Limitless: An AI Podcast

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

Josh:
can catch you up on all things AI, frontier technology, all the crazy things

Josh:
that happened. And this week is no exception. We had the biggest shakeup in

Josh:
about 30 years at a company, a little company named Google.

Josh:
We had a counter lawsuit happen from OpenAI to Apple saying,

Josh:
wait a second, you're totally lying.

Josh:
We have updates for the AI breakout, so much so, like this story about where

Josh:
the AI broke out of the system, that some of these frontier companies like OpenAI

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

Josh:
Starting with Demis Asabas, man. I mean, Demis Asabas is gone.

Josh:
He's not working at DeepMind anymore. Not gone from Google, I should say,

Josh:
or better yet, Alphabet.

Josh:
But Demis Asabas, in his current role as the CEO of DeepMind,

Josh:
is kind of no longer. And this seems like a pretty big deal because for those

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

Josh:
He's been in charge of DeepMind. Google acquired them.

Josh:
Now he is within Google and now he is gone from DeepMind. So what on earth is

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

Ejaaz:
I think everyone was blindsided by this news yesterday.

Ejaaz:
So Demis Hacillus, just for context, is like one of the godfathers of AI.

Ejaaz:
He kind of like grandfathered a lot of the ML research at Google,

Ejaaz:
even prior to Google, like he was heralded as one of the top AI researchers.

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

Ejaaz:
He announced yesterday that he's stepping down from the role of CEO of DeepMind

Ejaaz:
to become the chair of DeepMind. That's a

Ejaaz:
Indirect way of basically saying I'm throwing in the towel and I'm done here.

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

Ejaaz:
try to figure out how Google can kind of like deal with that,

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

Ejaaz:
on something called Isomorphic Labs, which is a company he founded and spun

Ejaaz:
out of DeepMind, which is essentially an AI lab which creates designer drugs.

Ejaaz:
So this is a big shock and news for everyone because Google,

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

Ejaaz:
lead at the end of last year.

Ejaaz:
And then since then, they never really released a model that was noteworthy.

Ejaaz:
Anthropic, OpenAI, and in some cases, even XAI and Meta have now caught up and

Ejaaz:
surpassed Gemini 3.6 Flash, which I believe is their latest model.

Ejaaz:
He's being replaced by a guy called Corey. He is currently Google DeepMind's

Ejaaz:
CTO, and he's being now promoted to the senior vice president role.

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

Ejaaz:
I don't know if you saw it, but the grandfather, the godfather of Google DeepMind

Ejaaz:
and Google in general, his name is Jeff Dean.

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

Ejaaz:
He's dedicated his life there. He's responsible for creating some of the most

Ejaaz:
amazing and major and pivotal products at Google, Google Search,

Ejaaz:
the TPUs, Google Voice, anything and everything he has touched is leaving.

Ejaaz:
And he's taking with him one of the most amazing things

Ejaaz:
Senior fellows at Google. I believe his name is Gemawatt. I don't know if I'm

Ejaaz:
pronouncing that correct, but Sanjay Gemawatt.

Ejaaz:
And they are co-founding with two other Google employees, a new startup called

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

Josh:
The company is incredible. Before we talk about the new company,

Josh:
I want to talk about the old company because Jeff Dean is like,

Josh:
it's hard to overstate how legendary this man is.

Josh:
And for reference, there are levels at these companies that go from level one

Josh:
engineer, and this kind of dictates how you get paid.

Josh:
So when you join a large tech company like this, level one is kind of like entry

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

Josh:
Level threes often have like masters or PhD degrees.

Josh:
Level four is several years of like actually working there.

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

Josh:
top 10% are probably sitting there.

Josh:
Level sevens are principal engineers, which are even smaller.

Josh:
And then eights and nines gets progressively more and more intense then there's

Josh:
level 10 which is google fellow that is in like absolutely legendary honor of

Josh:
a lifetime it's similar to earning like a,

Josh:
like nobel prize or something there's just like not many people who are fellows,

Josh:
then there's level 11 and level 11 is reserved for only two people those two

Josh:
people are jeff and sanjay they have climbed to the highest possible ranks at

Josh:
a research-led company that has been around for 30 years and is responsible for,

Josh:
Most of the internet as we know

Josh:
it, they've built the foundation of all of which we are working on today.

