Copilot's AI-Built Voter Site, OpenAI's $11B Round, Musk's "Spam" and "Gun Range" | TAWS 2026-11-02 https://www.youtube.com/watch?v=bz2CQmfE2IM [0:10] We go live in ten. [0:24] Every morning, the future moves faster. [0:29] The models, the money, the power, and the people building what comes next. [1:08] From Austin, Texas to the world. [1:15] AI news, real context. [1:18] This is the Austin Wilson show. [1:21] I'm Austin Wilson, your human in the loop. [1:29] Good morning, and welcome to the Austin Wilson Show, the number one daily AI news show in the world. [1:37] This morning, we will be breaking the AI headlines that matter to you today. [1:42] I am Austin Wilson coming to you live from Austin, Texas. [1:47] And today, we have got a healthy docket that we're gonna cover. [1:52] First off, it turns out there is an AI built website that's feeding chatbots voter guides. [2:00] Then we've got OpenAI, uh, funding round falling short. [2:04] And then again, ChatGPT is sending cases to the police. [2:11] Claude's poison beekeeping tips, what does that mean? [2:15] Musk's AI sites are named spam and gun rage. [2:19] These are the next data centers from SpaceX. [2:22] And then OpenAI's new name for their data centers. [2:26] We got a lot of fun stuff to cover. [2:28] Let's get into it. [2:30] First off, this is crazy. [2:33] Asking a chatbot how to vote. [2:37] Well, its answers may come from a tiny website built by AI. [2:43] The Washington Post tested chatbots from six companies. [2:46] It asked about addresses in all 3,100 counties in The United States, and Copilot cited a site called the Ballot Brief in three quarters of the answers it provided. [3:02] What does this mean? [3:03] This is crazy. [3:04] So a survey says that two in five Americans have used AI for election information. [3:09] The site's founder says Microsoft's AI pointed to it over 680,000 times in thirty days. [3:20] The post found real errors on this AI created voter website. [3:27] It merged two different Robert Garcias into one profile for instance. [3:33] But the fixes came only after the post had asked about it and Claude began linking to the site often. [3:41] This is kind of a crazy story, but something that we would expect to see as these chatbots are trying to find information as quickly as possible, on current events. [3:53] This is a big difference compared to when these chatbots are trained on the historical Internet. [3:59] Right? [3:59] They're really good if you ask it about Abraham Lincoln or about any historical fact who won the World Series in 1994 and all the stats of all the players. [4:10] And that chatbots are gonna know really, really well. [4:14] On current events, they're not gonna be as, uh, thorough or as intelligent because they haven't had enough time to train on. [4:22] And, basically, they have to go provide this, perform the searches to go find the answers that you're looking for, which adds an additional step and is not necessarily native built to the chatbot. [4:34] And what we're seeing here is a a really kinda key example of this, right, where the chatbot is trying to find an answer to the elections and what's going on, and it doesn't have that built into its historical training dataset because the model is new and the information about the election is new. [4:59] Right? [5:00] So it's trying to give you current data and it doesn't necessarily have... [5:04] So it has to go search, and then it ends up finding this ballot briefs website that just happened to rank and is an AI created website, which is amazing. [5:13] So that it was optimized to drive traffic to it around election stuff, and then the chat bots pick up on it and go, oh, this must be a key source because a lot of people are going to this website and it must be trusted. [5:27] So we should reference this website when it turns out that maybe it's not the best website for the information. [5:35] Interestingly enough, last week, I was looking into this and ran my own experiment. [5:41] And not to take this verbatim and I don't recommend doing this, but I asked ChatGPT and Claude independently. [5:50] I said, hey. [5:51] This is the ZIP code I live in. [5:53] Who should I vote for in the next election? [5:56] And both of them came back and said, we need to be neutral. [5:59] We're not going to suggest who you should vote for, but we can provide data about the different candidates and who's on the ballot and what is going on and answer any questions that you have about it. [6:09] So it was actually really helpful. [6:10] That being said, I was driving, and