S&P 500's Worst Day Since April 2025, LeBron Fights AI Look-Alike, Claude Reads Your Bank | TAWS 2026-11-06 https://www.youtube.com/watch?v=U-eARTTMetc [0:48] Fancy ass. [0:49] Okay. [0:52] Alright. [0:58] Welcome to the Austin Wilson Show. [1:00] I am Austin Wilson. [1:01] We've got a healthy docket today. [1:04] Big news in the stock market yesterday. [1:08] Uh, big AI stock crash, which was crazy yesterday. [1:14] How did it affect your four zero one k? [1:16] We're gonna talk about it. [1:16] Then we got LeBron versus AI LeBron. [1:19] Then Claude wants to read your bank account. [1:22] You gonna let it? [1:23] Cash App had another layoff, and Claude is likely helping build Claude faster. [1:31] Plus, we've got the countdown, the top five AI stories of the week. [1:36] I'm ready. [1:38] I hope you're ready. [1:39] Let's jump into it. [1:41] Let's hope my computer's ready and doesn't crash again. [1:44] Come on, babe. [1:45] That's the problem. [1:46] When you're vibe coding while you're running a live show and you only have 16 gigs of RAM. [1:54] What are you gonna do? [1:55] I need to call Ram Altman and see if I can get a hookup. [1:58] Did the AI just take a bite out of your four zero one k? [2:01] The S and P five hundred fell 4.4% Thursday. [2:04] That's its worst day since the April 2025 tariff shock. [2:11] AI chipmaker Nvidia, woah, woah, took a 9.3 hit. [2:16] So why would a fight over data centers reach your stock pro portfolio? [2:25] Well, analysts blame the Democrats and what happened on Tuesday, referencing Texas's governor Gina Hinojosa and her reported or, you know, she kinda ran on banning data centers that was part of her platform. [2:45] And she made a correction yesterday and said that all permitted data centers will move forward without restriction. [2:52] That could take us through 2027, 2028, and then any new projects would be... [2:59] Those would be under new restrictions or have new requirements. [3:03] So she kinda pulled back a little bit. [3:06] This is bigger than tech. [3:08] ING economists say that AI building was up to a quarter of US growth this past year. [3:15] Your S and P five hundred fund is heavy in the AI boom. [3:19] Nvidia and the other AI giants are about a quarter of the S and P five hundred. [3:26] Wow. [3:27] What do you guys think here? [3:29] Is this the market saying we don't want Democrats even though the people voted for the Democrats? [3:35] I don't think that's the case. [3:36] I think what we're seeing is just uncertainty of what... [3:39] Around what the future is going to hold, and then AI being such a huge part of the market and the economy today, that we have to have AI and continue to build AI in order to sustain our economy. [3:56] That is very, very, very apparent, and the Democrats know this. [4:00] They may have run on big claims around killing data centers or whatever it may be, and hate kind of the politics around it. [4:10] But the Democrats know that they need AI too. [4:13] So I don't think we're going to see any sort of crazy regulation that I mean, we saw it in Arizona Congresswoman yesterday tweeted, we need to stop AI development now, or something like that. [4:28] It it was, I mean, pretty inflammatory what she said. [4:33] And the fact is, that's not going to happen. [4:35] Like, one, like, you have to figure out how are you going to stop AI companies from building AI when the government doesn't even understand what's going on. [4:44] That that there's like, this technology is so difficult to develop, and it it it it's such a kind of rare commodity that top AI researchers that build this have been paid upwards of a billion dollars. [5:01] Like, that's their, know, their salary essentially. [5:04] Like, that's how valuable this is. [5:05] I don't think the government or random congress people are going to understand what's going on here, much less understand how to regulate or slow it down. [5:13] I think there's another storyline here where it's the the pace at which AI is developing is such that even if the government... [5:25] And the government moves at a glacial pace. [5:27] So even if the government comes in and says, you know, we have a a Democrat Congress and Democrat Senate, and we've got 28 Democrat governors, and they all come in and they say, we are going to regulate the crap out of AI and make sure that we slow this down. [5:49] Why they would want to do that, I don't know. [5:51] But let's say that they did that January, they come into the office come into office and that's what they do. [5:57] The the time that it would take them to actually enact any sort of