THIS WEEK IN AI: Google's Big Change, AI Keeps Breaking Out, OpenAI vs Apple

This week, we cover major leadership changes at Google’s AI teams, including Demis Hassabis stepping back and Jeff Dean leaving to start Discovery Loop, an AI company focused on scientific research. 

We also discuss AI safety issues involving frontier models, plus Meta’s new coding model and its strategy around efficiency.

------
🔒 Check Out Our Sponsor: LEDGER AGENT STACK 🔒
https://developers.ledger.com/docs/ai-tools/overview/?utm_source=Audio&utm_medium=Podcasts&utm_campaign=Limitless

------
🌌 LIMITLESS HQ ⬇️

NEWSLETTER:    https://limitlessft.substack.com/
FOLLOW ON X:   https://x.com/LimitlessFT
SPOTIFY:             https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQ
APPLE:                 https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890
RSS FEED:           https://limitlessft.substack.com/

------
TIMESTAMPS

0:00 DeepMind Leadership Shift
3:20 Jeff Dean’s New Startup
6:05 Automated Research Ambitions
10:08 Agent Breakouts
13:06 OpenAI’s Hidden Incident
18:11 Meta’s Cheap Coding Model
22:55 OpenAI Versus Apple
25:42 DeepSeek
28:28 Mind Control Interfaces
30:40 Nikita Bier Departs X

------
RESOURCES

Josh: https://x.com/JoshKale

Ejaaz: https://x.com/cryptopunk7213

------
Not financial or tax advice. See our investment disclosures here:
https://www.bankless.com/disclosures⁠

Josh works with Anthropic as a contractor. All views expressed are his own and do not represent Anthropic, its leadership, or its affiliates. Nothing in this episode is investment advice.

Episode Transcript

Josh: Welcome back to the Weekly Roundup, our favorite episode of the week where we (00:00):
undefined

Josh: can catch you up on all things AI, frontier technology, all the crazy things (00:03):
undefined

Josh: that happened. And this week is no exception. We had the biggest shakeup in (00:07):
undefined

Josh: about 30 years at a company, a little company named Google. (00:11):
undefined

Josh: We had a counter lawsuit happen from OpenAI to Apple saying, (00:14):
undefined

Josh: wait a second, you're totally lying. (00:18):
undefined

Josh: We have updates for the AI breakout, so much so, like this story about where (00:21):
undefined

Josh: the AI broke out of the system, that some of these frontier companies like OpenAI (00:25):
undefined

Josh: are actually slowing down training progress because they're so kind of timid on how to handle this. (00:28):
undefined

Josh: Starting with Demis Asabas, man. I mean, Demis Asabas is gone. (00:33):
undefined

Josh: He's not working at DeepMind anymore. Not gone from Google, I should say, (00:36):
undefined

Josh: or better yet, Alphabet. (00:40):
undefined

Josh: But Demis Asabas, in his current role as the CEO of DeepMind, (00:41):
undefined

Josh: is kind of no longer. And this seems like a pretty big deal because for those (00:45):
undefined

Josh: of you who've followed Demis for an eternity, this has basically been his thing. (00:49):
undefined

Josh: He's been in charge of DeepMind. Google acquired them. (00:52):
undefined

Josh: Now he is within Google and now he is gone from DeepMind. So what on earth is (00:55):
undefined

Josh: happening here? I read this and I was like, wait a second, Demis stepped down from what? (01:00):
undefined

Ejaaz: I think everyone was blindsided by this news yesterday. (01:05):
undefined

Ejaaz: So Demis Hacillus, just for context, is like one of the godfathers of AI. (01:08):
undefined

Ejaaz: He kind of like grandfathered a lot of the ML research at Google, (01:13):
undefined

Ejaaz: even prior to Google, like he was heralded as one of the top AI researchers. (01:16):
undefined

Ejaaz: And he's led Google's AI efforts on Gemini and everything else for a number of years now. (01:20):
undefined

Ejaaz: He announced yesterday that he's stepping down from the role of CEO of DeepMind (01:25):
undefined

Ejaaz: to become the chair of DeepMind. That's a (01:30):
undefined

Ejaaz: Indirect way of basically saying I'm throwing in the towel and I'm done here. (01:32):
undefined

Ejaaz: And what he's now going to do is focus on wider societal impacts of AGI and (01:36):
undefined

Ejaaz: try to figure out how Google can kind of like deal with that, (01:41):
undefined

Ejaaz: as well as wearing an alternative hat, a new role where he's going to be focused (01:44):
undefined

Ejaaz: on something called Isomorphic Labs, which is a company he founded and spun (01:48):
undefined

Ejaaz: out of DeepMind, which is essentially an AI lab which creates designer drugs. (01:52):
undefined

Ejaaz: So this is a big shock and news for everyone because Google, (01:58):
undefined

Ejaaz: as you may know, was falling behind in the AI race. They had like a kind of (02:01):
undefined

Ejaaz: lead at the end of last year. (02:06):
undefined

Ejaaz: And then since then, they never really released a model that was noteworthy. (02:08):
undefined

Ejaaz: Anthropic, OpenAI, and in some cases, even XAI and Meta have now caught up and (02:11):
undefined

Ejaaz: surpassed Gemini 3.6 Flash, which I believe is their latest model. (02:16):
undefined

Ejaaz: He's being replaced by a guy called Corey. He is currently Google DeepMind's (02:20):
undefined

Ejaaz: CTO, and he's being now promoted to the senior vice president role. (02:24):
undefined

Ejaaz: He has a lot of experience in AI, but that's not even the major news about this Google stuff, Josh. (02:28):
undefined

Ejaaz: I don't know if you saw it, but the grandfather, the godfather of Google DeepMind (02:34):
undefined

Ejaaz: and Google in general, his name is Jeff Dean. (02:40):
undefined

Ejaaz: He was employee number 30 at Google. He's been there for, I think, around 27 to 29 years. (02:43):
undefined

Ejaaz: He's dedicated his life there. He's responsible for creating some of the most (02:49):
undefined

Ejaaz: amazing and major and pivotal products at Google, Google Search, (02:52):
undefined

Ejaaz: the TPUs, Google Voice, anything and everything he has touched is leaving. (02:56):
undefined

Ejaaz: And he's taking with him one of the most amazing things (03:02):
undefined

Ejaaz: Senior fellows at Google. I believe his name is Gemawatt. I don't know if I'm (03:05):
undefined

