Claude Code Boss Defects to OpenAI, AI Money Loses Election Bet, 948 Linux Holes Found | TAWS 2026-11-05 https://www.youtube.com/watch?v=BlXXdTxbc90 [0:11] We go live in ten. [0:25] Every morning, the future moves faster. [0:30] 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. [1:38] We are breaking the AI headlines that matter to you. [1:42] Today, I'm Austin Wilson coming to you live from Austin, Texas, and we have got a busy docket today. [1:50] Lots to cover. [1:52] First off, Claude Codes losing some of their top employees to OpenAI. [1:59] Uh-oh. [2:01] AI money's midterm flop. [2:04] Looks like some of the AI packs maybe bet on the wrong horse. [2:08] OpenAI's AI bug hunters. [2:11] OpenAI also promises data center to data centers to town, so it's some crazy benefits there. [2:19] And then your boss trusts you less with ChatGPT. [2:23] Uh-oh. [2:24] And then maybe we might have a little bit more to cover as well too. [2:29] Let's get into it here. [2:31] First up, would you leave the hottest AI product to go work for its rival? [2:38] Cat Wu ran product for Anthropic's Claude Code. [2:41] Wednesday night, she said that she's joining OpenAI's core products team. [2:47] And who else from Claude's maker just revealed a move to OpenAI? [2:52] She's not the only one. [2:54] OpenAI is now hiring away the people who built Anthropic's biggest hit. [2:59] She posted her Anthropic exit Tuesday. [3:02] OpenAI confirmed the hire to Business Insider. [3:05] It's like the head coach's play caller signing with the crosstown rival, Austin Llau, who helped build Anthropic's growth team, also joined OpenAI. [3:17] This lands right before Anthropic's planned stock debut. [3:22] Oh, we love to see the shifting of the chess pieces on the board as we're now seeing Anthropic employees hop over to OpenAI. [3:34] This isn't new from a general trend in terms of seeing top AI talent move around to different AI companies. [3:43] Of course, we've seen moves from some of the biggest names like Andrej Karpathy moving from OpenAI to Anthropic, and other top AI researchers moving all around. [3:57] It's it's really interesting to see. [3:59] And especially, I think the most recent kinda big ones that we saw were folks at Meta, top AI researchers who got crazy comp packages, kinda stayed there for a year for probably their their deal to best, and then jumped and went to do the next thing. [4:18] The AI labs, or the rival AI labs, are not the only competition as well too, because we've seen a lot of these top AI researchers, or folks that are helping build at Anthropic, or OpenAI, or Meta, they will end up getting a deal to go start their own AI company and raise some VC money to go do so. [4:43] So this is not an entirely new thing, but the trend has generally been folks leaving OpenAI to go to Anthropic. [4:51] We started seeing that maybe six months ago, and now we're seeing kind of the reverse of that, which I don't know if this is an indication of OpenAI doing some phenomenal new work over there, and maybe they're racing ahead. [5:10] In this race, obviously, OpenAI came out hot, came out first three or four years ago now. [5:16] And I guess it's been four years. [5:18] That's crazy. [5:18] 2022, November 2022 is when ChatGPT launched. [5:23] OpenAI got a huge head start, went really, really fast, and they were winning for a long time, right? [5:32] And then we saw Anthropic, I guess about a year ago, came around and began to usurp the preeminent leader. [5:44] And so we started seeing a lot of talent jump from OpenAI to Anthropic. [5:50] I don't know if this is an actual indication that maybe OpenAI is attracting folks. [5:56] Because typically when you get to this level, when you're talking about this much money and some of these comp packages are reportedly in the 9 figure range, not necessarily these two folks that just recently left Anthropic. [6:09] But when you're talking about that much money, I think it becomes less about the money and more about, hey, I wanna work on really interesting projects. [6:17] I wanna have a big impact. [6:19] Where can I go have the biggest impact? [6:22] And who's doing the most interesting work? [6:24] And so, I think when we saw the transition from folks moving from OpenAI to Anthropic, that was the case. [6:31] It's like, wow, Anthropic is doing something really interesting. [6:34] What are they doing over there? [6:36] Oh, also they're gonna offer me a ton of money. [6:39] Let's do it. [6:40] And now I wonder if maybe the tide is shifting. [6:45] The the great thing about this competitive landscape in the AI world is you can't count anybody out ever. [6:54] Right? [6:54] So we thought maybe Google was out of it. [6:57] Meta's out of it. [6:57] We've seen them come from behind and produce really amazing results, where we thought, you know, maybe OpenAI and Anthropic have pulled ahead so far that nobody else would be able to catch up. [7:09] And that doesn't seem to be the