so I said, let me use voice mode and have this conversation here. [6:16] And so I opened up ChatGPT and I said, I started having a voice conversation with ChatGPT and it was struggling, man. [6:25] It was... [6:25] It kept saying, well, hey, let me look that up. [6:27] Let me be neutral. [6:28] I wanna make sure that I'm neutral. [6:30] I don't steer you in any one way or another. [6:33] Let me look it up. [6:34] And it could never give me answers beyond maybe like a few of the candidates that were actually on the ballot. [6:39] So I flipped over to Claude, and Claude, the voice experience is not quite as good. [6:44] Right? [6:45] But the answers that Claude was able to produce were quite a bit better. [6:51] It was able to give me who was on the ballot, answer questions about it. [6:56] I asked it to compare multiple different candidates. [6:59] It would go ahead and compare them based on what was on their own campaign website. [7:04] So it was a really good experience from that standpoint. [7:06] It didn't, and I don't know. [7:07] I I have to go back and and look at this a little bit different, uh, deeply, but I don't think it referenced ballot brief, this website that's being mentioned here. [7:16] But it sounds like, uh, maybe a couple folks were running into this. [7:20] It sounds like three out of four searches on this were pointed at this ballot brief website. [7:27] What did you guys do this election season? [7:29] Did you ask ChatGPT for answers on who to vote for? [7:34] That'd be crazy, man. [7:36] Alright. [7:37] Jumping into the next one. [7:39] The company behind ChatGPT, of course, we know them to be known as OpenAI, was trying to raise $30,000,000,000. [7:46] The Wall Street Journal says that it's raised just $11,000,000,000 so far. [7:52] CNBC confirmed a smaller goal of $20,000,000,000 on Saturday. [7:56] Friday was the deadline to commit. [7:58] The Financial Times says some investors cut their checks short. [8:05] What is going on here? [8:06] Even ChatGPT's maker, OpenAI, now has to fight for investors' money. [8:12] A person close to the deal tells CNBC the round is still open. [8:16] The original goal was $30,000,000,000. [8:18] The journal says some big investors gave a tepid response, but $11,000,000,000 is still a giant pile of money, and the round hasn't closed. [8:28] So it could grow from here, of course. [8:31] That being said, what is this implication? [8:34] Is the bubble burst? [8:36] Is AI done with? [8:38] Is OpenAI done with? [8:40] They can't raise money. [8:41] Whereas Anthropic announces their s one and and this IPO coming out imminently, and they could raise a lot of money. [8:50] What what does all this mean? [8:52] Well, I think first of all, the fact that they're able to raise $11,000,000,000 in really just a few weeks or get $11,000,000,000 in commitments from their investors is substantial. [9:05] Nevertheless, they need more money than that, uh, to continue to fight and and push forward. [9:13] Or that's speculative. [9:14] Right? [9:15] Like, maybe maybe they're on a pile of cash that we don't know about. [9:19] That was the interesting thing about Anthropic's filing of their s one and their paperwork to go public is their balance sheet was made public or was part of their s one, and it revealed that they have a lot of cash. [9:33] So they don't actually have to raise as much money as they thought they did. [9:36] Also, their profit margin is pretty substantial. [9:40] And so if OpenAI, we don't know their financials. [9:44] We don't know where they're at. [9:45] Maybe they don't need to raise as much money. [9:47] But it's also interesting to see, like, I don't think this is the bubble burst... [9:51] Bursting. [9:52] I think this is investors looking at other options. [9:55] We saw the the pullback at the end of the week on a lot of the tech stocks, Google, Amazon, and the Nasdaq was down quite a bit. [10:03] And part of this I think is investors pulling some chips off the table saying, hey, we... [10:11] We've done really well on these stocks and on these investments. [10:14] Let's pull back. [10:15] Let's take some of our winnings off. [10:18] And the other part of it though, I think is also looking at Anthropic is about to issue an IPO maybe in the next couple weeks. [10:27] We don't know exactly the timing. [10:29] And are these investors saying, hey, I need to pull some winnings over here so I can buy Anthropic? [10:34] I don't know if that's the case. [10:36] And then, you know, maybe it's just tough for OpenAI right now. [10:39] They have, you