Jensen Huang's Reported Spy Problem, Google's AI Dodged 2020, Microsoft's $10B OpenAI Haul | TAWS 2026-10-29 https://www.youtube.com/watch?v=v9kC9mr86WQ [0:14] We go live in ten. [0:28] Every morning, the future moves faster. [0:33] The models, the money, the power, and the people building what comes next. [1:11] From Austin, Texas to the world. [1:18] AI news, real context. [1:22] This is the Austin Wilson show. [1:24] I'm Austin Wilson, your human in the loop. [1:32] Good morning, and welcome to the Austin Wilson Show, the number one daily AI news show in the world. [1:41] We are breaking the AI headlines that matter to you today. [1:46] I'm Austin Wilson coming to you live from the AI capital of the world, Austin, Texas. [1:52] And we have got a very interesting story plot for you today. [1:59] Jumping into it at the top, a reported spy plot around Nvidia's chief executive. [2:06] Google's AI is dodging who won the twenty twenty presidential election. [2:11] Microsoft's $10,000,000,000 OpenAI payday. [2:16] Musk's AI company or SI company is buying up a neighborhood reportedly. [2:23] New York's new rules for kids and their chatbots and China's AI labs are supposedly copying OpenAI's top model. [2:35] Hey. [2:36] No beating around the bush. [2:38] Let's get into it. [2:39] We've got a storied, uh, docket today. [2:43] Lots of interesting stuff going on. [2:45] Nvidia's Jensen Huang reportedly had a Chinese spy in his circle. [2:51] The White House knew about this, but he didn't. [2:55] The New York Times says US spies tracked a Singapore executive named Alice Wong. [3:02] They believe that she passes information to Beijing. [3:06] Trump was reportedly briefed on this about a year ago, but he took no action. [3:13] Wow. [3:13] Nvidia's chips power most of the world's AI as we all know. [3:19] US officials actually suspected that Alice Wong's company Mega Speed helped China get Nvidia chips legally. [3:28] A sister company in Malaysia had purchased at least $4,800,000,000 of Nvidia powered gear, but Nvidia says mega speed gets no sensitive information. [3:41] China's embassy calls it a smear tactic. [3:45] This is, uh, pretty crazy. [3:47] We've talked about spy activity in the past and corporate espionage, and this kind of blurs the lines here a little bit. [3:55] Um, Nvidia, of course, is... [4:01] You would expect to be a target of corporate espionage, but this maybe goes a step further in actual state sponsored espionage with, you know, reportedly state sponsored, of course. [4:14] Right? [4:15] That being said, it looks like Chinese companies may have been able to get Nvidia chips through a backdoor. [4:25] This is something that has been speculated on for a long especially after the DeepSeek moment a couple years ago, almost two years ago now. [4:34] DeepSeek dropped their their open source model that was really, really performant, and it was rumored that somehow maybe Chinese companies were getting Nvidia chips through a backdoor. [4:50] If you're not aware, there is currently in place a trade embargo, if you will, on chips between The United States and China. [5:02] China is allowed to get Nvidia chips, but they are essentially nerfed chips. [5:08] They are not the top performance chips. [5:12] They are throttled, if you will. [5:15] And, uh, this is... [5:17] Some see this as a national security issue. [5:21] Others think that free trade should reign and we should open up and sell chips to China because either way they're gonna figure out their way into building their own chips. [5:32] I think Huawei, uh, as well as maybe, uh, maybe Alibaba is working on chip development as well too. [5:40] So we know China's already working on this, but this is a crazy story that... [5:47] And and how far does this actually go? [5:49] You know, do we think that Jensen was being... [5:52] You know, when you think about spy movies or something like that, was he being bugged? [5:56] Was this like, you know, hiding in the ceiling rafters, like listening in on conversations? [6:04] I'm sure it didn't go that far. [6:06] It seems like somebody was at Alice Wong at Mega Speed, a customer of Nvidia was close in and may have been reporting back to Beijing what was going on in inner circle conversations. [6:21] It seems like if Trump knew that this was happening over a year ago and that maybe there was some questionable things going on, but if it was actually a national security thing... [6:36] Because Trump and the Trump administration really... [6:39] They they have been the ones, uh, locking down a lot of the trade... [6:45] The the the trip chip trade, uh, and preventing Nvidia from selling chips to China. [6:53] And so if this was really an espionage thing, I would think that the Trump