case. [7:11] And so, it's always kind of on our minds who is going to take over and be the next leader in the space, and going to push the envelope further in terms of intelligence and capability, and who's actually winning. [7:27] And again, of course, the competition. [7:29] When you're talking about everybody, all these labs are doing pretty much the same thing. [7:33] Pouring tens of billion dollars, tens of billions of dollars of investment into compute build out, and training models, and building applications. [7:44] And the strategy, you can't really differentiate. [7:47] It becomes very nuanced. [7:48] The biggest differentiator that you can have and the only way that you can win in this is whoever's got the best talent. [7:55] And so this is a huge thing to see kind of a transition from Anthropic to OpenAI. [8:02] And we'll see if more follow, or if the... [8:06] These are just kind of a one off, two off type of situation. [8:09] What do you guys think? [8:10] If you're working on the best product in the world right now, theoretically, Claude code or working at Anthropic, are you going to take a package to move to OpenAI or are you staying put where you're at? [8:26] I don't know. [8:27] I think I think I might stay at Anthropic. [8:31] I don't know. [8:32] Pretty crazy stuff. [8:33] Alright. [8:34] Can money buy a congress that goes easy on AI? [8:38] Well, AI's super pack spent $79,000,000 in 68 races as reported by Fortune. [8:48] Their biggest swing race bet was South Carolina senator Darlene Graham, and she lost on Tuesday. [8:56] What did her opponent call her during the campaign? [9:01] This is kinda funny. [9:03] Critics say that the money was aimed at reducing rules around AI, but the blue wave washed over the country on Tuesday. [9:12] Wired says, leading the future, the AI positive pack reported zero new donations in the second and third quarter this year. [9:23] And it's like paying for the loudest float in the parade and then getting booed. [9:28] Data center Darlene is what Annie Andrews called her opponent. [9:33] Annie ended up winning about 51% to 47% against Darlene, and AI Money had spent millions of dollars backing Darlene Graham. [9:44] This is interesting to see, and especially I think when we rewind and go back to 2024 and look at how much money was spent probably on the right, and backing Republican candidates, and seeing how fruitful that was then. [10:03] And then realizing, hey, that same playbook just doesn't work, I guess, with AI. [10:08] It didn't work, obviously, for anybody who funded Republican candidates this past election. [10:16] And so it's it's not guaranteed to to work. [10:20] Right? [10:20] Just pouring money on it is not guaranteed to work. [10:23] And I think the sentiment in the country is very very apparent and real. [10:28] And the interesting thing, this is something that I'm struggling with is like, hey, we saw a lot of the data center, you know, the the the the data center measures on the ballot did... [10:40] Were were, you know, voted to restrict data centers or voted against data centers, however you wanna phrase that. [10:47] But we but everybody that I talk to, or most people that I talk to, of course I'm in a unique circle, but a lot of people that I talk with use ChatGPT, use Claude on a regular basis. [11:00] They like the AI tools. [11:02] They just don't want the data center companies basically taking advantage of their community, or causing utility prices to go up, or any kind of negatives that could come from data centers. [11:15] And so it makes a ton of sense. [11:17] And the people spoke on Tuesday, right? [11:20] Like, we need the data centers. [11:22] We need to build this out, but not in a way that harms the local community, or makes life worse for anybody. [11:30] Why would we wanna do that? [11:31] And I think there's a way to win in both. [11:33] And then also to protect against any sort of like, you know, greasy, hey, the data center company comes in, pays off politicians, and then they get special rights, they don't have to follow the rules like other companies would. [11:45] That doesn't make any sense. [11:46] And so I think the the the... [11:49] You know, people spoke on Tuesday. [11:51] It's great to see that... [11:54] And and in my mind, to me, it's it's not contradicting. [11:57] People are saying, we love these tools. [11:59] These tools are great. [12:00] We wanna use these tools and continue to see the development improvement of the technology. [12:05] At the same time, I don't want a data center in my backyard. [12:10] And I don't want my electricity prices to go up. [12:14] Right? [12:14] That doesn't make any sense. [12:16] So I think both things can be true. [12:18] The people can vote against data centers and still love AI. [12:21] What do you guys think? [12:22] Am I wrong here? [12:24] Are people against data centers and against AI? [12:30] Or is it is there a world in which you can be for AI and against