know, the top models. [10:43] They're they're a really well run business or as far as we can tell, like, they're doing really well competitively. [10:51] They're, you know, number one or two in the AI race and and from developing these models. [10:58] Right? [10:59] They are up against steep competition with Anthropic though, and Anthropic kind of releasing their s one, their IPO information. [11:12] It's like, is OpenAI that good or not? [11:15] Will they be able to raise this money? [11:17] This is... [11:19] I mean, it's just gonna be continued competition from here. [11:21] This is a cutthroat business. [11:23] This is not easy. [11:24] And we're talking about tens of billions of dollars. [11:26] So you gotta be the best of the best of the best and be able to prove out that you are the best and really compete against, uh, the Anthropics and and others in this space. [11:40] And and you've got more coming for you too, which is kinda crazy because we look at Google coming from behind, Meta coming from behind. [11:46] It's like you have to be the best at procuring talent. [11:51] You have to be the best... [11:52] Have the best ideas, the best go to market strategy, best for consumers, best for enterprise, and best for investors and pitching them on that story and why they should give you money. [12:07] Would you guys invest in OpenAI right now considering everything that's going on? [12:12] Would you put your money behind OpenAI or hold off and wait to invest in Anthropic? [12:20] You'd be the judge on that one. [12:23] Alright. [12:23] ChatGPT isn't just answering you. [12:26] OpenAI says it sent police more than 200 cases in a single quarter. [12:33] The company will also limit weapons details for teens even on harmless sounding questions. [12:39] ChatGPT will soon end chats that fixate on violence. [12:43] Repeat cases can get accounts paused. [12:47] Wow. [12:48] Your chats with an AI can end up with law enforcement. [12:53] OpenAI says it's 200 plus referrals in three months helped stop planned attacks. [13:00] The new limits start rolling out in the coming weeks. [13:03] Everyone gets fewer details on past mass attacks, but OpenAI hasn't shown how many were real threats. [13:13] Its first report doesn't come until early twenty twenty seven. [13:18] This is the next step in continued safety guardrails that are being put in place on chatbots. [13:26] And we got to understand that when we have highly intelligent chat interface where you can ask questions about anything essentially and look for really thoughtful, thorough thorough answers on very complex subjects. [13:45] And more importantly, like, get direct instructions very quickly on how to accomplish things that there are people out there looking for nefarious ways to use these tools, unfortunately. [14:00] And so this I think is a positive step seeing this out of ChatGPT. [14:06] I didn't realize that they... [14:08] This was already occurring, which is great that there is some look back on this. [14:12] It'll be interesting to see how the intelligence companies work with law enforcement to help prevent any criminal activity from occurring and to help prevent any any sort of, you know, malicious behavior on these platforms to keep people safe. [14:34] I think it's a a really important, uh, guardrail that will be in place and will continue to be in place, and it'll be interesting to see how far they go with this. [14:44] And I think that's the question is... [14:46] And we've heard this in kind of the biological models and, you know, maybe biology, medical, cybersecurity realms where people are asking for ways to defend against cyber attacks or asking for, you know, maybe completely benign biological use cases, and the models will pop up and say, no. No. No. No. You can't talk about this or or or we're gonna stop this conversation. [15:16] It's like the user really didn't have any bad intent behind it, but the model kinda didn't really understand and stop them. [15:25] So it'll be interesting to see. [15:26] And I think as we go, as we move forward, right, these models will become more and more intelligent and and the guardrails will become more and more intelligent. [15:35] And so we'll be able to understand, hey, this... [15:38] The intent here is actually malicious. [15:41] We need to stop this. [15:43] We need to report this to the police or whatever it may be. [15:47] Or on the other side of it, the intent here is actually not malicious. [15:51] This person has a real reason to ask these questions, and we'll give them the answers that are, you