administration would have locked this down a year ago. [7:01] The fact though that Jensen and Nvidia are just finding out about this now or that this story is just breaking now, That would put me on guard if I'm in in Nvidia shoes, if I am Jensen Huang, if I am the CEO of any tech or AI company right now that that matters, I would really be guarding my inner circle and making sure that these are high stakes. [7:29] Right? [7:29] We see tens and hundreds of billions of dollars being committed to the build out of AI and the competition is fierce. [7:37] We talked about this, uh, previously about the talent wars and fighting to get the best talent. [7:43] And if there are, you know, potential state actors spying on me. [7:51] I'm I'm gonna be very, very careful about this. [7:54] So what do you guys think? [7:56] Should someone have warned Jensen ahead of time? [8:00] Is this on Jensen Wong? [8:02] How did he not know? [8:05] Leave your thoughts in the comments. [8:07] Uh, this is huge or kind of huge. [8:11] Ask Google's AI mode. [8:13] Who won the twenty twenty election? [8:15] Until last night, it told you to go look it up yourself. [8:18] Ars Technica, a publication tested it Wednesday. [8:22] It pointed to people, uh, to a federal election website. [8:26] Instead, it gave the same dodge for 2024. [8:30] Asked from France or Canada, it gave you the answers, but The United States. [8:35] Woah. [8:36] Woah. [8:36] Woah. [8:36] You better go look that up yourself. [8:38] Millions now get facts from AI Answers, especially leading up to the midterms in just a few days. [8:45] Google blamed a technical gap with the 2020 results. [8:50] It also hit twenty twenty two senate races. [8:54] Google's own Gemini app names Joe Biden right away. [8:59] So do ChatGPT and Claude, but it only changed after the outcry from the masses. [9:07] And, uh, only US users were the ones that were getting redirected. [9:13] This is it, uh, this is not necessarily a brand new issue. [9:18] We've seen some gatekeeping in the chat world and the chat features here. [9:27] Um, and whether this is nefarious, I don't know if this is really a feature or a bug. [9:36] Right? [9:36] It sounds like it's more of a a bug, especially leading up to the elections here in a few days. [9:44] And these tech companies, I mean, historically, there have been some impact from how the the big tech companies have, you know, maybe or maybe not impacted elections. [9:56] Right? [9:56] So there could be a chance here or or or maybe it's possible that Google was trying to prevent, you know, directing anything. [10:10] And this sounds more like a bug because it's like what what would the point of not answering who won the twenty twenty election be? [10:17] I I don't know how that's gonna have a huge impact on anything moving forward. [10:22] That being said, it's like it's another one of these checks and balances. [10:26] Our society is great at this. [10:27] When you have a free open society, you get checks and balances and people spoke up and they said, hey, what the heck, man? [10:35] When I'm in Canada, I can jump on Google and Google's AI mode is gonna tell me exactly who won in 2020. [10:42] But when I'm back home in The States, it tells me to go figure it out myself. [10:47] Why why is that happening? [10:49] So... [10:50] And and it sounds like they've fixed it already. [10:52] You guys should go check it out, test it, see if it works for you or doesn't, who won the twenty twenty election. [10:58] Also, is a really good point too. [11:00] We should be... [11:02] Like, have a level of skepticism when you're using these tools, especially if you're using a free AI tool like Google's AI mode. [11:11] You're getting lower level models, lower intelligence, maybe older models that are less performant, that have some of these like, yeah, who cares about the twenty twenty election right now? [11:25] But, yeah, maybe in 2023 when these AI tools started coming out, maybe they did care more about it. [11:31] So that was just a little bug that was stuck in there for the past few years, and you're using an older model. [11:38] So hey, go pay $8 a month or $20 a month for the more advanced versions of AI. [11:47] You're gonna get better results. [11:48] I guarantee you on that. [11:49] And then the the other piece of this is be skeptical. [11:53] Right? [11:53] Ask questions. [11:54] We should be pushing these chatbots and not just openly trusting everything that they say. [11:59] Push them a little bit. [12:00] Ask questions. [12:02] Question where their answers are coming from. [12:04] I I like the Google AI mode because it gives you links directly to the sources usually. [12:11] Right? [12:11] So you can get right to the source of what they're