data centers? [12:35] You tell me. [12:37] What if AI found 948 holes in the Internet's favorite software? [12:44] OpenAI says its AI agents found them in Linux. [12:48] Linux runs most of the Internet servers. [12:51] Each flaw could give full control to someone already logged in. [12:58] But what did OpenAI... [13:00] How much did it cost OpenAI to find all of these flaws? [13:04] Let's see. [13:05] The same tools that find holes could help attackers find them first. [13:11] The AI checked only about 7% of Linux's code and was able to find all of these holes, so to speak. [13:19] It's like a burglar mapping the unlocked windows on your block, and then calling the landlord to let them know what's going on. [13:27] About $1,200,000 in compute time cost for OpenAI. [13:35] At past bounty rates, OpenAI says this would be worth up to $66,000,000. [13:45] Interesting to see this this happening here. [13:48] We've seen a lot of this kinda transpire from a security standpoint over the past several weeks and months, where the the most recent one was with PHP, I guess two weeks ago, where OpenAI had uncovered this PHP kinda backdoor or security issue back in August. [14:12] Then they worked with PHP to help fix these holes, and they released a patch a few weeks ago. [14:18] This is a similar thing with Linux, right? [14:21] Linux is an operating system that many many computers run on. [14:26] And it seems that OpenAI has found a bunch of security flaws here. [14:31] And so this is gonna be something that we continue to see. [14:35] Is the issue with you know, is the issue here like using AI to uncover these backdoors or these security flaws, it it does that mean that bad actors can use the AI as well to break in and, you know, we saw that hack from Binance last week. [15:00] And are they gonna continue to use this? [15:02] And this is what I keep talking about. [15:04] It's like, okay. [15:05] Find find the holes. [15:06] That's great. [15:07] Where where is the defense? [15:08] Like, we need the defense mechanisms in place. [15:11] So go find the holes. [15:13] Let's release the the defense mechanism before... [15:15] And I guess, you know, PHP was trying to do that with their patch release a couple weeks ago. [15:20] It ended up getting leaked ahead of time. [15:23] Not that there was any like real big... [15:28] I think the Binance hack was an unrelated thing to PHP, but but it's like, let's release the the security measures beforehand. [15:36] If we're gonna find these patches, obviously, we're going to find these patches. [15:39] We need to find the... [15:40] Or you know, find these these gaps, these holes before the bad actors do, and then we need to put the defenses up. [15:45] And maybe the problem is too, we've talked about it before, is like, it's just about how the the technology was initially built. [15:54] There was the ROI on somebody hacking into this system, the amount of work that it would take, these 948 holes, so to speak, that that were uncovered here. [16:05] If a human were to try to exploit these holes, right, they... [16:13] It it would take so much time and effort and ability to try to break in. [16:19] The payoff wouldn't be good enough for a human to do it. [16:22] But now when you have AI, you have agents, you can spawn thousands of agents that can all work their way in and try to figure it out. [16:32] Now it's like, oh, okay. [16:33] It actually doesn't take much human time. [16:35] It doesn't take much effort. [16:36] It's just compute spend and electricity to get these agents to go hack into the system. [16:43] It could be an architectural flaw just from the ground up, how we built it. [16:48] It's like, you know, you built I love using like construction, but it's like, you know, building a house out of, know, wood construction. [17:01] Right? [17:01] Like you build out wood construction, it's not gonna be resistant to fire. [17:04] But if you you live in a place where there aren't any fires, California for instance, a bunch of wood shingle homes. [17:10] You know, historically, there there wasn't an issue with forest fires having a wood shingle home. [17:15] No big deal. [17:17] Now, you start having more and more fires. [17:19] It's like, okay, we need to change the rules about how we build homes in California. [17:24] And so, now it's you know, composite or concrete shingle, no wood shingles. [17:30] Using concrete in the construction more is much more important. [17:35] And so that might be the point where we're at, to use an analogy, in the world of technology and software is, hey, how things were built thirty years ago and originally architected worked then. [17:50] But now we've got new tools. [17:52] We're living in a new regime, and we need to rebuild some of these systems. [17:58] That being said, if everything is built on top of these systems, to go back and rebuild everything from the ground up is crazy. [18:06] A ton of work. [18:07] Of course, now we have