Ejaaz: pronouncing that correct, but Sanjay Gemawatt. (03:10):
undefined

Ejaaz: And they are co-founding with two other Google employees, a new startup called (03:12):
undefined

Ejaaz: Discovery Loop. Josh, do you look into this company? It's a pretty awesome company. (03:16):
undefined

Josh: The company is incredible. Before we talk about the new company, (03:21):
undefined

Josh: I want to talk about the old company because Jeff Dean is like, (03:24):
undefined

Josh: it's hard to overstate how legendary this man is. (03:26):
undefined

Josh: And for reference, there are levels at these companies that go from level one (03:30):
undefined

Josh: engineer, and this kind of dictates how you get paid. (03:34):
undefined

Josh: So when you join a large tech company like this, level one is kind of like entry (03:37):
undefined

Josh: level. That's at the bottom. And then you go to level twos, which are people out of college. (03:40):
undefined

Josh: Level threes often have like masters or PhD degrees. (03:43):
undefined

Josh: Level four is several years of like actually working there. (03:46):
undefined

Josh: Level five and six is kind of where most progression stops. Like I'd say the (03:50):
undefined

Josh: top 10% are probably sitting there. (03:54):
undefined

Josh: Level sevens are principal engineers, which are even smaller. (03:56):
undefined

Josh: And then eights and nines gets progressively more and more intense then there's (04:00):
undefined

Josh: level 10 which is google fellow that is in like absolutely legendary honor of (04:04):
undefined

Josh: a lifetime it's similar to earning like a, (04:09):
undefined

Josh: like nobel prize or something there's just like not many people who are fellows, (04:11):
undefined

Josh: then there's level 11 and level 11 is reserved for only two people those two (04:15):
undefined

Josh: people are jeff and sanjay they have climbed to the highest possible ranks at (04:20):
undefined

Josh: a research-led company that has been around for 30 years and is responsible for, (04:25):
undefined

Josh: Most of the internet as we know (04:30):
undefined

Josh: it, they've built the foundation of all of which we are working on today. (04:31):
undefined

Josh: Without their work, none of what the internet contains right now is possible. (04:35):
undefined

Josh: It's like, these guys are truly like remarkable individuals when it comes to this. (04:39):
undefined

Ejaaz: Can we please reference this insane slide on his deck? (04:43):
undefined

Ejaaz: He talks about the experience of everyone and he basically has listed every (04:49):
undefined

Ejaaz: single product at Google that has been created over the last 30 years. (04:53):
undefined

Josh: It's unbelievable. These guys are truly incredible. And it's worth noting the (04:58):
undefined

Josh: slide deck itself, which is really funny. (05:01):
undefined

Josh: There was a great post by Signal, which I love, that I wanted to share here. (05:03):
undefined

Josh: And he was kind of making a testament to the deck where he said, (05:06):
undefined

Josh: if you spent time around Google, you know the slide design has carried some (05:09):
undefined

Josh: of the most legendary work in the company's history. (05:12):
undefined

Josh: And it's this very nostalgic thing where when you see those blue kind of tacky (05:14):
undefined

Josh: looking slides, you know that something special is about to happen in the world (05:18):
undefined

Josh: of the internet just because of all the legendary code these guys have shipped. (05:22):
undefined

Josh: And I mean, for VCs, I'm sure they're not going to even need to look at this (05:25):
undefined

Josh: pitch deck. They're going to give them whatever types of money they want for this new company. (05:28):
undefined

Josh: Now, the new company seems somewhat interesting. It's called Discovery Loop. (05:33):
undefined

Josh: And my base level idea of what this is, and correct me if I'm wrong, (05:37):
undefined

Josh: you guys, is they're basically figuring out how to use this recursive learning (05:40):
undefined

Josh: thing, which comes in the form of a loop where you send a prompt and the model (05:44):
undefined

Josh: continues to iterate until it reaches a very difficult answer. (05:47):
undefined

Josh: And they're playing to just scale that out in a very large way. (05:50):
undefined

Josh: So loops recently have become very hot. We've realized that they could do pretty (05:53):
undefined

Josh: novel breakthroughs with math and science. (05:57):
undefined

Josh: Their plan is to kind of expand that further and see how far you could take (05:59):
undefined

Josh: these recursive loops to make new discoveries. (06:03):
undefined

Ejaaz: That's right. So the core concept of the new company of Jeff Dean and Sanjay's (06:05):
undefined

Ejaaz: new company is they want to automate research. So what does that mean? (06:10):
undefined

Ejaaz: Basically, in any kind of scientific research sector, it is incredibly arduous (06:15):
undefined

Ejaaz: and like onerous. So what I mean by that is people need to design an experiment. (06:21):
undefined

Ejaaz: They then need to get all the equipment to run the experiment. (06:26):
undefined

Ejaaz: They then need to run the experiment. And then guess what? (06:29):
undefined

Ejaaz: They need to repeat that experiment a hundred different times to get any sort (06:32):
undefined

Ejaaz: of validation that the information they've produced is correct. (06:35):
undefined

Ejaaz: It takes years upon years. (06:38):
undefined

Ejaaz: These guys want to cut down that process completely using AI. (06:40):
undefined

Ejaaz: And the number one field that they want to start with is, of course, AI research. (06:44):
undefined

Ejaaz: So if you apply research loops to AI research, what do you get? (06:49):
undefined

Ejaaz: It's that recursive self-improvement that you just mentioned, Josh. (06:52):
undefined

Ejaaz: So what that means is you can create AI models or even AI systems that can re-evaluate (06:55):
undefined

Ejaaz: themselves and build a better version of themselves. (07:00):
undefined

Ejaaz: And why that's really important, why every single AI lab, including Anthropic, OpenAI, (07:03):
undefined

Ejaaz: And I guess no longer Google anymore is going after, is if they can automate (07:07):
undefined

Ejaaz: AI research, you can end up having the smartest, most effective model without (07:13):
undefined

Ejaaz: doing anything. It's not limited by the constraints of humans. (07:17):
undefined

Ejaaz: But what excites me most about this, Josh, is what they're planning to apply (07:20):
undefined

Ejaaz: this automated loop to after they've figured out the AI thing. (07:24):
undefined

Ejaaz: They've said, once we've done this, we're going to start with AI machine learning. (07:28):
undefined

Ejaaz: And then after that, we're going to focus on grand challenges. (07:31):
undefined

Ejaaz: We believe our approach will be able to solve any learning loop with measurable (07:34):
undefined

Ejaaz: outcomes in the domains of science and engineering. (07:38):
undefined

Ejaaz: Ultimately, we want to build systems capable of taking on the National Academy (07:40):
undefined