referencing, and that's a great way to check what they're actually reporting there. [12:20] What do you guys think? [12:21] Do you trust AI blatantly, or are you a little bit skeptical like me? [12:28] Alright. [12:28] Microsoft owns about a quarter of ChatGPT's OpenAI. [12:34] Now we know what the partnership brings in. [12:37] Microsoft booked $10,000,000,000 from OpenAI last quarter. [12:41] That's about 11% of its total sales. [12:45] Its overall profit hit $36,600,000,000. [12:50] This is huge. [12:52] The OpenAI bet is now paying Microsoft serious money every single quarter. [12:58] Microsoft says it now owns about 24% of OpenAI. [13:02] The filing also shows OpenAI owed Microsoft $7,200,000,000 at the end of September, but Microsoft offered buyouts to about 60,000 workers just days earlier. [13:16] We talked about this a couple days ago. [13:18] Right? [13:18] Microsoft is profitable, doing really well, but they did offer this buyout. [13:24] And again, I think we see this as a mitigation step, but the bigger story here is almost a third of their revenue is coming from the OpenAI partnership. [13:38] This is not going to last forever. [13:41] I think there is an end date. [13:46] Don't quote me on this, but I believe it's 2030 is the end date for the OpenAI Microsoft relationship, and Microsoft is trying to get their money's worth out of this cash cow. [14:00] The fact that they own 24% of OpenAI is pretty crazy as well too. [14:05] This is... [14:06] These are like, big numbers, um, and it does actually kind of change a little bit when you observe or or think about OpenAI as an entity and how they operate considering that Microsoft is such a big part of that group and that partnership is really, really tied in together. [14:28] And you... [14:30] You've gotta think that that could benefit if hardware is the constraint, if the the build out... [14:37] The infrastructure build out of these data centers is the big constraint for the AI, the intelligence companies. [14:45] You've got to wonder how much that benefits OpenAI. [14:50] Microsoft is fantastic. [14:52] It has a fantastic cloud business in Azure and their data center business and has a lot of capability there. [15:01] And you gotta wonder if that gives OpenAI some of an edge. [15:05] We know that Anthropic has investments from Amazon and has partnered with Amazon in the past. [15:12] So I wonder if there's a similar relationship there as well too. [15:14] Obviously, I don't think the same scale, but, um, there there could be, you know, kind kind of competing benefits there. [15:23] Man, this is fascinating. [15:24] Isn't it crazy to watch all this stuff where it's like, uh, man, just intertwined. [15:30] You got big old tech, Microsoft. [15:32] You got Amazon. [15:34] Still older big tech, but you know, partnering with the young guns, the new guys, the hot guys coming in Anthropic, OpenAI, blazing, and then Microsoft, $10,000,000,000 in revenue from OpenAI last quarter is crazy. [15:50] The other thing too though, it gets a little wonky is like, man, you you get these accountants in here trying to figure out, hey, what's actual revenue and profit? [15:59] How do we make this make sense? [16:00] What's pass through revenue? [16:01] What are your expenses? [16:02] And as OpenAI goes to raise another round, how does this $10,000,000,000 that's going to Microsoft impact things? [16:14] Whose profit is that actually? [16:17] I know they've got that all worked out theoretically, but it does get a little bit wonky. [16:22] What do you guys think? [16:23] Alright. [16:24] Next up, uh, Elon Musk's AI company or SI company as we should call it, wants to buy people's homes. [16:32] The catch is that people can't sue. [16:36] The Guardian says homes in South Haven, Mississippi are already being bought. [16:43] One realtor said the goal is 418 homes. [16:48] Sellers must drop claims in a noise lawsuit against the company's power plant. [16:55] Wow. [16:56] This is crazy. [16:57] So AI's hunger for power is now reshaping whole neighborhoods, not just power bills. [17:04] The company denies the lawsuit's claims and says it wants to be a good neighbor. [17:10] Three neighbors sued in June over the noise. [17:13] A piano teacher says it wakes her up at 3AM, but sellers give up all claims over the plant. [17:21] Renters get no buyout offer. [17:25] Yeah. [17:25] This is kind of a fascinating one where we're seeing worlds collide between big tech and AI, super intelligence, and real estate world, and small town live in South Haven, Mississippi, and communities kinda up in arms about this stuff. [17:42] I did see a a piece about this, about the noise and the the noise level, I don't know. [17:51] It seems to be blown out of proportion to