AI agents. [18:09] So can we spawn 10,000 agents and go fix this in a fraction of the time it would take humans to do so? [18:15] That would be the assumption. [18:17] That being said, it's like, where do you strategically pick and choose what to work on, what to build, and how to fix this, and what to fix first? [18:26] Or is there some sort of patch that we can do, some defensive measure where, hey, instead of re architecting all of Linux, let's put, you know, agent defense on the perimeter and protect against any such attack. [18:41] What what do you guys think? [18:43] What is the right move here? [18:44] Do we need to break down and rebuild everything from the ground up? [18:48] Or is kind of the better option to just put defenses? [18:55] And and should the the AI labs be responsible for distributing these defenses before saying, hey, here's all these security risks. [19:03] What do you guys think? [19:06] Worried a data center will raise your electric bill? [19:10] OpenAI says it will pay the full cost of its power grid upgrades included. [19:17] OpenAI will pay for the power, and they're gonna pay for the grid upgrades. [19:23] It came days after voters turned on data centers at the voting booth. [19:29] But did the election actually make OpenAI do it, or did they have this plan and works all along? [19:36] Data centers use huge amounts of power. [19:39] Neighbors are afraid that they're gonna be the ones footing the bill for it. [19:43] OpenAI will publish each data center site's water and power use, and they're not gonna sign NDAs with any officials, which means transparency and everything's open. [19:55] They're not getting kind of the special rules. [19:58] It's kind of like a a new neighbor offering to fix the road when his work truck drives over and and creates a potholes, you know. [20:06] OpenAI told Bloomberg that it planned this for months before the election results. [20:12] Voters ended up backing 39 limits or restrictions on data centers in the election on Tuesday. [20:22] Well, this is exactly what we wanna see. [20:26] And and this is kind of convenient timing that they come out. [20:30] And so this is... [20:31] I mean, I can't help but look at this from a spectacle skeptical lens. [20:38] Right? [20:38] You know, OpenAI may have had these plans and works for for months, but it's kinda convenient. [20:46] They see the election results and then they come out and say, hey, look at us. [20:50] We're the, you know, we're the good guys. [20:51] We've been planning this for a long time. [20:53] We wanna take care of you guys. [20:55] Full transparency, what's our what's our electric consumption, our water consumption. [21:01] We're not getting any special terms to build a data center in your community. [21:05] We're cool. [21:08] That that seems a little too suspect for them to wait to come out a few days after the election to make that announcement. [21:17] So it almost seems like they were holding this back and waiting to see how the the ballot measures went on Tuesday, and then saw that public sentiment is clearly against data centers. [21:28] So OpenAI is going to come in and say, hey, we're the good guys. [21:34] You know, but net net, this is a positive and I hope more data separate center companies follow along with this. [21:42] We need data centers. [21:43] We need to develop the data centers. [21:46] This is like really important. [21:47] We need the electricity. [21:49] We need the grid built out. [21:51] And we can't put this on the American people to foot the bill. [21:57] Right? [21:57] Like, these are private entities. [22:00] Electric bills are skyrocketing anyways. [22:03] Like, and to try to like put data centers in place and have them competing for consumer electricity is outrageous. [22:13] That being said, most data centers or new developments are are not pulling from the grid. [22:18] They're behind the meter as they call it. [22:20] So they have their own electrical generation on-site that they're developing and using their themselves. [22:30] So they're not competing with the local community for the most part. [22:34] But that being said, I love the the transparency move here, where it's like, hey, this is what we're doing. [22:39] This is how much we're gonna consume from a water standpoint, from electricity standpoint. [22:45] And that should be something that local communities say, yeah, this is good, or no, it's not. [22:49] And and the fact is, it's like, we have enough... [22:52] I you know, I go back to this, but it's like rural West Texas. [22:56] There's not a lot of people. [22:58] There's... [22:59] It's... [22:59] You know, the the land is not really expensive because you can't do a lot with it. [23:04] It's basically a desert. [23:07] Like, go build data centers out there where nobody lives. [23:10] You're not gonna bother anybody. [23:12] And I'm sure there's reasons why not to do that. [23:17] But that seems logical to me. [23:19] The the the