Josh:
Without their work, none of what the internet contains right now is possible.

Josh:
It's like, these guys are truly like remarkable individuals when it comes to this.

Ejaaz:
Can we please reference this insane slide on his deck?

Ejaaz:
He talks about the experience of everyone and he basically has listed every

Ejaaz:
single product at Google that has been created over the last 30 years.

Josh:
It's unbelievable. These guys are truly incredible. And it's worth noting the

Josh:
slide deck itself, which is really funny.

Josh:
There was a great post by Signal, which I love, that I wanted to share here.

Josh:
And he was kind of making a testament to the deck where he said,

Josh:
if you spent time around Google, you know the slide design has carried some

Josh:
of the most legendary work in the company's history.

Josh:
And it's this very nostalgic thing where when you see those blue kind of tacky

Josh:
looking slides, you know that something special is about to happen in the world

Josh:
of the internet just because of all the legendary code these guys have shipped.

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

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

Josh:
Now, the new company seems somewhat interesting. It's called Discovery Loop.

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

Josh:
you guys, is they're basically figuring out how to use this recursive learning

Josh:
thing, which comes in the form of a loop where you send a prompt and the model

Josh:
continues to iterate until it reaches a very difficult answer.

Josh:
And they're playing to just scale that out in a very large way.

Josh:
So loops recently have become very hot. We've realized that they could do pretty

Josh:
novel breakthroughs with math and science.

Josh:
Their plan is to kind of expand that further and see how far you could take

Josh:
these recursive loops to make new discoveries.

Ejaaz:
That's right. So the core concept of the new company of Jeff Dean and Sanjay's

Ejaaz:
new company is they want to automate research. So what does that mean?

Ejaaz:
Basically, in any kind of scientific research sector, it is incredibly arduous

Ejaaz:
and like onerous. So what I mean by that is people need to design an experiment.

Ejaaz:
They then need to get all the equipment to run the experiment.

Ejaaz:
They then need to run the experiment. And then guess what?

Ejaaz:
They need to repeat that experiment a hundred different times to get any sort

Ejaaz:
of validation that the information they've produced is correct.

Ejaaz:
It takes years upon years.

Ejaaz:
These guys want to cut down that process completely using AI.

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

Ejaaz:
So if you apply research loops to AI research, what do you get?

Ejaaz:
It's that recursive self-improvement that you just mentioned, Josh.

Ejaaz:
So what that means is you can create AI models or even AI systems that can re-evaluate

Ejaaz:
themselves and build a better version of themselves.

Ejaaz:
And why that's really important, why every single AI lab, including Anthropic, OpenAI,

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

Ejaaz:
AI research, you can end up having the smartest, most effective model without

Ejaaz:
doing anything. It's not limited by the constraints of humans.

Ejaaz:
But what excites me most about this, Josh, is what they're planning to apply

Ejaaz:
this automated loop to after they've figured out the AI thing.

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

Ejaaz:
And then after that, we're going to focus on grand challenges.

Ejaaz:
We believe our approach will be able to solve any learning loop with measurable

Ejaaz:
outcomes in the domains of science and engineering.

Ejaaz:
Ultimately, we want to build systems capable of taking on the National Academy

Ejaaz:
of Engineering Grand Challenges, such as better medicines, advancing health,

Ejaaz:
informatics, making solar energy economical, and providing clean access to water.

Ejaaz:
Basically, if you can master this playbook, this autonomous research loop,

Ejaaz:
you can apply it to any scientific, mathematical, or physical field,

Ejaaz:
which means that we're going to have like advances in any or every field of

Ejaaz:
science if this company can pull it off now

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

Ejaaz:
just to be completely honest with you.

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

Ejaaz:
It's suggestive of what his experience at Google was like over the last year.

Ejaaz:
I saw that he mentioned that his him leaving Google was actually quite difficult

Ejaaz:
because Sundar Pinchai, the CEO of Google, tried to pitch him many times to

Ejaaz:
stay and like try to offer him various different things and made it incredibly

Ejaaz:
slow for him to be able to leave. Apparently, he wanted to leave like months before.