some extent. [17:54] They were saying it was like the, uh, I can't remember the exact level, but they're reading the decibels at night in the back of this guy's backyard, and they said it was the equivalent of like traffic driving by or something like that. [18:09] I can't remember what what they referenced it to, but I'm like, that's... [18:12] Doesn't everybody have that? [18:15] You know, that's kind of a normal thing, traffic driving by. [18:19] Maybe it's the droning nature of the data centers and that the generators are just constantly running versus like traffic is, you know, one or two cars at a time. [18:29] So maybe maybe that's a little bit annoying. [18:31] But it seems like, uh, Space xSI, the super intelligence company, is buying out these homes. [18:40] They're... [18:40] Instead of dealing with complaining neighbors, whether they just say, we'll just buy your house, which is crazy. [18:49] Um, and I wonder how this reshapes data center communities moving forward. [18:55] This doesn't seem to be helpful or beneficial overall. [19:01] Like, why don't we work on, like, lessening the noise? [19:08] Right? [19:09] Like, how do we dampen the noise or something? [19:11] That seems like a better solution than going and buying 418 homes and displacing a whole community. [19:18] Obviously, they're gonna sell by their free will if they agree to sell, and I'm sure they're getting above market offers potentially. [19:27] And then what does SpaceX SI do with all these homes? [19:31] Are they gonna house their workers there or something like that? [19:33] And then the workers just wear earplugs at night. [19:36] I don't know. [19:36] Lot of questions here. [19:37] This is interesting to see. [19:40] It'll be interesting to see if we, like, start figuring out the data center business. [19:47] Just like data center setup and how we manage this moving forward. [19:52] And like we've we've got plenty of rural air... [19:58] I mean, look at West Texas. [20:00] Like West Texas, there is nothing out there. [20:02] I would go build data centers out there. [20:05] Like, you know, building data centers near any kind of population doesn't seem to make a lot of sense to me. [20:13] Seems like we would rather just put them in remote areas. [20:16] It doesn't necessarily matter how close they are to any, you know, physical location. [20:23] Right? [20:23] So, yeah, this may be a case study in in how to handle things or not handle things. [20:31] And maybe we do see a benefit. [20:33] Maybe, you know, those folks at South Haven, they've been trying to get out of small town South Haven for a long time. [20:39] They just got a buyout, sold their house, and they're moving to the big city and they're gonna do something with their life. [20:44] Let's hope. [20:44] I don't know. [20:45] It's interesting to see though, we will continue to track stories around data centers and what's happening here and the continued developments. [20:52] Obviously, we need the data centers. [20:54] We need to win in data centers as a country. [20:57] This is a strategic play. [20:58] It's very important. [21:00] What Elon and these folks are figuring out is really, really important. [21:04] But we need to be considerate of the people that are being impacted, their day to day lives being impacted. [21:11] What do you guys think? [21:12] Is a buyout fair? [21:14] If you were offered over market price to sell your house because a data center moved in, would you take it or are you standing your ground? [21:27] New York just made it illegal for chatbots to act like your kid's boyfriend or girlfriend. [21:35] Governor Kathy Hochul signed the bill Wednesday. [21:39] Chatbots can't claim to be alive or have feelings with minors. [21:43] They can't push self harm, eating disorders, or drugs. [21:48] Companies must verify ages. [21:52] Fines reach $25,000 per violation. [21:58] Most teens have tried an AI companion, and so this rule reaches a ton of kids. [22:05] Uh, common sense media calls it the strongest kids AI safety law in the country. [22:11] A 2025 common sense survey found seventy two percent of teens used AI companions. [22:20] Rules start within months, but every user may now have to prove their age. [22:26] Companies must delete that data after checking the age requirement. [22:34] This is an interesting one, I think, and something that's actually really important. [22:40] This is a this is a great regulation that we wanna see. [22:47] Yeah. [22:48] It's kinda it's kinda creepy if kids are having boyfriends or girlfriends with the chatbot. [22:53] Right? [22:53] Like, that's a little... [22:56] I mean, obviously, this is this is, like, obvious use case. [23:01] Right? [23:02] Companionship, you know, lonely, whatever, and