Trading the Markets With AI: Claude Leak, AI News Takeover

0:05Kris Bullock:Happy Wednesday, everyone. Welcome back to another episode of Trading the Markets with AI with Kris Bullock, a.k.a. BlastoPlast. We've got quite a show for you today. It's the AI space, so there's like breaking news seemingly every few hours. But we're going to talk about all of that because OpenAI just closed the largest fundraise in history at$122 billion with a valuation of$852 billion, basically making it like a small nation state. We're going to break down what that means for the industry and for investors. And then Anthropic is having a little bit of a rough week. Their source code for Claude Code was leaked publicly.

0:55Kris Bullock:And this is actually the second leak in as many days with Anthropics. So we'll get into what that means as well. We're also going to talk about something that might make you question everything you're reading. A study shows that the New York Times allegedly, I have to say allegedly here, has been publishing AI-generated content. Then we're going to move into the physical world and talk about Rivian because they are putting a full AI assistant and autonomous driving system in their vehicles. And we're going to talk about what this could mean for Compute going forward. We're also going to check back in on Chris Bullock's dashboard, which we debuted a couple of weeks ago, and see how that's looking.

1:42Kris Bullock:He's got some updates for us. And then we're going to talk about something that I think is a little bit funny, captures a little bit about where AI is right now. A realtor used AI to clean up a listing photo and accidentally publish a demonic figure. We're going to show you that photo and talk about it. Plus, can AI make you a better chef? We're going to talk about that today, too. Happy Wednesday, Chris. How are you?

2:10Bijan Maleki:Doing well, doing well. So, yeah, the Anthropic story, it's blown up this week, basically. And kind of to recap, what happened is Anthropic accidentally exposed basically their full source code for their flagship Cloud Code CLI tool. Not like everything, but just the primary CLI tool. They mistakenly attached it to a package and it got out there. It leaked. And what's interesting that happened was the leak happened. And at first I thought it was kind of fake news because I saw it coming around on Twitter and I'm like, what is this? And then people were posting sort of satirical posts about it and things like that.

2:48Bijan Maleki:And so I kind of dismissed it at first. But then I stumbled on this post here. And this one was what caught my eye. And this was interesting because the fork happened, or rather the leak happened, which was bad. But then this guy, this developer in Korea basically took the code and essentially reworked it, like rewrote it in a different language. He took the whole code release, rewrote it in Python before anybody could catch it and posted it on GitHub. And so when Claude tried to issue like DMCA sort of takedown notices, they couldn't take this down because it was deemed like new creative content that wasn't their code.

3:32Bijan Maleki:And so now it just exists and it's out there and it's for anybody to access and it can't be taken down because it was rewritten in a different language, which is interesting in and of itself, I think. That the person that saw this was clever enough to take it, immediately rewrite it and post it on GitHub. So that was big. And that's like that was interesting. So I kind of dug into this. I was using Grok. I was using a bunch of stuff to break this down and figure out how much of this is actually true. What's real and what's not? What's relevant here? How big of a deal is this really? And so I spent quite a bit of time digging into this.

4:05Bijan Maleki:And it is a pretty big deal. It's definitely a very big deal for the open source world. Like it's a huge win for open source enthusiasts because now there's a free full version of this, you know, alternate version of the of Claude basically, you know, available for them to view, for them to inspect, for them to fork, for them to improve. And it gives them a really solid breakdown just on how Claude was operating and structuring a lot of their things like that. As for Anthropic themselves, yes, it's a big deal. It's certainly not catastrophic. It's not existential. It's not going to take them down or anything like that.

4:39Bijan Maleki:It was not their whole code base. It didn't include their weightings or anything like that. It was just their cloud code CLI tool, like this client side CLI tool. So it was a portion of what they do. It gave insight into how they operate. But it wasn't like the keys to the castle necessarily. So embarrassing for sure. Problematic, yes. Really good for, again, the open source community. But yeah, fun, interesting, pretty wild, really.

5:09Kris Bullock:yeah definitely not great for claude that's uh no for sure and i think claude needs to um figure out where these leaks are coming from and and really patch that up because um it'd be one thing if it was just this which is bad enough but this isn't the first leak uh that they've had this week even right um so i think i think they said sorry they came out and said it was sort of

5:34Bijan Maleki:human error it wasn't like a like a corporate espionage type scenario it was just like an accident.