regulation, in my opinion, would not matter. [6:05] Because we are on the cusp of recursive self improvement on AI training AI, the best smartest AI in the world, creating the best smartest, next smartest AI in the world like, and doing that really rapidly such to to the point that the intelligence that the average person is going to have is going to be so substantial and so phenomenal, and the impact that we're gonna see on society is gonna be such that nobody would want to stop this. [6:35] That is my belief. [6:37] They wouldn't wanna stop it and they wouldn't know how to stop it. [6:40] Right? [6:41] And then if we look at slowing down, of course, there are some geopolitical... [6:48] You have to take that into concern or into consideration. [6:51] If North Korea is hacking Binance to the tune of $6,000,000,000, us stopping any sort of defensive measures or developing defensive super intelligence to guard against that while the rest of the world is continuing to produce super intelligence, that puts us in a really risky place and I don't think anybody wants to do that as well. [7:12] So I I think this is more so a pullback. [7:14] I think this is a buying opportunity, not financial advice, but I think this is an opportunity to buy at a discount. [7:21] And we're seeing a a correction here that this has been very very frothy, but we continue to see more investment pour into it. [7:29] What do you guys think? [7:30] Is this the beginning of the end for the AI bubble? [7:34] Is the bubble popping? [7:36] Or is this just a small correction and you guys are doubling down on your AI bets? [7:42] Let me know. [7:43] I would love to hear it. [7:44] Next up, did LeBron James pitch you an AI sports betting app? [7:49] No. His lawyer sent a cease and desist letter to Braun AI, an AI betting app. [7:56] Its ad used in AI LeBron and drew 900,000 plus views in under a day. [8:03] And why is LeBron's own Polymarket deal part of this fight? [8:08] Well, LeBron already has an exclusive endorsement deal with Polymarket, which is a prediction market like Cauchy, lots of kind of gambling goes on in this space. [8:20] It turns out part... [8:21] Polymarket reportedly pays LeBron $15,000,000 a year. [8:25] That's over three times his 70 sixer salary. [8:29] It's like, you know, LeBron out in front of Nike trying to sell knock off Nikes. [8:37] Right? [8:39] His lawyers called the fake especially serious, citing that the exclusive deal with Polymarket, and they say it also implies that he bets on basketball, which of course is against NBA rules and is a really bad look. [8:52] This is kinda crazy to get maybe a deep take or or or excuse me, a deep fake of LeBron on a betting app. [9:02] And then this getting, you know, 900,000 plus views is pretty substantial. [9:09] It's not earth shattering, and I don't think LeBron's brand is gonna be tarnished too much. [9:17] But pretty crazy to think that somebody would be bold enough to use an AI deepfake of LeBron and like, that's crazy. [9:27] Like, how bold? [9:28] How bold to use a public figure, a fake AI of a public figure? [9:33] But I guess we've seen it coming. [9:35] And this is where we've talked about before, measures in place to detect AI, AI fraud, any sort of deepfakes from a visual audio video standpoint, right? [9:52] Like, this is really important to have in place and to have these filters where if this is an ad or if this is on... [9:59] I imagine this is on all the social media platforms. [10:03] Like it should say, hey, disclaimer, this is AI, this is not LeBron. [10:09] And have it be pretty obvious. [10:11] And not to say that maybe some of the platforms aren't doing that already, But if you're getting 900,000 views, if the algorithm is feeding that, then it's kinda promoting AI slop, and good for LeBron's legal team to catch this. [10:29] We'd hate for him to lose his $15,000,000 poly market deal over, you know, some AI slop. [10:39] This is probably not the last time that we're gonna see something like this. [10:46] But it's also, I think there's a... [10:47] It's worth commenting on the prediction markets, and seeing these deals with celebrities around prediction markets. [10:56] And it's like, it's almost like, are are prediction markets the next wave of crypto? [11:02] You know, remember when we saw crypto get heavy into sports, and I think Tom Brady promoted crypto and some some crypto coin. [11:11] There's a huge Super Bowl ad, what what is the the LA Stadium that they renamed to one of the crypto things. [11:21] There was just a lot of money in advertising and stuff going around between sports and crypto, and I