you've got somebody that can talk to you like a real person and, you know, it's not real, but, like, have perceived connection with another entity. [23:19] It's got that. [23:21] That's a real thing and a real use case. [23:23] But, yeah, totally agree. [23:25] This should be for adults. [23:26] I think we've struggled as a society working through the Internet age and social media age with kids and how this impacts kids and lots of issues there that we still haven't solved yet, I don't think. [23:46] And then you add chatbots into it. [23:51] And I think this is obviously a huge disservice to the next generation because as we talked about before, it's like the human connection, the... [24:00] Your your ability to connect as a human with other humans is going to be your superpower moving forward. [24:08] And humans are imperfect. [24:10] Right? [24:11] And if I'm growing up and, you know, I'm a young teenager, whatever, and and I'm connecting with a chatbot that just agrees with me on everything and likes what I like and is kinda like perfect from a conversational standpoint, I feel like that's a horrible precedent to set in terms of just human interaction and relationship, and then opens up a whole new can of worms on social behavior and issues from a social standpoint that when you actually go out in the real world and connect with real humans, there's gonna be, you know, probably a disconnect there because you're expecting a real human to act like your chatbot companion. [24:58] So I totally agree with this law, um, and protecting the kids. [25:05] That being said, the... [25:08] There's there's still an issue for adults, and there's gonna be a lot of adults that just don't wanna interact with humans at all in the future, which could be a sad turn of events. [25:20] That being said, there's always another side to the coin. [25:23] And I think there is a huge benefit here for people that are lonely right now, don't have connection, don't have anybody to connect with, and this gives them an opportunity to hopefully, not this particular law, but chatbots. [25:41] And and and as we see AI develop and get better and better and better at conversations and connecting with humans, seeing a real positive impact there, I think, from, like, a mental health standpoint. [25:53] And if we can get models that are, you know, capable of providing that maybe emotional, relational, psychological support and know what they're doing from, like, a clinical standpoint, that could be a really positive impact, I think, on society in the future. [26:20] Um, what do you guys think? [26:25] Should kids be able to have kinda these AI companions? [26:31] Do you think AI companions are gonna, you know, be a net positive or net negative me moving forward. [26:42] The other the other piece of it too, this is kind of a second, third order effect here. [26:50] Would you show your ID to ChatGPT in order to meet the age requirement? [26:58] And that opens up something else completely different, which is like, is the state trying to chat and track what you're doing on ChatGPT and remove the anonymity from... [27:13] Or perceived anonymity from Internet, which has been a kinda hot topic for a few years now. [27:22] I don't know. [27:23] You'd be the judge on that one. [27:27] China's leading AI labs reportedly trained their newest models on OpenAI's own work. [27:36] The information says DeepSeek and Moonshot fed answers from GPT six Astra into their own training. [27:44] Kimmy's latest model and a DeepSeek update both benefited, neither lab commented. [27:52] Oh, getting spicy here. [27:54] Copying a rival's answers. [27:57] Let's labs catch up for a fraction of the cost. [28:00] OpenAI already tied a July copying campaign to people linked to Moonshot. [28:08] Moonshot released that Kimi model just last week, but the report rests on unnamed sources. [28:15] OpenAI, DeepSeek, and Moonshot did not comment. [28:21] This is a spicy one for a number of reasons. [28:24] Number one, like, obviously... [28:29] I mean, we've seen this for a long time. [28:30] China has been distilling US models and frontier based models. [28:34] And basically, what they do or what's perceived what they do or reportedly what they do or theoretically, okay. [28:44] I'm in China, and I want to understand what, in this case, like OpenAI's answers are to a particular question. [28:56] And maybe if I can understand the answer to that question and the chain of thought that is given in the... [29:03] I can reverse engineer how they model their AI. [29:08] Right? [29:09] And so what they do is they've got like... [29:12] And this is reportedly right. [29:14] Tens of thousands of devices that are all... [29:18] Have their own subscription to