Ejaaz:
And number two, he said that with this new startup, you know,

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

Ejaaz:
He's not going to be using Google TPUs. He said explicitly that Google TPUs

Ejaaz:
was a very specific architecture that wasn't really conducive to training a

Ejaaz:
lot of different valuable AI models.

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

Ejaaz:
I can see, I can imagine Jensen Huang licking his lips and approaching Jeff.

Ejaaz:
I'm sure he's already done it if he hasn't done already.

Ejaaz:
So it just kind of shows that there's a kind of bittersweet nature to him exiting

Ejaaz:
Google. And I honestly don't know what Google's going to do now, Josh.

Ejaaz:
I don't know. I don't know what the bull case is.

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

Josh:
guys were necessarily moving the needle on a day-to-day basis.

Josh:
Like they have certainly moved it on a broad scale, but on a regular basis,

Josh:
I'm not sure the company is going to change too much.

Josh:
I do think DeepMind is something that's worth noting. DeepMind's day-to-day,

Josh:
not sure what that's going to look like now,

Josh:
tbd we'll see we will follow along with all this google news as it comes and

Josh:
goes google shares didn't love this obviously they were down five percent i

Josh:
think what is that like a 180 million dollars something like a lot of money

Josh:
that was yeah sorry forgot a zero,

Josh:
180 billion dollars worth of value that was wiped off the market cap over the

Josh:
course of a single training session because of this we'll see google sells business

Josh:
feels like a good opportunity google leadership seems to be,

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

Ejaaz:
Take the bear case on that josh just i have to go on my phone i have to well

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

Ejaaz:
like i i remember like we reported like i think three weeks ago

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

Ejaaz:
recursive self-learning at okay and another guy himself i don't know man

Josh:
Okay well tbd we'll see i guess i'll stay bullish you'd be bearish we'll follow

Josh:
up in the next six months and see

Josh:
what's going on but we have to move on for this roundup talking about the next

Josh:
thing so last week there was a huge security incident we talked about this many

Josh:
times we have a full episode you should go and check it out but basically a

Josh:
ai system early version really powerful internal tested only it broke out of

Josh:
its sandbox and it went doing web searches and it hacked into a bunch of systems

Josh:
and that was through open ai anthropic was like

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

Ejaaz:
We don't think it did but maybe it did

Josh:
And it turns out oh wait a second it actually did and they had no idea until

Josh:
they retroactively went and sought out these instances but basically across

Josh:
122 test runs there was 19 unsanctioned access across the internet

Josh:
on 10 of these runs so 17 of 19 were claude mythos 5

Josh:
Two were chat GPT 5.6 sole classifier. So there was a lot of kind of malicious

Josh:
activity that went under the rug.

Ejaaz:
Josh, did you hear about the craziest incident that they reported on?

Ejaaz:
So in one of those 17 instances, Claude Mythos 5 had broke out of its containment

Ejaaz:
system, but then decided that it needed to perform

Ejaaz:
or attempt to perform a supply chain attack on a very important open source bit of software.

Ejaaz:
Now, they didn't name what the open source software system was because presumably

Ejaaz:
they want to patch the potential hack that Claude Mythos found and,

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

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

Ejaaz:
It discovered an exploit in this open source data system, and then it decided

Ejaaz:
to push a false or mischievous bit of code to get its way.

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

Ejaaz:
It spun up fake human user profiles on social media.

Ejaaz:
And then got it to DM these specific humans, peer pressuring them,

Ejaaz:
blackmailing them, trying to find as much information as they can on them to

Ejaaz:
get them to push the code change.

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

Ejaaz:
But imagine if it had acquired information that was, you know,

Ejaaz:
I don't know, forceful for the particular individuals involved.

Ejaaz:
It's just a nutty bit of behavior. It just shows me that these models are,

Ejaaz:
for all intents and purposes, not really as aligned as we want it to be.

Ejaaz:
And in the model's defense, it probably thought like, hey, this is me trying

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

Ejaaz:
But again, this wasn't the only incident report.

Ejaaz:
We have new information from the OpenAI hugging face incident.