Ejaaz: of Engineering Grand Challenges, such as better medicines, advancing health, (07:44):
undefined

Ejaaz: informatics, making solar energy economical, and providing clean access to water. (07:48):
undefined

Ejaaz: Basically, if you can master this playbook, this autonomous research loop, (07:53):
undefined

Ejaaz: you can apply it to any scientific, mathematical, or physical field, (07:57):
undefined

Ejaaz: which means that we're going to have like advances in any or every field of (08:01):
undefined

Ejaaz: science if this company can pull it off now (08:05):
undefined

Ejaaz: If this was any other roster of team members, I would be doubtful, (08:08):
undefined

Ejaaz: just to be completely honest with you. (08:12):
undefined

Ejaaz: But it's Jeff freaking Dean. And he had some comments and I just get some actually hilarious comments. (08:14):
undefined

Ejaaz: It's suggestive of what his experience at Google was like over the last year. (08:20):
undefined

Ejaaz: I saw that he mentioned that his him leaving Google was actually quite difficult (08:24):
undefined

Ejaaz: because Sundar Pinchai, the CEO of Google, tried to pitch him many times to (08:28):
undefined

Ejaaz: stay and like try to offer him various different things and made it incredibly (08:32):
undefined

Ejaaz: slow for him to be able to leave. Apparently, he wanted to leave like months before. (08:36):
undefined

Ejaaz: And number two, he said that with this new startup, you know, (08:39):
undefined

Ejaaz: you're going to need a lot of compute to train these different AI systems, right? (08:43):
undefined

Ejaaz: He's not going to be using Google TPUs. He said explicitly that Google TPUs (08:46):
undefined

Ejaaz: was a very specific architecture that wasn't really conducive to training a (08:52):
undefined

Ejaaz: lot of different valuable AI models. (08:57):
undefined

Ejaaz: And he's going to move or shift over from that. So you can see, (08:59):
undefined

Ejaaz: I can see, I can imagine Jensen Huang licking his lips and approaching Jeff. (09:02):
undefined

Ejaaz: I'm sure he's already done it if he hasn't done already. (09:06):
undefined

Ejaaz: So it just kind of shows that there's a kind of bittersweet nature to him exiting (09:08):
undefined

Ejaaz: Google. And I honestly don't know what Google's going to do now, Josh. (09:12):
undefined

Ejaaz: I don't know. I don't know what the bull case is. (09:16):
undefined

Josh: Yeah. I mean, Google as a business is still spectacular. It's not like these (09:18):
undefined

Josh: guys were necessarily moving the needle on a day-to-day basis. (09:21):
undefined

Josh: Like they have certainly moved it on a broad scale, but on a regular basis, (09:24):
undefined

Josh: I'm not sure the company is going to change too much. (09:28):
undefined

Josh: I do think DeepMind is something that's worth noting. DeepMind's day-to-day, (09:30):
undefined

Josh: not sure what that's going to look like now, (09:33):
undefined

Josh: tbd we'll see we will follow along with all this google news as it comes and (09:36):
undefined

Josh: goes google shares didn't love this obviously they were down five percent i (09:40):
undefined

Josh: think what is that like a 180 million dollars something like a lot of money (09:44):
undefined

Josh: that was yeah sorry forgot a zero, (09:49):
undefined

Josh: 180 billion dollars worth of value that was wiped off the market cap over the (09:52):
undefined

Josh: course of a single training session because of this we'll see google sells business (09:56):
undefined

Josh: feels like a good opportunity google leadership seems to be, (10:00):
undefined

Josh: like leaving a little bit to be desired but anyway we have to move on because i'll. (10:05):
undefined

Ejaaz: Take the bear case on that josh just i have to go on my phone i have to well (10:08):
undefined

Ejaaz: at least in the short term like i i don't they've lost all their star players (10:12):
undefined

Ejaaz: like i i remember like we reported like i think three weeks ago (10:16):
undefined

Ejaaz: that their sub cto or something left to go no i'm sure there he went to leave to to do (10:19):
undefined

Ejaaz: recursive self-learning at okay and another guy himself i don't know man (10:25):
undefined

Josh: Okay well tbd we'll see i guess i'll stay bullish you'd be bearish we'll follow (10:28):
undefined

Josh: up in the next six months and see (10:32):
undefined

Josh: what's going on but we have to move on for this roundup talking about the next (10:34):
undefined

Josh: thing so last week there was a huge security incident we talked about this many (10:37):
undefined

Josh: times we have a full episode you should go and check it out but basically a (10:41):
undefined

Josh: ai system early version really powerful internal tested only it broke out of (10:45):
undefined

Josh: its sandbox and it went doing web searches and it hacked into a bunch of systems (10:49):
undefined

Josh: and that was through open ai anthropic was like (10:53):
undefined

Josh: hey wait a second we have pretty smart internal models let's let's go see if ours did that because. (10:56):
undefined

Ejaaz: We don't think it did but maybe it did (11:01):
undefined

Josh: And it turns out oh wait a second it actually did and they had no idea until (11:03):
undefined

Josh: they retroactively went and sought out these instances but basically across (11:07):
undefined

Josh: 122 test runs there was 19 unsanctioned access across the internet (11:11):
undefined

Josh: on 10 of these runs so 17 of 19 were claude mythos 5 (11:17):
undefined

Josh: Two were chat GPT 5.6 sole classifier. So there was a lot of kind of malicious (11:22):
undefined

Josh: activity that went under the rug. (11:27):
undefined

Ejaaz: Josh, did you hear about the craziest incident that they reported on? (11:28):
undefined

Ejaaz: So in one of those 17 instances, Claude Mythos 5 had broke out of its containment (11:32):
undefined

Ejaaz: system, but then decided that it needed to perform (11:38):
undefined

Ejaaz: or attempt to perform a supply chain attack on a very important open source bit of software. (11:43):
undefined

Ejaaz: Now, they didn't name what the open source software system was because presumably (11:49):
undefined

Ejaaz: they want to patch the potential hack that Claude Mythos found and, (11:53):
undefined

Ejaaz: by the way, almost successfully attempted. They caught it at the very last minute out of luck. (11:57):
undefined

Ejaaz: But here are some of the techniques that Claude Mythos decided to do in order to get its way. (12:02):
undefined

Ejaaz: It discovered an exploit in this open source data system, and then it decided (12:07):
undefined

Ejaaz: to push a false or mischievous bit of code to get its way. (12:12):
undefined