5:42Kris Bullock:Negligence is not good either. Chris, speaking of big deals, we had the biggest deal in private company history just happened yesterday. It just came out yesterday. We talked about this in the intro that OpenAI raised$122 billion at an$852 billion valuation. which in and of itself is a big headline. But where I think it starts getting really interesting is that now OpenAI is going to be included in ARK Invest ETFs. So now you're giving public market exposure to a company that's still private. So I think that's also very interesting. OpenAI is one of the first ones that people think of in the whole LLM AI space.

6:37Kris Bullock:And the fact that you're now going to be able to pick those up in the ARK Invest ETF, I think is huge news, right? Because OpenAI is basically prepping us for an IPO, it seems like, in real time with this. So pretty big news. And also when you're growing at this stage, I read somewhere that at this stage, OpenAI is growing at four times the rate that Meta or at that point, Facebook and Alphabet that were growing in their same stage. So they're not a startup at this point. They're literally inevitable in this space.

7:11Bijan Maleki:It's kind of interesting to me. What does this say about public markets that companies like OpenAI and like Anthropic can come along, generate billions and billions of dollars in rounds and then still not be a public company, still not IPO'd? And you have to wonder, as soon as they do IPO, what's going to happen to all this money that's already in there? Like, is it just going to be all dumping on retail, you know, like a new token launch does? I feel like something, the system is broken here to some degree. Like the fact that we've got, you know, public companies that exist that trade at a fraction of these valuations.

7:48Bijan Maleki:Meanwhile, we've got private companies that are raising literally trillions of dollars practically in earnings, you know, in rather funding rounds before even going public. You know, something's not right here.

8:03Kris Bullock:Yeah, it is interesting. And like, I mean, no, open AI, they, the article also says they're, they're, they're bringing in$2 billion a month in revenue. They, their ads, which, you know, was met with a lot of controversy in six weeks, it brought in$100 million of ad revenue to all this to say that I don't know if open AI is actually profitable, which is interesting if you're making that much money, right? Where is it going? I guess, to the energy that is needed for all of this, to the very high salaries for your AI engineers. A lot of things can go into paying back debts. Interesting for sure. Yeah, all the compute power.

8:46Kris Bullock:Yeah. Interesting for sure. But I mean, if this is a sign, I mean, if people are saying Anthropic is better, can you imagine what they will raise?

8:56Bijan Maleki:Yeah, it's so weird. I really want to see how this plays out when they actually do IPO. And honestly, I wouldn't want to touch either one of them. Even though I both think that they're good companies, I think they're both good companies. The structure here is backwards. And yeah, they'll continue to do well, but it just seems like typically companies go public in order to generate this kind of money, not like generated in private funding rounds and then go public. So yeah. Yeah.

9:26Kris Bullock:Well, we'll keep an eye on it and bring you the latest as that story develops for sure. But speaking of, you know, kind of AI and, you know, doing some crazy stuff, Rivian and Volkswagen are bringing AI into vehicles now. Chris, what's going on there? Yeah.

9:45Bijan Maleki:So I've been talking about edge compute for a while now in both this show and the Monday show, how one of my main investment thesis is going forward is that we're going to see a general transition away from the centralized cloud-based compute, like these central models, like ChatGPT and Anthropic. And of course, they'll still be used, but we're gonna see a lot of that compute moving out to edge devices. And what I mean by edge devices are our phones, our cars, our desktop computers, our kitchen appliances. Everything is going to sort of have AI compute power built into it so that it can process its own, have its own intelligence basically built in at the edge device so that it takes some of the need away from having these massive data centers and these massive compute power engines floating in the sky.

10:34Bijan Maleki:We can, because really to run a local model on a little device, you don't actually need a ton of compute power. But when you have a billion people trying to hit it all at once, yeah, you need a ton of compute power. So like you can process a lot of this stuff locally. And so I think we're going to see a transition to that. And as such, there's going to be massive hardware requirements and things like memory and storage and local processing and things like that that are going to be required to build out these devices. And so that's reflective in a lot of the semiconductor stocks and things like that.