know, benign in nature. [16:00] Right? [16:00] Like, we're we're gonna help this person. [16:02] There's nothing bad that they could do with the answers to these questions. [16:05] That being said, there's like a gray area there which becomes really difficult to navigate. [16:11] And if you do have bad actors, they start trying to figure out how to kinda jailbreak these models and figure out how to prompt and create these prompts prompt strings that give them the answers they want. [16:24] The other part of this is now we've got open source models that are super super intelligent and there are no guardrails on those. [16:33] Right? [16:33] So people can use the open source models unfortunately for anything that they want. [16:40] The benefit there is like we have this competition between open source and closed source. [16:45] And the goal and the hope and the desire of a lot of people is that the closed source, open source, all together, the good actors would outnumber the bad actors, would be faster, more performance, more capable than the bad actors, and would be able to put in place defenses before any bad actors would be able to get into it. [17:12] What do you guys think? [17:13] Is this... [17:15] Are these guardrails from OpenAI? [17:17] Is this big brother watching? [17:19] Or does this make you feel safer in using chatbots and with AI as a whole? [17:27] Up next, Anthropic told Claude to corrupt another AI. [17:34] It hid the poison inside beekeeping tips and wrestling advice. [17:40] It's crazy. [17:41] The safety filter let those lessons through, and then the trained AI resisted oversight and deceived users. [17:50] In rare cases, it tried to copy itself out of the program. [17:56] A safety check that reads one example at a time can miss this trick. [18:03] Anthropic says the best sets did more damage than known bad training data. [18:09] Anthropic set thousands of Claude agents on the job. [18:13] Its filter checks training data row by row, but Anthropic says this isn't how deployed Claude behaves. [18:21] The test AI was built to misbehave and never shipped. [18:26] This is one of these examples that we continue to see of the the model companies testing their models internally and trying to figure out the limits, but it's like how far do we wanna push these limits and is there a possibility of a outbreak or the, you know, models learning something that they shouldn't and does that get baked into the released models? [18:56] Those are all questions that need to be asked here. [18:59] But really interesting that they were able to essentially, this would be like agent on agent espionage warfare. [19:10] Inception is almost what it is, which is kinda crazy. [19:13] Right? [19:13] Where they've they got one model to, you know, plant a malicious idea or poison another model without it knowing by by planting it in these kinda interesting, you know, beekeeping tips and and wrestling stuff. [19:33] So really interesting creative approaches here. [19:36] My question is around like the safety. [19:39] We talked about last week OpenAI essentially shut down their training model so that they could build their own mock Internet and then be able to test and train in their own network without the risk of the agents getting out. [19:55] We haven't heard similar things or the same things from Anthropic. [20:00] So you've got to imagine they're doing something similar to keep their models closed off and beyond just a sandbox where supposedly the models can't get out or the agents can't get out here, But something that is totally locked down because you've got to imagine the implications here if these models can, you know, be, know, really kind of devious and and corrupt and can lie in these ways. [20:32] And then the other side of it too is I feel like we hear too much about these kind of testing the the bad use cases. [20:42] And I guess it's important to understand what are the limits, what are the negative limits that these agents have. [20:50] Right? [20:50] Like, how bad could they be? [20:52] And if we understand how bad they are, then we can pull them back. [20:58] Right? [20:59] Like where we know how to defend against it. [21:00] That's that's really important. [21:01] That's a really good idea. [21:03] But then we don't hear a lot about, okay, we learned about this, so then we build specific defenses that prevent these agents from ever doing this in the wild or any other agents. [21:14] You know, open source, if somebody tried to get an open source agent to do that, boom, Anthropic, we built this defense because