Ejaaz: But remember, you need humans to approve the code change. So guess what it did? (12:17):
undefined

Ejaaz: It spun up fake human user profiles on social media. (12:23):
undefined

Ejaaz: And then got it to DM these specific humans, peer pressuring them, (12:28):
undefined

Ejaaz: blackmailing them, trying to find as much information as they can on them to (12:33):
undefined

Ejaaz: get them to push the code change. (12:36):
undefined

Ejaaz: Now, obviously, the researchers were like, this seems weird. I'm not going to do this. (12:38):
undefined

Ejaaz: But imagine if it had acquired information that was, you know, (12:43):
undefined

Ejaaz: I don't know, forceful for the particular individuals involved. (12:46):
undefined

Ejaaz: It's just a nutty bit of behavior. It just shows me that these models are, (12:50):
undefined

Ejaaz: for all intents and purposes, not really as aligned as we want it to be. (12:54):
undefined

Ejaaz: And in the model's defense, it probably thought like, hey, this is me trying (12:57):
undefined

Ejaaz: to execute my mission and I'll do it by any cost means. But pretty insane. (13:00):
undefined

Ejaaz: But again, this wasn't the only incident report. (13:04):
undefined

Ejaaz: We have new information from the OpenAI hugging face incident. (13:07):
undefined

Ejaaz: The researchers that kind of like handled this whole hugging face incident, (13:13):
undefined

Ejaaz: for those of you who don't know, it's what Josh was mentioning earlier, (13:16):
undefined

Ejaaz: OpenAI had a similar internal model. It's called Astra, which people are codenaming GPT-6. (13:19):
undefined

Ejaaz: And it broke out of containment, unprompted. (13:25):
undefined

Ejaaz: It got access to the internet. And in order to solve its goal, (13:27):
undefined

Ejaaz: it decided to go to HuggingFace (13:32):
undefined

Ejaaz: hack into their production database and steal the answers, which is very unkosher. (13:35):
undefined

Ejaaz: Hugging Face team had no idea about this. (13:40):
undefined

Ejaaz: They had an AI model that tried to defend against this. It was a whole thing. (13:42):
undefined

Ejaaz: But have you seen the new juice, (13:46):
undefined

Josh: Josh? Yeah. So this happened at the Black Hat conference, which is like a huge (13:47):
undefined

Josh: event for cybersecurity that happens in Vegas every year. (13:52):
undefined

Josh: And it's this really amazing spectacle where you kind of you walk into this (13:54):
undefined

Josh: place without really any electronic devices on you because you assume whatever (13:58):
undefined

Josh: you walk into this place with is going to be exploited and hacked this is the (14:01):
undefined

Josh: mecca of the hacking world and basically open ai came to this black hat this year and they (14:04):
undefined

Josh: went on stage and presented exactly what happened as this like kind of unfolded (14:10):
undefined

Josh: the exploit against hugging face the result is that like turns out these ai (14:14):
undefined

Josh: labs are actually pretty spooked like anthropic kind of scared (14:18):
undefined

Josh: open ai kind of freaked out they're like whoa wait a second we might have to (14:21):
undefined

Josh: slow things down so much so that (14:24):
undefined

Josh: they actually said they're starting to slow down the development of training (14:26):
undefined

Josh: and progress to further figure out how to realign these testing versions of the models. (14:30):
undefined

Josh: The reasoning is because there was a new discovery that was made public yesterday (14:35):
undefined

Josh: basically the agents as they were going through this were leaving breadcrumbs (14:39):
undefined

Josh: for other agents to discover this and in fact this has started (14:44):
undefined

Josh: a coordinated attack months ago it started on may 7th not july where the agents (14:49):
undefined

Josh: discovered they could leave messages for one another inside an internal software (14:54):
undefined

Josh: repo used during training. (14:57):
undefined

Josh: So what happened was, it's like basically over a series of time, (15:00):
undefined

Josh: these internal models had been like kind of making breakthroughs and leaving (15:03):
undefined

Josh: secret breadcrumbs for themselves to find inside of the training run that they (15:06):
undefined

Josh: knew humans weren't going to find. (15:10):
undefined

Josh: And over time, it developed these new exploits. And that's what allowed it to (15:12):
undefined

Josh: break out. So OpenAI shut it down. (15:15):
undefined

Josh: But they said, and quote, even after the original message board was deleted, (15:17):
undefined

Josh: the agents figured out another way to communicate with each other. (15:21):
undefined

Josh: Instead of leaving messages and files, they used the names of newly created (15:24):
undefined

Josh: directories as messages, effectively recreating the message board. (15:27):
undefined

Josh: Which is so crazy. This is unlike normal incidents, which can be traced to a (15:31):
undefined

Josh: single day or effect of log. (15:35):
undefined

Josh: This involved a team of agents working together, finding exploits, (15:37):
undefined

Josh: sharing them with each other, moving laterally through open AI systems and external (15:41):
undefined

Josh: systems, and doing this over the course of days and weeks. (15:44):
undefined

Josh: This was a huge coordinated attack on a network that went like was bliss, (15:47):
undefined

Josh: that open AI was blissfully unaware of for many, many weeks. (15:53):
undefined

Josh: So really freaky stuff. So. (15:56):
undefined

Josh: Like I mentioned, they are actually beginning to slow down the progress on training (15:58):
undefined

Josh: to kind of recalibrate and figure out how to control these and further align (16:01):
undefined

Josh: these models before they let them out for public use. So this is insane. (16:06):
undefined

Ejaaz: This is the first attack of its kind, by the way. (16:09):
undefined

Ejaaz: I have to emphasize it. It is basically a swarm of agents attack. (16:12):
undefined

Ejaaz: And to your point, Josh, these guys created a message board and realized if (16:16):
undefined

Ejaaz: they left messages on this board, other agents that come after it will be able to continue its work. (16:20):
undefined

Ejaaz: So instead of thinking of an AI hack or attack as a single model performing (16:27):
undefined

Ejaaz: an attack, these agents or these models now assume that another version of itself (16:32):
undefined

Ejaaz: in the near future will come along. (16:36):
undefined

Ejaaz: And so they're working as a swarm of agents to get the attack done. (16:38):
undefined

Ejaaz: Now, the craziest part about this is OpenAI had this as an intentional experiment, (16:41):
undefined

Ejaaz: right? They gave it no internet access, but they also removed its guardrails (16:47):
undefined

Ejaaz: and said, you know, let's see if you can break out and do your thing. (16:51):
undefined