11:04Bijan Maleki:But anyway, I digress. A perfect example of this that we're seeing right now is the likes of Volkswagen and Rivian building out just that, building out local inference in their cars, you know, so that they have AI built in that can run diagnostics, that can optimize your engine, you know, settings and things like that. your engine ECU settings to optimize fuel efficiency, to optimize performance, you know, that can be processing all this stuff in real time and sort of redefining, I guess, what a smart car actually is. And then Rivian is actually taking it a step further. And they've really built in a full on in-house AI assistant right into their infotainment center, you know, they're right in their dashboard.

11:49Bijan Maleki:And so kind of, and rather than just like bolting on something and having it talk to Gemini or whatever, they've actually, they've built out their own framework. It basically controls the whole nervous system of the vehicle. It has a custom processor and it not only manages, again, vehicle diagnostics, it can control anything. You can basically just talk to it. It can integrate with your accounts, your apps, like your Google calendar, your email, if you wanted to. And then you can just talk to it. You can be driving around and be like, hey, find me a coffee shop that sells XYZ, you know, and it'll just use its built-in AI to do whatever.

12:29Bijan Maleki:And or, you know, it's cool. You're going to have your own AI system built right into your car that you can drive around and talk to it, and it'll do any kind of processing you want. And all of the compute power and inference is occurring locally right there in your car without a cloud connection. And I think that's really cool. I think that's the future. It really makes me want a Rivian at this point. And I think it's only going to continue. I think we're going to start seeing this more and more. I think that all cars, you know, Rivian and I mean, Tesla started with their sort of main central operating system, you know, where the car is controlled by like one big computer.

13:03Bijan Maleki:You're essentially driving a computer around on wheels, you know, whereas prior to that, cars were just like these dumb machines and they had all kinds of little computers from all over from different vendors. And none of them really had centralized management. And so Rivian has taken that and, you know, they're working on the same thing and they're building it out and it's going to be the future. It's going to be what all cars eventually, I think, coalesce to is this main central computer. So we're essentially driving around, driving around a computer on wheels, you know, and it does everything all from one centralized spot and it has AI built into it.

13:35Bijan Maleki:And honestly, I can't wait. I think it's, I think it's great. I think it's super cool. And I, I'm looking forward to it.

13:41Kris Bullock:Yeah, definitely cool. Your move now, Tesla. But is this so when Jordy Visser at the beginning of the year when he was saying that 2026 is going to be AIs moving from software to a hardware world, is this basically what he was implying? Things like this happening?

13:59Bijan Maleki:Absolutely. A hundred percent. And I mean, the same thing with Tesla, the robo taxi. That's that's, you know, edge compute AI. Rivian also is building in autonomous driving. They've actually signed a deal with Uber to utilize their new R2 vehicles as robo-taxis as well, using the same general concept that Tesla does. So this is absolutely a transition to humanoid robots, edge compute, things like that. All of that is going to be local inference is what you can call that. And that is the evolution of AI. And I think that for people who missed the initial, you know, buying the MAG7 stocks, the initial AI run of a lot of those early stocks that ripped, you know, multiples, if you want to still get in on this and have exposure to AI, the next evolution of that is to focus on companies that are building this localized compute, that are focusing on edge compute, local inference type technology that are building the semis for that, the memory for that, the storage for that.

15:04Bijan Maleki:the glass for that, the plastics for that, all of those companies are going to, that's where everything is going. And that's where you need to focus on right now. Yeah.

15:13Kris Bullock:Chris, you said robo taxi, and I couldn't help but think you have a very recent and intimate experience with just that, right? How was it? Were you scared? Was it a smooth ride? Tell us a little bit about what you did.

15:29Bijan Maleki:Yeah, yeah. So I rode in a Waymo for the first time last week. And it was really cool, actually, I got to say. I honestly felt a little bit safer in it because, like, I had a contrast in that I traveled last week. I took an Uber from the airport to where I was staying. You know, a human drove it. A human picked me up. And, I mean, not to disparage humans or drivers or Uber or anything like that, but, like, it's a human driver. And, you know, a person that's been driving all day long in busy traffic, they're going to have mental fatigue, especially if they do this all day, every day. It's just a it's just a matter of fact, you know.