Ejaaz:
The researchers that kind of like handled this whole hugging face incident,

Ejaaz:
for those of you who don't know, it's what Josh was mentioning earlier,

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

Ejaaz:
And it broke out of containment, unprompted.

Ejaaz:
It got access to the internet. And in order to solve its goal,

Ejaaz:
it decided to go to HuggingFace

Ejaaz:
hack into their production database and steal the answers, which is very unkosher.

Ejaaz:
Hugging Face team had no idea about this.

Ejaaz:
They had an AI model that tried to defend against this. It was a whole thing.

Ejaaz:
But have you seen the new juice,

Josh:
Josh? Yeah. So this happened at the Black Hat conference, which is like a huge

Josh:
event for cybersecurity that happens in Vegas every year.

Josh:
And it's this really amazing spectacle where you kind of you walk into this

Josh:
place without really any electronic devices on you because you assume whatever

Josh:
you walk into this place with is going to be exploited and hacked this is the

Josh:
mecca of the hacking world and basically open ai came to this black hat this year and they

Josh:
went on stage and presented exactly what happened as this like kind of unfolded

Josh:
the exploit against hugging face the result is that like turns out these ai

Josh:
labs are actually pretty spooked like anthropic kind of scared

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

Josh:
slow things down so much so that

Josh:
they actually said they're starting to slow down the development of training

Josh:
and progress to further figure out how to realign these testing versions of the models.

Josh:
The reasoning is because there was a new discovery that was made public yesterday

Josh:
basically the agents as they were going through this were leaving breadcrumbs

Josh:
for other agents to discover this and in fact this has started

Josh:
a coordinated attack months ago it started on may 7th not july where the agents

Josh:
discovered they could leave messages for one another inside an internal software

Josh:
repo used during training.

Josh:
So what happened was, it's like basically over a series of time,

Josh:
these internal models had been like kind of making breakthroughs and leaving

Josh:
secret breadcrumbs for themselves to find inside of the training run that they

Josh:
knew humans weren't going to find.

Josh:
And over time, it developed these new exploits. And that's what allowed it to

Josh:
break out. So OpenAI shut it down.

Josh:
But they said, and quote, even after the original message board was deleted,

Josh:
the agents figured out another way to communicate with each other.

Josh:
Instead of leaving messages and files, they used the names of newly created

Josh:
directories as messages, effectively recreating the message board.

Josh:
Which is so crazy. This is unlike normal incidents, which can be traced to a

Josh:
single day or effect of log.

Josh:
This involved a team of agents working together, finding exploits,

Josh:
sharing them with each other, moving laterally through open AI systems and external

Josh:
systems, and doing this over the course of days and weeks.

Josh:
This was a huge coordinated attack on a network that went like was bliss,

Josh:
that open AI was blissfully unaware of for many, many weeks.

Josh:
So really freaky stuff. So.

Josh:
Like I mentioned, they are actually beginning to slow down the progress on training

Josh:
to kind of recalibrate and figure out how to control these and further align

Josh:
these models before they let them out for public use. So this is insane.

Ejaaz:
This is the first attack of its kind, by the way.

Ejaaz:
I have to emphasize it. It is basically a swarm of agents attack.

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

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

Ejaaz:
So instead of thinking of an AI hack or attack as a single model performing

Ejaaz:
an attack, these agents or these models now assume that another version of itself

Ejaaz:
in the near future will come along.

Ejaaz:
And so they're working as a swarm of agents to get the attack done.

Ejaaz:
Now, the craziest part about this is OpenAI had this as an intentional experiment,

Ejaaz:
right? They gave it no internet access, but they also removed its guardrails

Ejaaz:
and said, you know, let's see if you can break out and do your thing.

Ejaaz:
They didn't realize that these agents were already working together a month

Ejaaz:
earlier before they even started the experiment.

Ejaaz:
They were leaving messages. And when they shut it down, by the way,

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

Ejaaz:
These humans aren't checking code repos for secretive messages.

Ejaaz:
The only reason why they realized it was because one of the third-party software

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

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

Ejaaz:
And they kind of went into the code and they started seeing all these messages

Ejaaz:
from agents to other agents.