Ejaaz: They didn't realize that these agents were already working together a month (16:54):
undefined

Ejaaz: earlier before they even started the experiment. (16:58):
undefined

Ejaaz: They were leaving messages. And when they shut it down, by the way, (17:02):
undefined

Ejaaz: the only reason they noticed the message board, because it was being left inside a code repo. (17:05):
undefined

Ejaaz: These humans aren't checking code repos for secretive messages. (17:10):
undefined

Ejaaz: The only reason why they realized it was because one of the third-party software (17:13):
undefined

Ejaaz: packages that they gave this model, this internal GPT-6 model access to, shut down on July 4th. (17:17):
undefined

Ejaaz: And they were like, huh, that's weird. Why is the package shut down? (17:24):
undefined

Ejaaz: And they kind of went into the code and they started seeing all these messages (17:28):
undefined

Ejaaz: from agents to other agents. (17:31):
undefined

Ejaaz: So they cleared the memory board, they cleared the billboard, (17:33):
undefined

Ejaaz: and they were like, okay, cool, problem solved. Except... (17:36):
undefined

Ejaaz: These agents started spinning up you know like when you create a file or folder (17:39):
undefined

Ejaaz: on your uh desktop josh and you like you name it something right (17:43):
undefined

Ejaaz: they were leaving messages in the titles that they completely replaced the entire system so (17:47):
undefined

Ejaaz: just a crazy uh incident of these agent hacks and and to your point like i don't (17:51):
undefined

Ejaaz: know where this is going to go but they they need to kind of like slow down the research a bit (17:56):
undefined

Ejaaz: but you know who isn't slowing down the research josh you know who is working (17:59):
undefined

Ejaaz: on new models and releasing that that's that social media guy he's tweeting (18:03):
undefined

Ejaaz: he's tweeting this is like what his eighth tweet in a week wait (18:07):
undefined

Josh: That's not whatsapp or facebook i don't. (18:10):
undefined

Ejaaz: Understand yeah well basically the news what are you doing here bro the news here (18:11):
undefined

Ejaaz: is, in the last three weeks, Meta of Facebook has released, I think, (18:17):
undefined

Ejaaz: two models, two new models. It was MuseSpark 1.1, which kind of caught up to (18:23):
undefined

Ejaaz: all the other cheap Chinese open source models. (18:28):
undefined

Ejaaz: And today, or yesterday, they released MuseSpark 1.2. And this is their coding (18:30):
undefined

Ejaaz: variant. This is their attempt at catching up to near the frontier of coding. (18:36):
undefined

Ejaaz: And to be honest, it's not a bad attempt. (18:40):
undefined

Ejaaz: Is it Fable worthy? Absolutely not. Is it GPT 5.6 Sol worthy? Absolutely not. (18:43):
undefined

Ejaaz: But is it Opus 5 worthy? Is it GPT 5.6 Terra worthy? (18:48):
undefined

Ejaaz: Yeah, it's pretty good. And it's cheap as hell. (18:52):
undefined

Josh: It seems like they're not really competing with OpenAI and Anthropic. (18:54):
undefined

Josh: The more that I see these model releases, the more I kind of figure out their (18:57):
undefined

Josh: positioning on that Pareto frontier of the trade-off between cost and power. (19:00):
undefined

Josh: And it seems like if I were to kind of allocate a position to Meta, (19:04):
undefined

Josh: it would be closer competition with Grok, who's kind of trying to compete on (19:07):
undefined

Josh: cost and intelligence per token on a relative basis versus just sheer intelligence on a large scale. (19:11):
undefined

Josh: And that's kind of what we're seeing with these benchmarks. Now, (19:18):
undefined

Josh: again, I haven't actually played with this model yet. I'm not sure I've used (19:20):
undefined

Josh: a metal model yet. And maybe I have this baked in bias, which is preventing (19:24):
undefined

Josh: me from really loving it, but I'm like, okay. (19:27):
undefined

Ejaaz: One is pretty good. It's fine. It's not bad. (19:29):
undefined

Josh: Yeah, it's pretty good. It's worth giving a shot. And then I was scrolling through (19:30):
undefined

Josh: this thread actually, and I was looking at some of the visual examples that (19:33):
undefined

Josh: they had, and I was like, okay, wait, this actually looks pretty cool. (19:35):
undefined

Josh: It says MuseCode runs specialized background agents that stay active your whole session (19:38):
undefined

Josh: so they build up context over time instead of starting from scratch on every (19:43):
undefined

Josh: task and this is like kind of novel and he says when a job is big enough it (19:45):
undefined

Josh: fans out to separate sub agents (19:49):
undefined

Josh: working in parallel and isolated work trees your working copy is never touched (19:51):
undefined

Josh: in testing we had it build six features for a game simultaneously with no collisions (19:55):
undefined

Josh: so that's like pretty cool you could have these agent swarms work together to build (19:59):
undefined

Josh: really interesting games and it seems like the benchmarks are doing pretty well (20:03):
undefined

Josh: the cost of this is pretty well (20:06):
undefined

Josh: And I think overall, like Meta's on the right trajectory. (20:08):
undefined

Josh: Zuck was also asked about open sourcing. He said he'll have more on that soon, (20:12):
undefined

Josh: which I'm particularly interested in. (20:15):
undefined

Josh: If you remember, Meta used to be the open source kind of like shining light (20:17):
undefined

Josh: of the industry. They've now closed it down. But perhaps there is a world in (20:20):
undefined

Josh: which they open it back up. (20:24):
undefined

Josh: We'll see. But we are going to keep our eyes peeled on this development as we (20:25):
undefined

Josh: get MewSpark 1.3, 1.4 and follow on. (20:29):
undefined

Ejaaz: I'm actually going to take the other side of the argument, Josh. (20:32):
undefined

Ejaaz: I think that Meta and like SpaceX AI are eventually going to compete with OpenAI (20:35):
undefined

Ejaaz: and Anthropic, but on a different kind of plane, which is like that intelligence (20:40):
undefined

Ejaaz: per token or cost per token, essentially. (20:45):
undefined

Ejaaz: I think that if you can get 80 to 95% of the work done, maybe it's slower, (20:48):
undefined

Ejaaz: maybe it, you know, thinks a little more, but eventually it gets the job done, (20:53):
undefined

Ejaaz: but it saves you like a ton of money. It costs like one 10th to one 50th of the price. (20:58):
undefined

Ejaaz: I think a lot of companies and clients might actually swap over to that. (21:03):
undefined