OpenAI. [29:21] They're all asking the same questions, and they're paying for it. [29:26] Right? [29:26] They're paying for the subscription or they're paying for the API to OpenAI, and they're paying as the... [29:32] They ask these questions, but they're doing this at a massive scale. [29:35] And then they're able to programmatically understand the answers to each of the questions and kind of reverse engineer that. [29:42] And that's how this distillation has been happening for a while. [29:45] It's like, okay. [29:46] Actually, if we can just brute force and ask all these questions, we can train our own model for a lot less than it would cost to train a model from scratch. [29:58] Right? [29:58] And so this is a similar story here, not really new, but it just sounds like that maybe they've found a direct connection to, um, a couple of these Chinese labs and what they're actually doing directly, not theoretically. [30:19] Right? [30:19] So that... [30:20] That's an interesting thing. [30:21] The the, I think kind of the more interesting thing, if we wanna take it to a second layer, right, is over the summer, we heard a lot from Anthropic. [30:34] And specifically, I think there was one public... [30:39] Either it was a blog post or something that Badario Ahmaday, CEO of Anthropic, had called out this Chinese distillation and complained about it and said that we need the US government to regulate these Chinese open source models and prevent them from distilling the American made frontier models, OpenAI, Anthropic, and prevent them from distilling these models because they're stealing the IP. [31:10] There's two problems here. [31:11] There's two problems. [31:12] Number one, Anthropic and OpenAI and all of these frontier AI companies, they distilled the whole internet without anybody's permission. [31:21] Right? [31:21] So they went and scraped the whole internet, and that's how they trained their their models. [31:26] They've also reportedly been buying books and scanning books and using books to train their models as well too. [31:34] The more data that you get in, the better cleaner data that you get in, the better model that you get on the back end. [31:41] And Anthropic has been doing this since its inception to train and build their models. [31:49] They're using basically freely available data or purchasing the data and, like, books for instance. [31:57] They're purchasing the books. [31:59] There's a free use law where it's like, hey, if I own this book, I can kinda do what I want with it. [32:05] Just to some extent, there there are some restrictions here. [32:10] But Anthropic did that, and then they're complaining when somebody else, okay, I bought a subscription. [32:17] I paid a subscription. [32:18] This is how I wanna use it. [32:20] Their terms of service say you're not allowed to distill it. [32:23] So they're breaking the terms of service. [32:24] They can use their subscriptions, lose their subscriptions, and Anthropic and OpenAI on a daily basis are looking for this and they are canceling subscriptions because they... [32:35] They're figuring out who is distilling this. [32:37] Anyways, getting really deep into it here. [32:40] But, over the summer said, hey, we need the government to regulate this, please. [32:45] And that was a little bit hypocritical. [32:48] Right? [32:49] We need the government to regulate this and prevent these open source people from stealing our models. [32:55] Right? [32:57] The government didn't do anything. [32:59] That's fine. [33:00] A few months later, this was like a couple weeks ago, Anthropic released that they have basically figured out how to prevent, I think it said 99.99 of distillation. [33:13] So nobody... [33:14] Like, these Chinese companies can't distill their models anymore. [33:18] Anthropic figured out internally how to solve this problem themselves. [33:22] They didn't need a third party regulators. [33:24] They didn't need the government to come in and stop these open source models. [33:29] Anthropic figured it out themselves, which is awesome. [33:32] And I love that. [33:33] And Anthropic is amazing, like, to figure that out. [33:38] Now we look at OpenAI. [33:40] Doesn't seem like they have figured it out. [33:42] I'm sure they will though. [33:43] I'm sure they'll get their... [33:45] You know, I'm sure they're pretty close here. [33:47] It seems like they're getting close if they're figuring out kind of a trail here of who did what, but, uh, interesting to see. [33:56] You know what we should do though? [33:57] We should get Ron's take on this real quick. [34:00] Let's see. [34:01] Is Ron available? [34:03] Oh, I don't think Ron's available. [34:05] Let me let me see if he's around. [34:07] Hold on. [34:08] No. I