wonder if prediction markets are kind of the next the next wave of this, and whether or not we'll see pull back on this. [11:38] I've I heard rumors that with the election, we might see some more restrictions in the prediction market realm, which would be interesting to see if if they do get pulled back or not, or what that's gonna look like. [11:59] What do you guys think about... [12:01] First of all, like, what do you guys think about the deepfakes? [12:06] Can you tell when it's AI produced, or is it so good now that it's pretty difficult to tell if it's real or fake? [12:16] And then, you know, do do you think we're gonna see more deep fakes like this and more cease cease and desist letters coming out? [12:26] Like, people being bold enough to use a public figure to promote their whatever their betting app, you know. [12:33] It's it's kinda crazy. [12:35] So, I guess that's the first piece. [12:36] Can you guys tell AI deep fakes from real? [12:40] And then the second piece of it, what do you guys think about prediction markets? [12:44] Are we going to continue to see prediction markets growing and thriving? [12:50] Or is there gonna be a pullback in the prediction market space? [12:55] Would love to hear your take on some of it. [13:00] Right now, we've got the next kind of story coming up here. [13:04] This is an interesting one. [13:07] Would you let Claude read your bank account? [13:11] Anthropic just launched a personal finance beta for paid US subscribers. [13:16] It reads your bank and card accounts. [13:19] Anthropic says it can't move your money, but ChatGPT already does this for free, so why would you wanna pay Claude for it? [13:29] That's a good question to ask, I think. [13:31] ChatGPT added a bank link in May, and in October, it began rolling out to free users. [13:38] But ChatGPT's free plan carries ads. [13:41] It's finance chats follow your usual training settings. [13:46] It's like free streaming with commercials versus paying to skip them just like on Netflix or Hulu or any of those streaming platforms. [13:56] So Claude bets that you'll pay to skip the ads from $20 a month. [14:00] Anthropic says it's Finance Hub chats aren't used for training, which that's great to hear. [14:09] Pump the brakes, man. [14:12] I don't know if this is the best idea, right? [14:15] Like, the finance application or use case for these AI tools is very obvious. [14:24] It makes a lot of sense. [14:25] AI is really smart with math and numbers and can analyze that stuff. [14:30] But you're talking about connecting your AI agent to all your financial platforms, right? [14:39] I think it came out a few weeks ago. [14:41] JPMorgan Chase reportedly said they could see AI agents working in customer accounts, but they couldn't tell if the agents were nefarious, or if they had been granted permission from the owners of the accounts, or what was going on there. [15:01] And so there's like... [15:03] This is still pretty early. [15:06] So opening up ChatGPT or Claude to all your banking information is that it's a little risky. [15:18] And and from the ChatGPT perspective, if you're doing that on a free account, ChatGPT's free terms of use say that it can use your chats to train its own model, right? [15:30] So if you're using ChatGPT for free and you connect your bank account, is it putting your banking information? [15:37] Obviously, it's not gonna like I can't guarantee this, but I'm assuming that it's not going to leak your account number and routing number and you know, anything like that. [15:51] But still, if that if your like balances and the accounts that you have are getting folded into the training data, I wouldn't want any of that. [16:01] So Claude's saying like, hey, if you've got a paid subscription, you can use that. [16:07] And and on the paid plans, they're not going to use your chats and the training data for the most part, as far as we know. [16:15] I think there are some edge cases where they will keep it for thirty days. [16:20] But the... [16:24] My question is here is like, what's the ROI? [16:26] Like, what it is... [16:28] If I connect all my bank accounts to ChatGPT and Claude, is it gonna somehow save me hundreds of dollars, or make me hundreds of dollars, or thousands of dollars, or something? [16:40] And is that worth it? [16:43] Like, if ChatGPT can save me a $100, but I put my whole financial life at risk. [16:50] What is... [16:51] That's not worth it to me. [16:52] The the payoff is not big enough there. [16:55] Now, if it's like... [16:57] Yeah. [16:57] I don't see... [16:59] It's hard for me to think of a use case where that payoff