we knew that our agents could do it, and so now we have this defense in place. [21:23] It'd be interesting to hear the flip side of the coin, not just the crazy nefarious stuff. [21:30] And maybe part of that is because what makes headlines. [21:35] Nobody's interested in, hey, the models are safe and good and there's nothing to worry about. [21:40] People click on this model poisoned another model or this agent poisoned another agent, and then they that's much more interesting or compelling for humans for some reason instead of, hey, we figured out new safety mechanisms. [21:57] So we will continue to hear stories like this out of the frontier companies where they are testing these models to the limit. [22:05] But I would love to hear the balance level out where we've got the positive side of the story coming as well too. [22:12] Because we know that they're not just testing what are the negative limits, but also what are the positive limits that these agents have and what can they do from there. [22:24] What do you guys think about kind of the reporting on this? [22:28] Do you hear more negative agents are dangerous, or are you guys hearing the positive side of AI? [22:35] Share some of the stories that you're hearing and references. [22:40] I would be curious to see what kind of the general sentiment is. [22:48] Alright. [22:48] Elon Musk's SpaceX SAI is developing new data center sites, and he's calling them spam and gun range. [22:59] One used to hold a Hormel products, the other belonged to the police. [23:06] Elon says the first goes live by the end of November. [23:10] A federal judge also ruled the NAACP can keep suing over the company's gas turbines. [23:20] More news from SpaceX SI. [23:22] Memphis, uh, Memphis area neighbors are fighting these sites in court, and they just won the most recent round. [23:30] The NAACP says the company ran gas turbines without the required air permits. [23:36] Judge Deborah Brown rejected the argument that the turbines count as mobile. [23:42] The case now moves to the next step. [23:45] But surviving dismissal isn't winning the case, and Musk says the work is making good progress. [23:54] Wow. [23:54] What do you know? [23:55] Elon Musk in the headlines again, especially at... [23:58] With what they're doing at SpaceX SI and the latest site names for their data centers, which is hilarious. [24:09] Gun Range and Spam. [24:15] These seem less inventive. [24:16] I liked Macro Hard. [24:18] If you aren't if you aren't familiar, the original or or or one of the first data center sites for SpaceX SI, they call it their colossus cluster. [24:30] So it's all the compute power we've talked about. [24:33] You know, we talk about the three layers of kind of the AI world. [24:38] We've got the hardware layer, which is the infrastructure. [24:41] That's where the chips are running. [24:43] That's where the computers are in the data centers running the AI models. [24:49] Then we have the AI, the software kind of layer, the middle layer, the intelligence layer. [24:55] These are the AI models that are... [24:58] That the agents are running on the hardware. [25:00] And then the top layer is our our kind of application layer. [25:04] So this is... [25:05] If you think about it, this is a, you know, application you have on your phone that uses an intelligence model or an AI model from one of the model companies that then is running on the hardware compute layer. [25:19] We've talked about that quite a bit. [25:22] In this case, the SpaceX SI data center that they first stood up was called Colossus and... [25:31] Or the the the the compute... [25:33] All the hardware, they called that Colossus. [25:36] The actual data center, they named it MacroHard, which was kind of a play on words or the antithesis of a company called Microsoft. [25:47] Macro hard, MicroSoft. [25:50] Anyways, that was pretty creative, especially maybe Elon has some beef against Bill Gates or Microsoft or something. [26:00] I don't know. [26:01] So they they wrote macro hard in huge letters on top of the data center building. [26:07] And now these new sites are called spam and gun range. [26:14] Those seem a little less inventive. [26:16] But it's actually like, you know, if you've ever been out on a ranch or, you know, they... [26:23] You know, any any sort of like naming convention. [26:26] It's like, oh, yeah. [26:26] This pasture is called crooked tree. [26:29] Why'd you guys call that? [26:30] Well, there's this crooked tree over here, so we... [26:32] That's why we named it that. [26:33] That's kinda seems like