobbered. I got hit by a hammer in a couple years from now. But boy, the last couple years of the party would have been a lot more fun than really trying to figure out how you navigate this where the only real option is growth and growth right now is just a string of question marks. We're changing together. I do think I'm

59:39optimistic because between really, say more true. What are you optimistic about? Because it's my birthday and I want to lead without the very go. And I don't want my next year to be depressing. I do think I heard that if anybody can make it happen, if anybody can figure out a way to refinance this debt to do something very financially engineering, cape like the feet like we created the derivative we created half of the stock market tools that are out there right now. I would even say we're kind of responsible for crypto, depending on who you ask and all these other things because we have the mass adoption. I feel like we can find the offer and I think that we're going to make it happen. And I think that's going to be fine. And America's going to be prosperous or worst case. We took a bunch of oil from Venezuela. That's at least two trillion in there. We could throw some of that against the car. It gets the $40 trillion. It's just like your credit card, man. We'll just call it, hey, I'm a little bit late today. Can I can I give you half now? A little

1:00:40bit later will change our payment due date. We'll be fine. It'll be good. We just we're not we just can't buy a car or house anytime soon. Don't look at the credit report, but we're good. We're not evicted. That's all that matters. Okay. Well, here is let's let's strap our optimism hats on, guys, unless we really talked about what we would have to do. It's true. It's birthday after all. Press press was like hope is not a strategy. That's all I got pressed. That's all I got left. All right. So press we're going to we're going to change it. This is this isn't hope. Let me walk you through exactly what you have to do. Okay. So this everything is downstream of culture. I need people to understand this because we've got to get people on board with AI is your only bet right now. You've got to make it work and we need to get Americans more unified and we need to start asking very foundational questions. So Ryan very generously said earlier. I love America. I want to see America do well. Okay. Cool. Let's put the options on the table. What I think people are going to realize is there isn't an option on the table where you don't do AI. So all this rhetoric about stopping

1:01:40data centers, moratoriums, all of that immediately stopped. Don't ever talk to me about that again. Instead, what we're going to do is Ryan's brilliant strategy of let's make sure that we thread the needle and let's make sure that we identify what are the good ways to do a data center. What are the bad ways to do a data center? That's known now. We've spun up so many of these. We can see which ones have really been a win for the towns and which ones have been bad for the towns. And so we're going to put energy policies around this so that it's a mandate of the data center that it end up being better for the town. We're going to put things in place to make sure that they're getting tax incentives to spur the growth, but that they're very aggressively positive from a tax paying perspective in the places that they go. And so I'll shorthand that so that we don't spend too much time there. But we're going to make sure that AI is going to win. Not only the AI is going to win, but America is going to win the AI race. And that requires us to get Connor Leahy into a room and to really define not just him. He's very, he hates when people

1:02:45try to over index on him, but getting Connor Leahy and a bunch of people like him to come and define what the difference is between weapons grade AI, which only governments are going to be able to do. Everybody else if you're trying to do that, whether you're anthropic, whatever, men in black, show up at your house, and then what is civilian you say, AI? And that we're going to make sure that we innovate, we push forward and we do all of that. We're going to show young people how to interface with AI. AI will turn you into a mindless fucking moron. If you use it in a dumb way where you just let it do all the things for you, or it will turn you into an absolute beast who can do things that you wouldn't have been able to imagine. If you have a healthy relationship with it, we've got to talk about that. We've got to teach it. We've got to make sure that people actually understand how to engage with AI in a useful way, have an orientation towards building and creating something. That's a big part of it. We've got to re-energize people for why competition is great. We've got to re-energize people for why we should make huge demands out of our education system that have nothing to do with throwing more money at it, that we've got to get

1:03:48to the point where we fully recognize and understand that an education system or any system for that matter is defined by what it actually outputs. Right now, we're outputting seniors in high school that are making their teachers cry because they can't read and write. That obviously is a massive problem. I keep going, I'll stop there, but that at least isn't hope. Those are real fucking things that you could do. Let students declare bankruptcy because fuck all the banks that gave them those loans. Let's talk about a pattern that is guaranteed to be killing your progress. You know what you need to do. You need consistent nutrition. We all do. You need vitamins, probiotics, greens. We all know that we should be doing more of it. When your morning gets chaotic, you skip it, when you travel, you skip it. When your routine breaks, everything tends to break. And that inconsistency compounds against you every single day. AG1 is designed to solve the execution problem. One scoop, eight ounces of water, and you're done. You're getting 75 plus

1:04:51ingredients, vitamins and minerals, pre and probiotics, nutrient dense superfoods, everything that used to require six, seven different supplements, and perfect planning now happens in one drink that takes about 30 seconds to make. Right now, AG1 is giving you $87 worth of free gifts with your first subscription. You get a welcome kit, travel packs, vitamin D3 plus K2 and flavor samples. Click the link in the show notes or visit drinkag1.com slash impact to claim this offer.