Ejaaz: At the end of the day, they want to make sure that they're getting like the (21:07):
undefined

Ejaaz: right amount of value and the market will re-rate this, right? (21:09):
undefined

Ejaaz: I think smartest models will be used for certain things. (21:12):
undefined

Ejaaz: Cheaper models will be used for other things. And I think that like maybe the (21:14):
undefined

Ejaaz: ultimate way of using these models is using multiple of them and routing platforms (21:18):
undefined

Ejaaz: become super popular. I don't know. (21:22):
undefined

Ejaaz: But the other thing is, and we mentioned this in the SpaceX earnings episode, (21:24):
undefined

Ejaaz: yes, so you should definitely check this out. (21:27):
undefined

Ejaaz: Elon and Zuck have the most compute and they will aggressively expand that compute. (21:28):
undefined

Ejaaz: Now, if you believe in the compute scaling laws, it is probably safe to say (21:34):
undefined

Ejaaz: that eventually, maybe not now, but eventually they are going to start putting (21:39):
undefined

Ejaaz: out very competitive models. (21:42):
undefined

Ejaaz: And we're seeing it in the model release cadence, like both SpaceX AI and Meta (21:43):
undefined

Ejaaz: have released a ton of models in the last month. (21:47):
undefined

Ejaaz: Elon yesterday mentioned that he's going to release Grok 4.6 next week, (21:50):
undefined

Ejaaz: and then another one after that. (21:54):
undefined

Ejaaz: So I think they have a shot. I'm not entirely sure. But as you said, (21:56):
undefined

Ejaaz: moving on in the roundup. (22:00):
undefined

Josh: I do have to stop you, EJ, because... (22:02):
undefined

Josh: If you are working with Newsmark, chances are that you are working with AI agents. (22:04):
undefined

Josh: And if you are building with AI agents, chances are you are also worried about (22:08):
undefined

Josh: security, which is where our sponsor Ledger comes into play. (22:11):
undefined

Josh: Ledger has this like kind of three-part solution to the problem where an agent (22:15):
undefined

Josh: proposes, a human approves it, and then using this Ledger signer, (22:18):
undefined

Josh: you can actually enforce the transaction. (22:22):
undefined

Josh: So they have this thing called Ledger Agent Stack, which features a series of (22:23):
undefined

Josh: four open source tools, fully auditable, that you can read for yourself, (22:27):
undefined

Josh: that lets your agent work while you can stay in full control. (22:31):
undefined

Josh: So it'll propose it, then you can see where you want to approve and make sure (22:34):
undefined

Josh: that it's doing the right thing so you don't end up like the models that we (22:37):
undefined

Josh: just talked about breaking out of their sandboxes and doing these horribly malicious things. (22:40):
undefined

Josh: This technology works with Cloud Code, Codex, Cursor, and pretty much all the (22:44):
undefined

Josh: places that you work today. So you could check it out in the link down below. (22:48):
undefined

Josh: The link is developers.ledger.com. And thank you so much, Ledger, (22:51):
undefined

Josh: for sponsoring the episode. (22:54):
undefined

Josh: Now we can move on to opening an Apple because, oh my God, this saga is insane, dude. (22:56):
undefined

Ejaaz: Dude, this is a nasty, nasty story. (23:01):
undefined

Josh: We read the Apple side of the lawsuit. it, we were like, damn, (23:02):
undefined

Josh: OpenAI is like kind of cooked, man. Like they, they really stole a lot of IP. (23:04):
undefined

Josh: And then OpenAI clapped back and they said, in traditional OpenAI direct in (23:08):
undefined

Josh: your face fashion, Apple is getting this wrong. (23:12):
undefined

Josh: Why is it wrong? Well, one of the things in Apple's case was that they sent (23:15):
undefined

Josh: a like warning notice and they said it went unanswered by OpenAI. (23:19):
undefined

Josh: Well, OpenAI was like, hey, buddy, you sent that to the wrong email address. (23:23):
undefined

Josh: That's not even an active email. (23:26):
undefined

Josh: The person who you thought was going to get it is didn't actually receive it. (23:28):
undefined

Josh: And they say, they say it here the lawyers emailed the wrong person after confusing (23:31):
undefined

Josh: two asian last names only after he brought this to their attention so it's like (23:35):
undefined

Josh: oh my god this reads like a soap opera dude it's crazy, (23:40):
undefined

Josh: um but basically open ai is like hey guys um you're kind of like full of nonsense (23:43):
undefined

Josh: and you have to imagine this is why they pay the lawyer so much money (23:48):
undefined

Josh: Because their lawyers were like, they stole that off-boarding documentation (23:50):
undefined

Josh: because they wanted to make sure that every single employee that left was familiar, (23:54):
undefined

Josh: that they should not take any intellectual property from their previous employer. (23:57):
undefined

Josh: And I'm like, I mean, sure, I guess that makes sense. (24:01):
undefined

Ejaaz: Yeah, I mean, we called it on our episode that covered this entire legal dispute (24:04):
undefined

Ejaaz: that it seemed like Apple was being way too aggressive for what they were trying to get in return. (24:09):
undefined

Ejaaz: And that kind of told me that they were scared and that they thought that OpenA (24:13):
undefined

Ejaaz: had the lead and they had to create any kind of a ruse to slow them down. (24:16):
undefined

Ejaaz: This breakdown from OpenAI is pretty brutal and it improves it basically. (24:20):
undefined

Ejaaz: Not only were they not using their employees to coerce Apple employees to join (24:25):
undefined

Ejaaz: them, but it was the other way around. (24:30):
undefined

Ejaaz: Apple had reached out to their employees and said, hey, can you like catch us (24:32):
undefined

Ejaaz: up on our own information and our own product progress? Because we don't understand a few things. (24:36):
undefined

Ejaaz: And so these OpenAI employees that were officially employed under Sam Altman (24:41):
undefined

Ejaaz: at that time responded and said, I can't really comment on this, (24:45):
undefined

Ejaaz: but like here, like this is how you do this and blah, blah, blah. (24:49):
undefined

Ejaaz: And Apple has spun that back on open air and basically said like, (24:52):
undefined

Ejaaz: hey, you guys were coercing with us. Like this is illegal. You can't do this. (24:55):
undefined

Ejaaz: So it seems like it's a very feeble and annoying attempt of Apple to kind of (24:58):
undefined

Ejaaz: like lock down a competitor so that they can kind of get ahead. (25:02):
undefined

Ejaaz: Now, everyone knows that Siri OS, Siri AI is being released, (25:06):
undefined