don't think he's around. [34:09] Okay. [34:10] No worries. [34:10] Um, Yeah. [34:11] What do you guys think? [34:12] Is copying AI's answers stealing? [34:18] If you're using your subscription the way that you wanna use it, is that considered stealing? [34:23] You be the judge. [34:26] Alright. [34:26] Uh, one last story I wanted to cover is in regards to Google's latest AI model, it is the Gemini four Carbon that they just dropped. [34:47] Top ranked model by artificial analysis intelligence index. [34:53] And the key thing here, we've talked about this a few times, but Sundar Pichai referenced RSI in the release of Gemini for Carbon. [35:05] This is a huge release and really, really, really massive in terms of what Google has done in such a short period of time, and we're seeing this in across the board, like... [35:20] Well, maybe not across the board. [35:21] I'm sorry. [35:21] We're seeing this in these laggards, Meta, Google. [35:28] Yeah. [35:28] Primarily Meta and Google, but they were far behind. [35:32] And now all of a sudden, they are producing really, really good models. [35:35] We just saw the Gemini four Argon come out a few weeks ago, and now they just released Carbon, which it beats all frontier models. [35:46] Uh, it is the top ranked model on artificial analysis intelligence index, which is, uh, a benchmark used to measure the intelligence of these models. [35:58] And Sundar said that that he referenced RSI. [36:01] So whether or not they used RSI to train this model or if they're close to achieving RSI. [36:08] And remember, RSI is recursive self improvement, which means that the the agent is training the agent. [36:14] The model is training the model. [36:16] The AI is training the SI, which means that these models can train much faster, develop much faster. [36:24] They're gonna be smarter. [36:26] Like, the smartest humans are developing AI, and then the AI is smarter than the smartest humans, and then we have the AI developing the AI. [36:37] You see where that goes. [36:39] And so this is a really, really big story. [36:44] Like, very, very big. [36:46] If they have achieved RSI, phenomenal to see. [36:50] Um, I think this means a lot for Google, but also for the broader economy, the market, and The United States Of America. [37:01] I wanted to touch that before the end of the show today because this is such a such a big, uh, big deal here. [37:10] That being said, we covered a lot today. [37:13] Uh, really, really, really great, um, news day. [37:17] Really interesting to see what's going on, what's coming out. [37:20] We're trying to get better and better. [37:21] I appreciate your comments and what you guys are giving me feedback on because I'm trying to learn and get better one day at a time, and I think we're doing so. [37:32] There was a comment yesterday that asked for us to reference the stories in the show clips. [37:41] Right? [37:41] So now if you go to taws.live, our website, that's taws.live, the austinwilsonshow.live, you can see all the news stories that we cover and links to the sources there so that you can go read for yourself. [38:00] Yeah. [38:00] Other than that, thank you so much for joining me. [38:03] Awesome day today. [38:05] It's gonna be a great rest of the day. [38:06] I'm excited. [38:07] As always, come back here for your AI news real context, and I'm your human in the loop. [38:13] I'm Austin Wilson. [38:15] We'll see you tomorrow morning. [38:17] Let's go. [40:04] Nvidia's chief... [40:06] Take two. [40:07] Take two. [40:08] Nvidia's chief executive reportedly had a spy nearby. [40:13] Trump was warned a year ago, Jensen Huang never was. [40:19] K. Take two. [40:21] Nvidia's chief executive reportedly had a spy in his midst. [40:26] Trump was warned a year ago, but Jensen never was. [40:35] You asked Google's AI who won the twenty twenty election. [40:39] It sent you to go find answers elsewhere. [40:42] But if you're in Canada, it gave you the answers. [40:46] Why? [40:50] Microsoft's OpenAI deal brought in $10,000,000,000. [40:56] Days earlier, it offered 60,000 workers buyouts. [41:05] Musk's SpaceX SI is buying up a whole neighborhood. [41:12] Sellers must drop their noise claims. [41:15] Renters get no buyout offer. [41:18] Is this fair? [41:22] Take two. [41:25] Musk's SpaceX SI is buying up a whole neighborhood in Mississippi. [41:32] Sellers have to drop their lawsuits for the noise and renters get no buyout. [41:40] Would you sell? [41:45] New York just banned flirty chatbots for kids. [41:49] To enforce it, you might have to show your ID too. [42:04] New York just banned chatbot companions for kids. [42:09] To enforce it, you might have to get carded too. [42:12] DeepSeek reportedly copied OpenAI's own work.