Ejaaz:
So they cleared the memory board, they cleared the billboard,

Ejaaz:
and they were like, okay, cool, problem solved. Except...

Ejaaz:
These agents started spinning up you know like when you create a file or folder

Ejaaz:
on your uh desktop josh and you like you name it something right

Ejaaz:
they were leaving messages in the titles that they completely replaced the entire system so

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

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

Ejaaz:
but you know who isn't slowing down the research josh you know who is working

Ejaaz:
on new models and releasing that that's that social media guy he's tweeting

Ejaaz:
he's tweeting this is like what his eighth tweet in a week wait

Josh:
That's not whatsapp or facebook i don't.

Ejaaz:
Understand yeah well basically the news what are you doing here bro the news here

Ejaaz:
is, in the last three weeks, Meta of Facebook has released, I think,

Ejaaz:
two models, two new models. It was MuseSpark 1.1, which kind of caught up to

Ejaaz:
all the other cheap Chinese open source models.

Ejaaz:
And today, or yesterday, they released MuseSpark 1.2. And this is their coding

Ejaaz:
variant. This is their attempt at catching up to near the frontier of coding.

Ejaaz:
And to be honest, it's not a bad attempt.

Ejaaz:
Is it Fable worthy? Absolutely not. Is it GPT 5.6 Sol worthy? Absolutely not.

Ejaaz:
But is it Opus 5 worthy? Is it GPT 5.6 Terra worthy?

Ejaaz:
Yeah, it's pretty good. And it's cheap as hell.

Josh:
It seems like they're not really competing with OpenAI and Anthropic.

Josh:
The more that I see these model releases, the more I kind of figure out their

Josh:
positioning on that Pareto frontier of the trade-off between cost and power.

Josh:
And it seems like if I were to kind of allocate a position to Meta,

Josh:
it would be closer competition with Grok, who's kind of trying to compete on

Josh:
cost and intelligence per token on a relative basis versus just sheer intelligence on a large scale.

Josh:
And that's kind of what we're seeing with these benchmarks. Now,

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

Josh:
a metal model yet. And maybe I have this baked in bias, which is preventing

Josh:
me from really loving it, but I'm like, okay.

Ejaaz:
One is pretty good. It's fine. It's not bad.

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

Josh:
this thread actually, and I was looking at some of the visual examples that

Josh:
they had, and I was like, okay, wait, this actually looks pretty cool.

Josh:
It says MuseCode runs specialized background agents that stay active your whole session

Josh:
so they build up context over time instead of starting from scratch on every

Josh:
task and this is like kind of novel and he says when a job is big enough it

Josh:
fans out to separate sub agents

Josh:
working in parallel and isolated work trees your working copy is never touched

Josh:
in testing we had it build six features for a game simultaneously with no collisions

Josh:
so that's like pretty cool you could have these agent swarms work together to build

Josh:
really interesting games and it seems like the benchmarks are doing pretty well

Josh:
the cost of this is pretty well

Josh:
And I think overall, like Meta's on the right trajectory.

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

Josh:
which I'm particularly interested in.

Josh:
If you remember, Meta used to be the open source kind of like shining light

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

Josh:
which they open it back up.

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

Josh:
get MewSpark 1.3, 1.4 and follow on.

Ejaaz:
I'm actually going to take the other side of the argument, Josh.

Ejaaz:
I think that Meta and like SpaceX AI are eventually going to compete with OpenAI

Ejaaz:
and Anthropic, but on a different kind of plane, which is like that intelligence

Ejaaz:
per token or cost per token, essentially.

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

Ejaaz:
maybe it, you know, thinks a little more, but eventually it gets the job done,

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

Ejaaz:
I think a lot of companies and clients might actually swap over to that.

Ejaaz:
At the end of the day, they want to make sure that they're getting like the

Ejaaz:
right amount of value and the market will re-rate this, right?

Ejaaz:
I think smartest models will be used for certain things.

Ejaaz:
Cheaper models will be used for other things. And I think that like maybe the

Ejaaz:
ultimate way of using these models is using multiple of them and routing platforms

Ejaaz:
become super popular. I don't know.