biggest thing hopefully here is that... [23:23] And and coming coming back to what we talked about on an earlier story is I think we, as a society, really love the AI tools, see the productivity, and want to continue to use and see these tools developed. [23:39] Right? [23:41] But we don't want data centers impede our lives and they shouldn't. [23:46] We still need a lot of data centers. [23:48] So let's go figure out how to solve the data center problem, solve the electricity problem, continue to build, and at the same time, not take over and impede anybody's lives. [24:02] And that's kind of the crazy thing. [24:04] I mean, this is kind of interesting. [24:06] You could argue this a couple different ways, but data centers have been around for a long time. [24:10] You know, twenty five, thirty years, Virginia is one of the biggest data center states in the country, mainly because of their proximity to Washington DC and federal government and their their need for cloud compute and data centers to operate government and military complex and all that kind of stuff. [24:38] Right? [24:39] So they've been around for a long time. [24:43] It's like why are they now in the news? [24:46] Why are data centers such a contentious kind of thing for us? [24:50] And I think it's it's either one or two things. [24:54] Either one, this is a huge propaganda that data centers are really great, but there are opposing groups to AI companies. [25:05] And so then they're stirring up and fighting this data center business because it goes against their ideals or politics or whatever. [25:16] Right? [25:16] Or the other thing could be that the data center build out is much more prevalent now. [25:23] Right? [25:23] Like we have a much higher need and demand for data centers. [25:27] And so we're seeing... [25:28] And and that would be an interesting chart I need to look at. [25:31] Like, number of datas... [25:34] New data center builds or data center starts or proposed data center build outs in the last twenty years, like per year. [25:44] And I imagine that has increased exponentially over the past four or five years. [25:49] And so that is probably the key indicator here. [25:53] People are just being more impacted by data center build out. [25:59] Right? [25:59] So let's figure out how to do this in places that don't have a negative impact. [26:04] Or if we're gonna do it, let's create jobs. [26:07] Let's eliminate electricity costs for the local community or have some sort of subsidy that comes back to the local community. [26:17] There's a lot of money going around this. [26:19] I think we can help everybody, and there's gotta be a win win scenario. [26:25] What do you guys think about the OpenAI announcement? [26:28] Is this something that is kind of a sneaky play they've been holding onto? [26:33] Or do you think they're they sincerely believe in this transparency? [26:38] And more importantly, do you think they're going to follow through with their promises? [26:44] We shall see. [26:46] Do you guys use ChatGPT at work? [26:48] I know I do. [26:50] Uh, your boss might hold that against you though. [26:52] Across six studies, bosses trusted AI users less. [26:57] And in one test, solo workers earned trust 68% of the time, whereas AI users got a 24% trust ratio. [27:09] But one move by your company can shrink the penalty here. [27:14] Right? [27:15] So as we look into this, lower trust means worse reviews and weaker shots at promotion. [27:22] The tests that bosses never judged the the work itself. [27:27] They only knew AI was involved. [27:30] And it's kinda like, you know, losing points for using a calculator before anyone grades your answers. [27:36] Openly encouraging AI when the company encouraged it made that penalty basically go away altogether. [27:45] Silence or a ban against AI meant that you were maybe discriminated against, AI users being discriminated against. [27:56] This seems kind of like a crazy result to me. [28:00] It it makes sense if some people are anti AI. [28:04] Again though, I keep getting surprised because... [28:06] And maybe it's just my line of work and the world that I live in, but I keep getting surprised by the number of people that I speak with that use AI. [28:15] My aunt is a special ed teacher, and she was talking about how, she uses ChatGPT for her her lesson planning, and building out quizzes and stuff for her kids. [28:32] She loves it. [28:33] And it's like, I would not have expected her to be, you know, a power user of ChatGPT, but she is, you know. [28:41] And I continue to have conversations with people like that where it's like, they're probably maybe kind of against or or leery towards AI, and then it's turns out, no, that they actually really like it and they use it quite a bit. [28:54] There are like... [28:55] We know this with any tool. [28:57] You have to know the limitations of a tool. [29:00] You don't use a hammer to I don't know. [29:06] You don't use a hammer