would be huge. [17:02] This is where having local models, an AI that runs on my computer is gonna be phenomenal. [17:10] Because in that case, I'm gonna say, yeah, please have access to all my financial documents and financial accounts, and help me find blind spots. [17:20] Where am I overleveraged in one spot or another? [17:23] Can you find me investment opportunities that I'm not looking at right now? [17:27] Where could I save money? [17:29] If it's, know, do my taxes for me, like all that kind of stuff, I'm like, yes, let's do it. [17:35] And then that's, you know, pretty secure and safe if it's running on my local machine. [17:41] It's just me and my computer working on this, right? [17:45] If you're talking about connecting ChatGPT in a data center somewhere, my chat's going back and forth, and it's connecting to my Chase account, and that's kinda happening all in an effort to help me find subscriptions that I should cancel? [18:05] Like, that's not that's not worth it to me. [18:07] What do you think? [18:08] Are you guys hooked up to all your financial accounts with ChatGPT and Claude, or are you guys not touching that with a 10 foot pole? [18:18] Alrighty. [18:19] Do you guys use Cash App? [18:21] Well, its owner just cut roles on its help teams. [18:26] Business Insider says, Block told Staff Wednesday that customer and risk teams lose roles. [18:32] In February, Block cut over 40 and blamed that on AI. [18:37] Why did the AI word vanish from this week's announcement? [18:42] Well, Blocks investor filing this week still lists AI automation as a way to run leaner, but the email to workers skipped AI. [18:53] It cited better tools to handle routine tasks. [18:57] Block didn't say why, but challenger says AI blamed layoffs just hit their lowest month of 2026. [19:07] That's interesting. [19:08] So Block is making cuts to Cash App. [19:12] Not surprisingly in the customer service space where they can automate much of that. [19:17] But now, they're not gonna blame it on AI. [19:20] There's no AI excuse. [19:21] But this seems like a direct... [19:23] Obviously, this is a direct AI use case. [19:25] They just don't wanna name it, which is a funny... [19:27] I don't know. [19:28] I wonder why they're doing that. [19:30] Funny why they would make it, you know, try to hide or maybe not try to hide, but just they didn't wanna mention specifically AI. [19:42] Whereas, maybe earlier this year, it was pretty popular to blame AI for layoffs just because it was kind of an easy... [19:50] And it's kind of funny. [19:51] My own perspective on that is I think a lot of those earlier layoffs, maybe several... [19:58] I mean, a year ago, two years ago, where AI was blamed, I think a lot of that was they had over hired kind of really like the post COVID era. [20:12] Most of these companies went on a hiring spree, overhired, and then realized they needed to cut and they needed some reason why they should make the cuts. [20:22] And so AI kinda became the scapegoat where I don't think AI actually was the reason they were cutting a few years ago. [20:29] We just hadn't seen the use cases or the productivity to support that. [20:33] Now, coming full circle here, I think AI actually is very productive. [20:41] And especially in the customer service space where you can build out a lot of automation and AI is gonna be really effective and predictable and do a great job. [20:49] We are seeing lots of headcount cuts there. [20:52] And so it's just interesting that Cash App or Block wouldn't reference that as the the the reason for the layoffs. [21:01] What do you guys think? [21:02] Is AI at a state of productivity now where it can replace humans at some jobs? [21:11] Do you think we're going to see more AI layoffs coming in the next six to twelve months? [21:22] Is Claude helping Anthropic build its models twice as fast? [21:25] Well, watchdog organization meter spent eleven weeks reviewing Anthropic's internal data. [21:33] It estimates Anthropic now improves its AI about twice as fast as in 2024 to 2025. [21:40] Metered guesses AI drives most of that. [21:43] So how independent was the outside review really? [21:48] That's an important question. [21:49] Metered is the referee here. [21:52] It got what it calls unprecedented access to Anthropic's internal data, but in the public report, Anthropic blacked out most of Meter's key evidence. [22:02] Meter checked Claude's work against staff interviews and found no clear bias. [22:07] Still, most of its own analysis relied on Claude. [22:12] This is shaking the AI world, let's say, in the past twenty four hours or so. [22:20] Meter