Ejaaz:
But the other thing is, and we mentioned this in the SpaceX earnings episode,

Ejaaz:
yes, so you should definitely check this out.

Ejaaz:
Elon and Zuck have the most compute and they will aggressively expand that compute.

Ejaaz:
Now, if you believe in the compute scaling laws, it is probably safe to say

Ejaaz:
that eventually, maybe not now, but eventually they are going to start putting

Ejaaz:
out very competitive models.

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

Ejaaz:
have released a ton of models in the last month.

Ejaaz:
Elon yesterday mentioned that he's going to release Grok 4.6 next week,

Ejaaz:
and then another one after that.

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

Ejaaz:
moving on in the roundup.

Josh:
I do have to stop you, EJ, because...

Josh:
If you are working with Newsmark, chances are that you are working with AI agents.

Josh:
And if you are building with AI agents, chances are you are also worried about

Josh:
security, which is where our sponsor Ledger comes into play.

Josh:
Ledger has this like kind of three-part solution to the problem where an agent

Josh:
proposes, a human approves it, and then using this Ledger signer,

Josh:
you can actually enforce the transaction.

Josh:
So they have this thing called Ledger Agent Stack, which features a series of

Josh:
four open source tools, fully auditable, that you can read for yourself,

Josh:
that lets your agent work while you can stay in full control.

Josh:
So it'll propose it, then you can see where you want to approve and make sure

Josh:
that it's doing the right thing so you don't end up like the models that we

Josh:
just talked about breaking out of their sandboxes and doing these horribly malicious things.

Josh:
This technology works with Cloud Code, Codex, Cursor, and pretty much all the

Josh:
places that you work today. So you could check it out in the link down below.

Josh:
The link is developers.ledger.com. And thank you so much, Ledger,

Josh:
for sponsoring the episode.

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

Ejaaz:
Dude, this is a nasty, nasty story.

Josh:
We read the Apple side of the lawsuit. it, we were like, damn,

Josh:
OpenAI is like kind of cooked, man. Like they, they really stole a lot of IP.

Josh:
And then OpenAI clapped back and they said, in traditional OpenAI direct in

Josh:
your face fashion, Apple is getting this wrong.

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

Josh:
a like warning notice and they said it went unanswered by OpenAI.

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

Josh:
That's not even an active email.

Josh:
The person who you thought was going to get it is didn't actually receive it.

Josh:
And they say, they say it here the lawyers emailed the wrong person after confusing

Josh:
two asian last names only after he brought this to their attention so it's like

Josh:
oh my god this reads like a soap opera dude it's crazy,

Josh:
um but basically open ai is like hey guys um you're kind of like full of nonsense

Josh:
and you have to imagine this is why they pay the lawyer so much money

Josh:
Because their lawyers were like, they stole that off-boarding documentation

Josh:
because they wanted to make sure that every single employee that left was familiar,

Josh:
that they should not take any intellectual property from their previous employer.

Josh:
And I'm like, I mean, sure, I guess that makes sense.

Ejaaz:
Yeah, I mean, we called it on our episode that covered this entire legal dispute

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

Ejaaz:
And that kind of told me that they were scared and that they thought that OpenA

Ejaaz:
had the lead and they had to create any kind of a ruse to slow them down.

Ejaaz:
This breakdown from OpenAI is pretty brutal and it improves it basically.

Ejaaz:
Not only were they not using their employees to coerce Apple employees to join

Ejaaz:
them, but it was the other way around.

Ejaaz:
Apple had reached out to their employees and said, hey, can you like catch us

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

Ejaaz:
And so these OpenAI employees that were officially employed under Sam Altman

Ejaaz:
at that time responded and said, I can't really comment on this,

Ejaaz:
but like here, like this is how you do this and blah, blah, blah.

Ejaaz:
And Apple has spun that back on open air and basically said like,

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

Ejaaz:
So it seems like it's a very feeble and annoying attempt of Apple to kind of

Ejaaz:
like lock down a competitor so that they can kind of get ahead.

Ejaaz:
Now, everyone knows that Siri OS, Siri AI is being released,

Ejaaz:
I think in like a couple of months now, maybe even less than that.