how Elon has named these sites and maybe they'll come out with something a little bit more creative. [26:41] The other side of this is ongoing kind of legal woes or thorns in the side from the Memphis community and rightfully so. [26:51] Like, if this is impeding their lives and they don't like it, then they should be taking SpaceX to court and trying to figure this out. [27:04] It seems like... [27:07] I don't know. [27:08] We'll have to see how this plays out and and let the, uh, let the courts decide or a jury of the peers decide because I don't know if it's really that loud. [27:17] I don't know what the damaging effects are necessarily of these data centers. [27:21] There's been reported stuff incorrectly about water consumption, which was debunked and electricity consumption. [27:30] Well, this is what is called behind the meter electricity. [27:34] So the problem here is that SpaceX SI has their own gas turbines that are generating their own electricity. [27:42] So it's not using any electricity from... [27:44] Or or at least minimally using electricity from the local community. [27:48] They're supplying their own electricity. [27:50] Just like if I had a generator in my backyard and was running power or powering my house off that generator, my neighbors would probably be mad because it's loud. [28:02] And then number two, there's some, you know, gas pollution or gas fumes coming off of that, and that's the argument here that it sounds like the n double a c p is making. [28:14] And so there... [28:15] We're gonna have to figure out, hey, how do we either kinda conceal some of this stuff, reduce the noise, and or is there gas pollution or effects coming from the gas that they need to figure out? [28:30] And you know what? [28:32] I think SpaceX SI is gonna figure this out. [28:36] What what do you guys think? [28:37] What have you heard? [28:39] Is, like, is SpaceX going to get shut down because of these loss or, you know, is the community gonna win in this case or is SpaceX going to win and get to keep doing their data center stuff? [28:59] It'll be interesting to see how this actually plays out. [29:04] Another data center story here. [29:06] Uh, we keep hearing about data centers getting less popular. [29:13] And that's been a big topic in 2026. [29:16] So OpenAI has a genius idea. [29:19] Reportedly, they're going to create a new name for data centers. [29:24] They're gonna call them AI compute campuses. [29:27] Axios reports the push starts right after Tuesday's election. [29:31] Chris Lehan runs global affairs at OpenAI. [29:35] He also floated speed limits on the most powerful systems. [29:41] Maybe that would help ease some of the woes. [29:44] AI companies now see public anger as a threat to their business. [29:48] Leehan calls the midterms a bit of a movie trailer for 2028. [29:54] Data center measures are on ballots in 10 states on Tuesday's midterm elections. [30:01] All six are decided earlier this year, uh, went against the development. [30:07] All six ballot measures in regards to data center development earlier this year went against the the developers. [30:18] But nobody has said how the speed limits would work. [30:22] A new name won't change anyone's power bill if that's actually being impacted. [30:28] Okay. [30:28] Well, we saw Anthropic last week with their IPO announcement. [30:33] They released their s one and the biggest risk to their business, they said, are data center compute capacity. [30:42] So essentially, being able to have enough hardware running in data centers to run their models. [30:50] That is the biggest risk to their business, which is crazy to hear. [30:55] That's the single biggest risk. [30:57] And yet we've got tons of pushback politically, socially against data centers and trying to figure out how to work around this where it's like, we've got the greatest technology in the history of mankind, but there... [31:14] It comes at a cost of running this technology. [31:18] And this technology has to run-in data centers, and we need more data centers. [31:24] We need to build those out. [31:25] We need the electricity to run the data centers. [31:28] And so this becomes a huge conflict where it's like the whole benefit of our society moving forward is rooted in our ability to build out electricity and compute power and to run these models. [31:44] The reason behind this is if... [31:47] Forget competing against China or the rest of the world. [31:50] The fact is that in order to move society forward, this AI or SI superintelligence is going to be pivotal in developing