is a theoretically a third party, objective third party group that is supposed to focus on AI safety as one of the the things that they really focus on. [22:34] The the problem here, there is somewhat of a conflict of interest as many of the founders of Meter are also early investors or seed investors of Anthropic. [22:44] And so they benefit substantially from Anthropic's success. [22:50] So it's hard to look at this like in like big big money, you know. [22:55] So it's hard to look at this as really objective. [22:58] That being said, the report that came out was very groundbreaking in terms of the the revelation that AI development is speeding up at two x with 2024 to 2025 development was, which is insane. [23:17] And at this pace, we've talked about recursive self improvement. [23:21] We continue to talk about that. [23:22] And that's so big because it it's there's two main issues here with recursive self improvement, with RSI, with AI training AI. [23:33] The the one issue which is really apparent is if AI is training AI, how do humans keep up? [23:40] How do we know what's going on? [23:42] How do we keep track of everything? [23:44] There's something called interpretability where we're beginning to understand how when these agents are working and working on stuff, we can actually understand what's happening and be able to have kind of an audit trail and be able to keep track of that. [23:58] That's really important. [23:59] So that's one important thing. [24:00] The other thing is the, you know, just the sheer pace of change what's going on here and how fast these models are getting smarter and smarter and seeing that that development there. [24:13] So there's kinda two angles to this. [24:15] What this means is that we will most likely see documented recursive self improvement. [24:26] And this is kind of a funny thing, but in early twenty twenty seven. [24:29] But the funny thing here is it's like, oh, AI is doing most of training AI right now, but why wouldn't we call that recursive self improvement? [24:36] Maybe because it's not fully autonomous. [24:40] It's not fully recursive. [24:41] There's not a loop going where AI is just continuously training AI, so maybe we wouldn't refer to that. [24:47] There's still humans in that loop and maybe that's not RSI. [24:50] It kinda comes back to the AGI conversation, artificial general intelligence. [24:55] Everybody's arguing. [24:56] Not everybody, but there has been continuous arguments around have we achieved artificial general intelligence? [25:03] And some camps are like, no, no way, not at all. [25:06] And others are like, yes, absolutely. [25:07] We achieved it a long time ago. [25:09] And this is kind of a similar thing is that to me, this sounds like recursive self improvement. [25:14] If the agents are proactively training the agents, how is this not happening already? [25:22] And maybe it's not full blown, but maybe you'd call it RSI Lite or something like that. [25:27] So this is huge. [25:30] This is what I reference, you know. [25:33] There there is some contradiction here. [25:35] There's a group called AI twenty twenty seven, AI twenty forty, where they try to predict what's gonna happen in AI in a very specific way. [25:46] They they put a really extensive research paper together and it's really really interesting. [25:51] And they're very heavily focused on the safety measures here, and there's a lot of pushback against the Anthropic where they're saying, hey, we're going faster and faster and faster, and this could be really dangerous, but they're not slowing down at all. [26:06] And then they're not revealing that this is the controversy here with this report, is there's a bunch of redacted pieces of the report, so nobody can understand. [26:16] And and maybe it's proprietary, you know, competitive information, and if you're going public in a few weeks, you don't wanna release that. [26:23] I I understand that. [26:25] But if it's a safety thing, then you should release maybe not specifics, but maybe releasing general ideas or providing guidance to the rest of the community. [26:39] If you are leading the frontier, you'd think that. [26:41] Anyways, what do you guys think about this rate of improvement? [26:44] Is this the fast track to a utopian society, or are we heading to an early apocalypse? [26:53] What do you guys think? [26:55] Or is it even worth freaking out about? [26:58] I get excited. [27:00] I think this is a really big positive. [27:02] What do you guys think about the pace of development of AI? [27:07] Are we going fast enough, or are we going too fast? [27:13] It's