16:07Bijan Maleki:And so needless to say, the ride was a little herky jerky. And when I got into this Waymo, it was the complete opposite. Not only it's like it follows the speed limits everywhere it goes. It has a preplanned route. But not only that, the cool thing with the Waymo, and I'm sure the Tesla Robotaxi is the same, but it on its display, both in the main dash and in the center console from the backseat, is a big display that shows the full radar scan of everything that the Waymo sees. So it sees all the traffic in all directions at all times. So it can measure things like where vehicles are coming from, the closing speed, if the car in front of you has its brake lights on, what cars are turning, which cars aren't, everything that's going on all around.

16:50Bijan Maleki:And I thought to myself, you know, no human, even a human who is completely, you know, refreshed and energized has that level of situational awareness, has that wherewithal where they can monitor everything in 24, you know, in 360 all around them at all times, you know. And so I was like, wow, this is actually really cool. And I actually feel safer in this because this car knows everything that's going on around me everywhere, you know. And I could see it visualized on the screen. and yeah, it knew all the traffic rules. It knew like, you know, when to turn, when to use signals, when to turn left, when to turn right.

17:25Bijan Maleki:It didn't do anything that even gave me pause that I was like, how's this going to work? Or, you know, there were a few things where I was like, oh, this will be interesting. I want to see how it does. And it did great. It did exactly what it was supposed to do. It not only didn't know where to go, it knew the most efficient route. It knew all the traffic rules. It didn't do anything scary or weird. It had cars going all around it, passing it, and it didn't freak out. It would just drove. It got me right to where I needed to be. I'm in the time that it told me that it was going to take. Um, it was super easy in terms of like, I had luggage in the back.

17:56Bijan Maleki:You just get in, you pop the trunk, throw the luggage in the back. It knew then when it got there, don't forget your luggage. It pops the trunk automatically as soon as it parks. Um, yeah, it was just, it was, it was really cool. Um, and it was like half the price of, of, you know, an Uber. And I didn't have to leave a tip of course, because it's a robot. So So I'm like, that's tough. You know, like I get it. I mean, you had my attention.

18:22Kris Bullock:Now you have my curiosity.

18:25Bijan Maleki:I know, I know. And so I don't know if, you know, I don't know that it can work everywhere. I don't know how comfortable I'd feel in, say, a winter environment, like on slippery roads, you know, or something like that. So I think some cities it works better than others. I happen to be in Phoenix and Phoenix is just a big, wide open city with big, it's just a big grid. There's no wonky turns and hills and windy road. It's just an easy grid, flat, you know, big roads, easy to get around. So it made sense there. I can see where in other scenarios it may be a little trickier. But yeah, for some places I can see where it's absolutely the future.

19:08Kris Bullock:Oh, it's coming. That's for sure. So let's switch gears a little bit here because definitely want to touch on this New York Times story that is developing. Allegedly, I have to say, you know, I'd be remiss. So allegedly, New York Times has been caught publishing extensive AI generated articles. You know, if you're you're someone that is worried about AI slop, you know, seeping into the creative world and now, well, apparently you should be worried about it. seeping into journalism as well. Of course, again, this is all alleged, but I think where this becomes a problem is a lot of people might, they trust these, you know, these reputable news sources.

19:55Kris Bullock:So they might not even have any idea that what they're reading is AI generated content or an article. And I think that has the implication for trust and for anyone who kind of, you know, wants signal and something that's actually like truly vetted as opposed to just like, you know, kind of noise. Chris, what's your take on this?

20:15Bijan Maleki:I do want to add a little bit additional detail to this. It wasn't just the New York Times. Also things like the Wall Street Journal, the Washington Post, et cetera. And it was generally more specific to opinion pieces as opposed to like factual news content. But it was, you know, yeah, so it was generally more relegated to that. So at least it's not generating facts, I guess you could say, or reporting news per se. It is more sort of opinion-based pieces. But yeah, a fair amount of that content is, you know, at the very least, you got to figure people are like putting their ideas out and running it through LLMs to refine things and improve things and spiff things up, I guess, however.

21:05Bijan Maleki:But yeah, I don't know how I feel about it. Like, on the one hand, I feel like it's going to be a thing. It's not going away. People are going to use this because it is going to make journalists work more efficient. But there's definitely a very gray area. There's a fine line between where does this become, where does this go from being a journalistic opinion to AI slop, like you said. And they're definitely walking the line with it for sure. But I think we're only going to see more of it, I have to believe. And I'm sure it's probably not just these guys. You got to think it's mass everywhere.