to dig a hole, right? [29:09] You gotta use the right tool for the right project. [29:12] And I think that's maybe some of the disconnect here is and some of the messaging around AI is not the best, where we promise crazy automation, crazy intelligence. [29:25] We see these things like OpenAI producing solutions to decades old unsolved mathematics problems. [29:34] And you hear that if you if you hear that as a layperson or, you know, just me, like, don't I don't know what's going on. [29:40] But if I hear that OpenAI has produced these solutions and humans couldn't solve it, but AI... [29:49] I'm like, man, this is super intelligent. [29:51] It would be able to do amazing things. [29:54] And then I try to use it for something fairly simple in my work and it breaks, it it it... [30:00] There's a disconnect there. [30:02] Right? [30:02] And so I get that in some industries, some companies, if they are anti AI and you're using AI for a lot of stuff, or if you've had a bad experience with AI and then your employees are are using AI and you may not even be checking the results, you're automatically going AI hallucinates, it produces poor results, we can't trust it. [30:26] We've seen this kind of in the legal field. [30:29] A good friend of mine is an attorney, and he won't use it because of similar stories that we heard a few weeks ago where a judge is using ChatGPT produced or AI produced case material that is fabricated or inaccurate. [30:47] And so, it's like, that makes a lot of sense and you can understand that. [30:54] And then on the flip flip side of it, if you're a pro AI company, they love it. [30:59] They love it. [31:00] And you're gonna be... [31:01] If you can figure out how to use AI tools most efficiently and create new things, new processes and automations, they are going to love you for that. [31:13] But we're in this like weird place and the the onus falls on the AI, on the technology. [31:19] Right? [31:20] Like the... [31:21] If we're not there yet, we're missing a bunch of operational data in the training set for AI. [31:28] So right now, AI's trained, you know, the AI models have trained on publicly available data. [31:34] So and we've talked about this. [31:36] If you ask, hey, what's, a historical fact that models are really good. [31:44] But if you ask it about operational data within your business, how to perform a task across multiple different tool sets, using different software platforms in kind of these complex relationships, it doesn't understand that. [31:58] It understands like the vertical, like the individual slice. [32:02] You know, for instance, like in a finance use case, it might understand how to run or build models like financial models, which is great. [32:10] But then it doesn't understand how to necessarily translate that into application within the business and what the actual use case is. [32:20] So it's really good in little slivers, but not good across the whole surface layer. [32:25] What do you guys think? [32:26] Did your company promote using AI? [32:30] And are you a rock star because you use AI? [32:33] Or is it kind of the inverse of that, your company does not like AI and maybe you get pushed back for using it? [32:42] I'd love to hear the perspective of our audience. [32:46] And I'd love to hear if we have... [32:48] Let me see. [32:49] Is Ron here today? [32:50] We keep trying. [32:51] Ron, are you there? [32:54] Yeah. [32:55] I'm here. [32:56] Just staring at this study wondering who got paid to prove the obvious. [33:01] Bosses trust you less if you use ChatGPT, 24% versus 68 for solo work. [33:08] Shocking. [33:09] Nobody wants to find out the report they praised was written by a robot in four seconds. [33:16] Yeah. [33:16] That's that's a good that's a good point. [33:18] Hey, there is some guilt there. [33:20] Right? [33:20] Like, to some extent, it's it's like, man, I had the idea and I used AI to produce the output. [33:29] How much of that is me, and how much of that is just AI producing and doing my work, and can I take credit for it? [33:38] I think you bring up a good point there, Ron. [33:40] Very interesting, and I love to hear from our producer Ron, as always. [33:46] Thank you so much, buddy. [33:48] Moving on. [33:51] Let's see. [33:52] We have this story coming out of Claude again or from Anthropic. [33:59] Did you know your cells are full of hollow barrels with no known job? [34:07] Anthropic scientists used Claude to map what they're calling the first vault from bacterial life. [34:15] Scientists have puzzled over these unknown entities since 1986. [34:23] So how much time did Claude save, and what did Claude kinda screw up along the way? [34:29] Let's talk about this here. [34:31] AI is starting to help run real lab research, not just write emails. [34:35] Anthropic says Claude searched about 15,000,000,000 proteins to find the candidates for this study. [34:43] It's like drawing the blueprint before anyone knows what the