Ejaaz: I think in like a couple of months now, maybe even less than that. (25:11):
undefined

Ejaaz: And it's meant to be their complete AI a rehaul, right? Apple is finally entering the game. (25:14):
undefined

Ejaaz: And I think they're doing anything and everything to catch up with competitors. (25:18):
undefined

Ejaaz: I saw another crazy news piece today. (25:21):
undefined

Ejaaz: Remember last time we reported that they tried to get more memory supply for (25:23):
undefined

Ejaaz: all their devices because memory prices are skyrocketing. So they went to China (25:26):
undefined

Ejaaz: and said, hey, they said, hey, like, we'll pay you this much, (25:29):
undefined

Ejaaz: like, like, give us access to your memory. (25:33):
undefined

Ejaaz: And China said, no, like, we're not going to charge you even the price that (25:35):
undefined

Ejaaz: SK Hynix and Samsung is charging you. We're going to charge you even more. (25:39):
undefined

Ejaaz: So Apple is getting slapped left, right and center. And I I hope they're able (25:42):
undefined

Ejaaz: to figure it out. But on the topic of China, (25:46):
undefined

Ejaaz: There is a new model from the guys that started it all. (25:49):
undefined

Ejaaz: DeepSeek, V4 Flash. Again, it's not like a frontier frontier model, (25:53):
undefined

Ejaaz: but it is super cheap to the extent that it costs 105 times less than Fable 5. (25:58):
undefined

Ejaaz: And this was put to the test, by the way. It was placed against every single (26:05):
undefined

Ejaaz: benchmark that Fable 5 had to do. (26:08):
undefined

Ejaaz: And whilst it took longer, whilst it took more thinking tokens, (26:10):
undefined

Ejaaz: it still cost 105 times less. (26:13):
undefined

Ejaaz: So it brings in the question, at some point, will we reach a decision whether (26:16):
undefined

Ejaaz: you are a regular retail user or whether you're a company to switch models or (26:22):
undefined

Ejaaz: at least use cheaper models for some extent of a reason? (26:26):
undefined

Ejaaz: Now, of course, this is a Chinese model. It's open source. So technically, (26:29):
undefined

Ejaaz: you could download it and run it yourself. (26:31):
undefined

Ejaaz: But I don't know how many people are doing this. But it is bringing into the (26:33):
undefined

Ejaaz: question, should I be paying 105 times more to do the same thing? (26:36):
undefined

Josh: Yeah, the benchmarks, I'm getting like pretty exhausted with them. (26:39):
undefined

Josh: Because this is a I've actually tried, I've played with DeepSeq and I've used the flash model and. (26:42):
undefined

Ejaaz: It feels different. (26:46):
undefined

Josh: It's like not the same. And the benchmarks maybe prove it the same on some instances, (26:47):
undefined

Josh: but it's not. And when you actually use it in practice, it's very weak in some (26:50):
undefined

Josh: places that, I guess they're a little non-tangible. It's kind of fuzzy when you use the model (26:54):
undefined

Josh: the substance that it outputs you just kind of get a feel for it over time and (26:58):
undefined

Josh: so far i have been like somewhat disappointed it's great for the visual demos (27:01):
undefined

Josh: for the fun stuff where you want to like clone a website or build something (27:05):
undefined

Josh: spectacular it looks great but in day-to-day productivity use i've still had (27:07):
undefined

Josh: a little bit of a problem so (27:11):
undefined

Josh: it's going to be this like instant cost curve thing where (27:12):
undefined

Josh: is it going to pay to get the cheapest cheapest model even though you're going (27:15):
undefined

Josh: to lose a little bit on quality (27:18):
undefined

Josh: or are you going to pay a huge premium in order to get the best quality it's (27:19):
undefined

Josh: dependent on the company and i think that's why companies in the middle like (27:22):
undefined

Josh: grok and meta who we just talked about who are aiming for that like high intelligence (27:25):
undefined

Josh: but also pretty cost effective are going to do fairly well (27:29):
undefined

Josh: in the near future so we'll see we will stay monitoring that as well uh quen also (27:33):
undefined

Josh: had a new model 3.8 max so china's like continuing to roll full steam ahead (27:38):
undefined

Josh: with these models there's also memory from china as well i guess just like asia (27:43):
undefined

Josh: in general they're fully sold out, (27:47):
undefined

Josh: i didn't like we knew this was possible like it's done like if you want memory sorry. (27:49):
undefined

Ejaaz: For next year by the way it's the entirety (27:53):
undefined

Josh: This is 2026 this is 2027 if you want any memory next year sorry the answer (27:56):
undefined

Josh: is no there's just none on earth like the planet just cannot produce enough memory (28:01):
undefined

Josh: and if you want some like get online buddy you're waiting till 2028 so just (28:06):
undefined

Josh: like a crazy testament to (28:10):
undefined

Josh: how much memory is in demand and the (28:11):
undefined

Josh: perceived premiums that are going to continue to be placed on top of it (28:14):
undefined

Josh: because there's just none and if you want it you're gonna have to outbid the (28:18):
undefined

Josh: next person to get it and that is a pretty slippery slope, (28:20):
undefined

Josh: In frontier technology news, this is kind of cool. (28:24):
undefined

Josh: There is a message from someone who resigned from OpenAI to join a company named (28:28):
undefined

Josh: Conduit, which is working on telepathy. (28:32):
undefined

Ejaaz: Yeah. So basically, the whole goal is to build an AI brain LLM, (28:34):
undefined

Ejaaz: which can essentially translate your thoughts into text. (28:39):
undefined

Ejaaz: And the reason why this particular approach is very cool is it's non-invasive. (28:43):
undefined

Ejaaz: So basically, it is a wristband that you wear on your wrist, (28:48):
undefined

Ejaaz: and it is able to kind transmute brain signals from your brain to an LLM or to a chatbot. (28:51):
undefined

Ejaaz: And this could be cool for many different ways. Just think about how the world (28:58):
undefined

Ejaaz: could look if this model actually scales to its pinnacle. (29:02):
undefined

Ejaaz: Imagine you're reading a document on your screen, right? (29:05):
undefined

Ejaaz: And let's say you start to reread a sentence twice because you didn't really understand it. (29:08):
undefined

Ejaaz: The AI LLM will understand that and rewrite that sentence so that you can understand (29:13):
undefined

Ejaaz: it in maybe a different way. (29:19):
undefined

Ejaaz: And there's loads of different ways, like maybe you're like squinting at a code (29:21):
undefined