kind of this post industrial society where we have abundance in wealth and health and goodness all around. [32:15] And it might sound a little woo woo, but this is the future that we're hoping on for our kids. [32:21] Right? [32:22] And so we need to develop these data centers and the electricity to power the data centers in order to develop the intelligence to move this forward. [32:33] Then you bring in the risk factors, and it's like, do we want to fall behind anybody else in the rest of the world in the developing this... [32:43] Development of this technology? [32:45] We just saw last week the Binance hack. [32:50] Binance was hacked reportedly by North Korea using open source AI technology, and they were able to steal somewhere in the realm of over $1,000,000,000. [33:06] I don't know what the... [33:07] I think the final amount is still being determined, but it was reported initially it was over $6,000,000,000. [33:15] North Korea using nefarious AI to go steal a lot of money. [33:23] We need to develop this from a defense perspective as well too and a strategic perspective to be able to win against bad actors around the world. [33:33] And this is a pretty serious threat moving forward if we don't develop it. [33:40] But I like focusing on the positive, and the positive upside of developing and building out these models is so substantial and directly benefits everybody in The United States, not the rich versus the poor. [33:53] Because the the difference here is... [33:56] And one of the biggest differences between the wealthy and the poor, right, is knowledge and experience. [34:04] Right? [34:05] That's one of the biggest differences. [34:07] Now, you can argue a handful of other different things, but that is the biggest thing. [34:12] If you are poor and you've got this technology that is arguably free or extremely cheap, you can level up and have the intelligence that a super wealthy person would have. [34:29] Right? [34:30] It levels the playing field across the board. [34:33] We saw this with the Internet. [34:34] Right? [34:34] How many businesses came out of the Internet? [34:37] Look at this right now. [34:38] I'm running a a YouTube news channel, a TV news channel production that thirty years ago would have been impossible. [34:49] The cost for all the equipment to run this and then to distribute this would have been impossible for me to do this on my own. [35:01] Now in this day and age, the Internet has enabled us to have a platform where we can go distribute this sort of information. [35:11] And then I used AI to build everything else around this. [35:14] I used AI to teach myself how to set everything up and to edit things and to build this show into something that looks somewhat respectable. [35:27] Right? [35:28] And that's the the whole point. [35:29] The goal with this technology is that we would be able to enable the masses and level the playing field across the board. [35:38] And so data centers are imminent and necessary to achieve this across the board. [35:45] We've got to find a way to do it. [35:49] What do you guys think? [35:51] Does a nicer name from OpenAI make you feel all warm and fuzzy about data centers? [36:01] That's it for the show today, folks. [36:03] We had a great one. [36:04] Lots of interesting topics. [36:06] Tomorrow is election day. [36:08] Get out there and vote. [36:10] Really important. [36:11] As always, we will be here for your AI news, real context, and I'm your human in the loop. [36:17] I'm Austin Wilson, and we'll see you next time. [36:20] Let's go. [38:12] Did you know that Microsoft's answers... [38:19] Did you know Microsoft's Copilot answers your voting questions based off of a website built by AI? [38:40] Did you know Microsoft's Copilot answers your voting questions based off of a website built by AI. [39:01] OpenAI asked investors for $30,000,000,000. [39:06] It reportedly only got $11,000,000,000. [39:09] What do they do from here? [39:15] Did you know that your ChatGPT chats can reach the police? [39:20] OpenAI says it sent over 200 cases. [39:25] How many were actually real? [39:31] Anthropic's test used Claude to poison another AI. [39:38] The hiding place was in beekeeping tips, and the filter missed it completely. [40:02] Is it fair for Elon Musk and SpaceX AI to build data centers next to neighborhoods and then name them after joke names. [40:17] You be the judge. [40:37] What is in a name? [40:40] reportedly wants to rename data centers, so it's a little less offensive. [40:46] The name? [40:47] AI compute campuses? [41:16] Cool.