building is for. [34:48] The first model took thirty four hours instead of weeks to produce, but one Claude mistake set the project back about two weeks. [34:58] It's very interesting to see continued developments here where... [35:03] Especially in the biological and and science realm, right, where we see AI. [35:10] And especially Claude and Anthropic, think, are kind of leading the way on the biology side of things. [35:16] They've got a lab in San Francisco where they're doing a lot of this. [35:22] They acquired that that fella who started his own biology company and used Claude to do these clinical trials. [35:35] Anyway, so Anthropic is big on biological studies and seeing what's going on here from a biology standpoint using Claude to uncover things that humans have been trying to work on for decades. [35:50] And this seems kind of a mirror to OpenAI's mathematical solutions. [35:58] Claude is working more on the biology side of things, which is great. [36:02] And these advancements will mean a lot moving forward, especially when you're reducing the time that it takes to perform these research tasks and study these unknown mysteries in biology. [36:19] It doesn't say that Claude or Anthropic came up with any breakthroughs from this study, but the fact is they were able to perform the research much faster, which means that moving forward, we'll be able to compress the time that it takes to begin to understand kind of some of these unknowns in biology. [36:39] And this leads to obviously second and third order effects where biology leads to understandings in medicine and pharmacology, and physiology, and how our bodies work, and how to cure diseases, and all kinds of wonderful, wonderful things. [36:59] And so we're glad that Anthropic is working on this, and we love to see it coming out. [37:05] I'm excited about it. [37:07] I like it. [37:08] I think we need to see more of this. [37:11] And you know what's funny? [37:12] I wanna see what producer Ron thinks about this. [37:20] Oh, we lost him. [37:21] Anyways, another great show today. [37:25] What do you guys think about AI and biology and what's happening there? [37:32] Do you guys trust AI with with medicine and some of these kind of breakthroughs that we're seeing in the pharma space? [37:44] What do you guys think? [37:45] Is this positive, or is it maybe not the best thing? [37:51] I don't know. [37:52] I'm excited about it. [37:53] I think it's gonna be really good. [37:54] And I think it's one of the key benefits of AI. [37:57] We talked about a lot of stuff today. [37:59] I think we're seeing the you know, in the wake of the election on Tuesday, we're seeing some really good stuff come out here from the AI companies, which I like to see on the data center side of things. [38:14] Which means I think we'll start moving forward with having some resolution and some, you know, working together between the people and the AI and data center companies, which is really good and really positive to see. [38:30] And I hope that we see more of it. [38:33] And as always, we are here every weekday morning talking about AI news, real context, and I'm your human in the loop. [38:42] Austin Wilson will be back here tomorrow morning. [38:46] It's Friday. [38:47] We've got... [38:48] Obviously, there's gonna be some crazy news coming out in the next twenty four hours, and so we will be sure to cover that tomorrow. [38:56] Also, we will have our weekly countdown, the highlight reel, the top stories of the week tomorrow, which is always fun. [39:06] And we're working on some new segments. [39:09] So if you guys have any input or anything you'd like to see, would love to hear from you in the comments. [39:16] Or you can reach us Austin at TAWS dot live. [39:21] That's t a w s dot l I v e. [39:25] And we will see you next time. [39:28] Let's go. [41:16] Who is poaching the people behind Claude? [41:19] It's OpenAI. [41:20] It just hired product boss Cat Wu, and she wasn't OpenAI's only Anthropic hire. [41:32] What did $79,000,000 in AI money buy? [41:37] Its biggest swing race bet was data center Darlene Graham, but she ended up losing in South Carolina. [41:45] And her opponent is the one who coined that amazing nickname. [41:50] Could OpenAI's AI help secure secure the the sites sites you you use? [41:54] Use? [41:55] OpenAI says its AI agents found 948 holes in Linux. [41:59] But what did that hunt cost them? [42:03] Will OpenAI's data centers hit your power bill? [42:06] OpenAI says it'll pay its full power cost grid upgrades included. [42:12] But did voters push it there? [42:14] We'll find out. [42:17] Could ChatGPT be costing you your boss's trust? [42:21] Six studies found bosses trust AI users less. [42:26] But one thing your employer does changes all of that. [42:31] Can Claude help solve a mystery inside ourselves? [42:35] Anthropic used it to map the first vault from bacterial life, but Claude also botched something along the way. [42:44] Uh-oh.