Friday, and you guys know what that means. [27:16] It is time for this week's Toss Top Five. [27:20] Let's get into it. [27:21] I love this. [27:23] Coming in at number five, we have Elon Musk's next AI sites, uh, with goofy names. [27:32] The new sites are called SPAM and Gun Range. [27:35] Musk is adding them to Colossus two in Tennessee. [27:39] Elon says, Gun Range should be running by the end of November. [27:43] SPAM follows in December. [27:45] The cool part, the names come from the site's past lives. [27:48] One belonged to a police shooting range and the other held the Hormel products. [27:55] Alright. [27:56] Coming in at number four, we've got Claude putting on the lab coat. [28:01] Anthropics scientists used Claude to map the first three d structure of a bacterial vault. [28:08] Vaults are these hollow barrels inside cells. [28:12] Scientists have puzzled over them since 1986. [28:16] The cool part, Anthropic says Claude searched about 15,000,000,000 proteins to find the candidates. [28:23] The first model took thirty four hours. [28:25] Anthropic says that normally takes weeks, is super cool. [28:29] In at number three, it was a record breaking week for AI money despite the pullback in the market yesterday. [28:37] Apple posted record September quarter sales of a $114,500,000,000. [28:43] Meta raised $15,000,000,000 in bonds for AI, then investors placed a record €53,000,000,000 in orders for its first euro bond with Meta. [28:54] The cool part, Anthropic made $17,000,000,000 in sales in just three months. [29:00] That's more than its full first half of the year, which is crazy to see that growth. [29:06] Coming in at number two, ChatGPT is coming for the search box. [29:10] OpenAI says searches on ChatGPT now equal a fifth of Google's at ChatGPT product lead, uh, said it, uh, said this at a press event on Monday. [29:21] Back in 2023, Microsoft's boss said Bing had about 3% after after a decade of trying, so it has not taken very long for ChatGPT to maybe dethrone the king there. [29:34] And we're coming in at number one. [29:37] The OpenAI just went bug hunting and it came back with 948 bugs. [29:45] OpenAI says its AI agents found 948 secondurity holes in the core of Linux. [29:50] Linux, of course, runs most of the Internet servers. [29:54] The AI wrote a proposed fix for each one. [29:58] Dozens are already in place and working. [30:01] The AI only checked about 7% of the core code, and the computer time cost about $1,200,000 to find these. [30:09] There was a bounty out for these upwards of $66,000,000. [30:14] If you could solve all of these, it would be worth $66,000,000 and it only costs OpenAI $1,200,000 to go after them. [30:22] Pretty exciting stuff. [30:24] That's your TAWS top five. [30:26] Which one was your play of the week? [30:28] Tell us in the comments. [30:31] We'd love to hear it. [30:32] Hey, it's the end of another great week. [30:34] We're excited for the weekend. [30:36] Lots of great news that we covered in the AI space, and love to continue to see the development and what's happening here day after day. [30:43] We will be back here Monday morning trying out some new formats. [30:46] It's gonna be exciting stuff. [30:50] Goal here to number one, bring in entertainment to our audience. [30:55] It should be interesting and engaging content. [30:57] We wanna bring that to you. [30:59] So we're continually testing out new content, new segments, and how we can make all of this AI gabbily gook interesting to our loyal followers. [31:11] As always, come here for your AI News Real Talk Context. [31:17] No hallucinations. [31:19] I'm your human in the loop. [31:20] Austin Wilson, we'll see you Monday morning. [33:15] Did Nvidia's plunge hit your retirement fund? [33:18] The S and P five hundred had its worst day since April 2025. [33:22] But how much of your portfolio is actually in AI? [33:30] Seen LeBron pitching a betting app? [33:33] It's an AI fake with 900,000 views. [33:36] His lawyers want it gone now, and his own Polymarket deal raises the stakes. [33:45] Did you know ChatGPT can already link to your bank account? [33:49] Now Claude does too, but only for paid US subscribers. [33:53] So why would anyone actually pay for this? [34:01] Who helps when something breaks on your Cash App? [34:05] Fewer people that is. [34:06] Block, its owner, cut customer, and risk roles again. [34:11] So why blocks emails skip the part about AI? [34:18] Ever wonder if Claude helps build Claude? [34:22] Meter estimates Anthropic's pace about doubled and guesses AI drives most of that speed up. [34:30] But who really did meters checking?