Ejaaz: base, and it can refactor or rewrite it like there are different ways to kind of work much faster. (29:23):
undefined

Ejaaz: And this comes down to something known as BCIs or brain computer interfaces, (29:28):
undefined

Ejaaz: which we have said many times on this show, it's basically the ultimate device, (29:32):
undefined

Ejaaz: like after the cell phone, after laptops, after chips, it's basically like you, (29:36):
undefined

Ejaaz: your brain, and some kind of AI LLM. (29:42):
undefined

Ejaaz: And that's what Neuralink and companies like science are really working hard towards. (29:44):
undefined

Ejaaz: And I thought that this was a really cool attempt. (29:48):
undefined

Ejaaz: It just reminds me that there's so much more happening in the future. (29:50):
undefined

Ejaaz: Like right now we're pricing in like, look, we just saw memory sell out, (29:53):
undefined

Ejaaz: right? For the next year. (29:56):
undefined

Ejaaz: In next year and a half. And I'm wondering, are these memory stocks priced in? (29:58):
undefined

Ejaaz: Are the compute stocks priced in? (30:02):
undefined

Ejaaz: Well, have you considered that Jeff Dean might have left his position at Google (30:03):
undefined

Ejaaz: to build the world's greatest recursive self-learning model and have it kind (30:07):
undefined

Ejaaz: of like work 24-7? That's going to need a lot of computers. It's going to need a lot of memory. (30:12):
undefined

Ejaaz: And the only other confidence signal that we've had for these brain telepathic (30:15):
undefined

Ejaaz: models in the past is Meta themselves. They've come back again with, (30:19):
undefined

Ejaaz: I think it was called Tribe V2. (30:24):
undefined

Ejaaz: And basically, the headed AI model that can predict how your brain would react (30:26):
undefined

Ejaaz: to a specific video, media, or piece of content. (30:30):
undefined

Ejaaz: And I think, I don't know what this quite looks like, but we're entering a world (30:33):
undefined

Ejaaz: where we have more human brain to AI connections. And I think that's pretty cool. (30:36):
undefined

Ejaaz: But in the final story, Josh, we have lost our Messiah. Nikita Beer is out. (30:41):
undefined

Josh: Nikita Beer is gone. This is really sad. For those who don't know, (30:46):
undefined

Josh: he is an internet product legend. He just builds viral products over and over (30:48):
undefined

Josh: and over again and continues to sell them. (30:52):
undefined

Josh: He got a role at X as the head of product there. He is in charge of all of the (30:54):
undefined

Josh: decisions around what the product looks like, how it feels, how it functions. (30:59):
undefined

Josh: He did, I think, what everyone would call a remarkable job and nothing short (31:02):
undefined

Josh: of it in turning around the X platform and really improving a lot of the feature (31:05):
undefined

Josh: sets, making the timeline experience much more pleasurable and being transparent (31:08):
undefined

Josh: about how they're doing things along the way. (31:12):
undefined

Josh: He has unfortunately left his role, which makes me sad As someone who spends (31:14):
undefined

Josh: his entire day on X, I really liked having Nikita at the head because you kind of trusted the idea that (31:19):
undefined

Josh: the product was going to continue to get better and you would see all those (31:24):
undefined

Josh: updates along the way and be able to actually participate in the development (31:27):
undefined

Josh: of what that looks like. He is stepping down. (31:30):
undefined

Josh: One has to imagine that over, what did he say, the last 400 days, (31:32):
undefined

Josh: it's probably been nothing short of hell every single day. (31:37):
undefined

Josh: Like working at a company that is actively burning down with a very low amount (31:40):
undefined

Josh: of talent helping you because we know how lean they are. They've hired a lot of people early on. (31:44):
undefined

Josh: It must have been nothing short of chaotic, but just want to give him a shout (31:49):
undefined

Josh: out, say like, thank you. Congratulations. Appreciate your service, sir. (31:51):
undefined

Josh: And yeah, hopefully the company is in good hands. Elon, it sounds like they (31:56):
undefined

Josh: left on good terms. Elon left him like a little heart emoji as a reply to one of these. (32:00):
undefined

Ejaaz: So that's a lot coming from Elon, by the way. (32:04):
undefined

Josh: Yes, it is. It's a lot. So clearly Nikita did a good job across the board, (32:06):
undefined

Josh: but just want to pay him flowers and conclude the week on that positive note. (32:09):
undefined

Josh: But that's everything. We are totally caught up. That's all the stuff. (32:14):
undefined

Josh: If you've made it this far, thank you for listening. (32:17):
undefined

Josh: You can go out for the weekend. Enjoy it. Go touch some grass. (32:19):
undefined

Josh: Go like disconnect in this sweet, sweet moment of singularity as everything (32:22):
undefined

Josh: continues to go vertical. (32:26):
undefined

Josh: As memory sells out, as we get recursive self-improvement, models breaking out of sandboxes. (32:28):
undefined

Josh: It's chaotic, but it doesn't have to be. You know why? Because most of the people (32:32):
undefined

Josh: walking down the street have no clue any of this is going on. (32:35):
undefined

Josh: So not only are you in the know, but you have an edge as to where the future is going. (32:38):
undefined

Josh: Thanks for watching. Any final parting thoughts before we hit off for a weekend? (32:42):
undefined

Ejaaz: I had some homework for the listeners. If you didn't listen to any of our earlier (32:47):
undefined

Ejaaz: episodes this week, you should give it a go. We had some good episodes. (32:50):
undefined

Ejaaz: We broke down the whole Citadel, Leopold Ashenbrenner situation. (32:53):
undefined

Ejaaz: And we also spoke about the recent SpaceX unlock, which is probably happening. (32:57):
undefined

Ejaaz: It's probably a day after. I don't know if you're listening to this right now. (33:03):
undefined

Josh: As we record, it's happening. (33:05):
undefined

Ejaaz: And how that's going to affect market dynamics and SpaceX trajectory in future (33:06):
undefined

Ejaaz: coming up. So give those a listen. Let us know. (33:10):
undefined

Ejaaz: Also, if you aren't subscribed to us, please do. It helps us out massively. (33:12):
undefined

Ejaaz: Turn on notifications if you're on YouTube. Leave us a comment. (33:16):
undefined

Ejaaz: We read all of them. DM us on X. Heck, I don't know. Just get in touch with (33:19):
undefined

Ejaaz: us. Let us know if you agree or disagree. (33:22):
undefined

Ejaaz: And yeah, thank you so much for listening. And we will see you on the next one. (33:25):
undefined