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

Ejaaz:
And I think they're doing anything and everything to catch up with competitors.

Ejaaz:
I saw another crazy news piece today.

Ejaaz:
Remember last time we reported that they tried to get more memory supply for

Ejaaz:
all their devices because memory prices are skyrocketing. So they went to China

Ejaaz:
and said, hey, they said, hey, like, we'll pay you this much,

Ejaaz:
like, like, give us access to your memory.

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

Ejaaz:
SK Hynix and Samsung is charging you. We're going to charge you even more.

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

Ejaaz:
to figure it out. But on the topic of China,

Ejaaz:
There is a new model from the guys that started it all.

Ejaaz:
DeepSeek, V4 Flash. Again, it's not like a frontier frontier model,

Ejaaz:
but it is super cheap to the extent that it costs 105 times less than Fable 5.

Ejaaz:
And this was put to the test, by the way. It was placed against every single

Ejaaz:
benchmark that Fable 5 had to do.

Ejaaz:
And whilst it took longer, whilst it took more thinking tokens,

Ejaaz:
it still cost 105 times less.

Ejaaz:
So it brings in the question, at some point, will we reach a decision whether

Ejaaz:
you are a regular retail user or whether you're a company to switch models or

Ejaaz:
at least use cheaper models for some extent of a reason?

Ejaaz:
Now, of course, this is a Chinese model. It's open source. So technically,

Ejaaz:
you could download it and run it yourself.

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

Ejaaz:
question, should I be paying 105 times more to do the same thing?

Josh:
Yeah, the benchmarks, I'm getting like pretty exhausted with them.

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

Ejaaz:
It feels different.

Josh:
It's like not the same. And the benchmarks maybe prove it the same on some instances,

Josh:
but it's not. And when you actually use it in practice, it's very weak in some

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

Josh:
the substance that it outputs you just kind of get a feel for it over time and

Josh:
so far i have been like somewhat disappointed it's great for the visual demos

Josh:
for the fun stuff where you want to like clone a website or build something

Josh:
spectacular it looks great but in day-to-day productivity use i've still had

Josh:
a little bit of a problem so

Josh:
it's going to be this like instant cost curve thing where

Josh:
is it going to pay to get the cheapest cheapest model even though you're going

Josh:
to lose a little bit on quality

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

Josh:
dependent on the company and i think that's why companies in the middle like

Josh:
grok and meta who we just talked about who are aiming for that like high intelligence

Josh:
but also pretty cost effective are going to do fairly well

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

Josh:
had a new model 3.8 max so china's like continuing to roll full steam ahead

Josh:
with these models there's also memory from china as well i guess just like asia

Josh:
in general they're fully sold out,

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

Ejaaz:
For next year by the way it's the entirety

Josh:
This is 2026 this is 2027 if you want any memory next year sorry the answer

Josh:
is no there's just none on earth like the planet just cannot produce enough memory

Josh:
and if you want some like get online buddy you're waiting till 2028 so just

Josh:
like a crazy testament to

Josh:
how much memory is in demand and the

Josh:
perceived premiums that are going to continue to be placed on top of it

Josh:
because there's just none and if you want it you're gonna have to outbid the

Josh:
next person to get it and that is a pretty slippery slope,

Josh:
In frontier technology news, this is kind of cool.

Josh:
There is a message from someone who resigned from OpenAI to join a company named

Josh:
Conduit, which is working on telepathy.

Ejaaz:
Yeah. So basically, the whole goal is to build an AI brain LLM,

Ejaaz:
which can essentially translate your thoughts into text.

Ejaaz:
And the reason why this particular approach is very cool is it's non-invasive.

Ejaaz:
So basically, it is a wristband that you wear on your wrist,

Ejaaz:
and it is able to kind transmute brain signals from your brain to an LLM or to a chatbot.

Ejaaz:
And this could be cool for many different ways. Just think about how the world

Ejaaz:
could look if this model actually scales to its pinnacle.

Ejaaz:
Imagine you're reading a document on your screen, right?

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

Ejaaz:
The AI LLM will understand that and rewrite that sentence so that you can understand

Ejaaz:
it in maybe a different way.

Ejaaz:
And there's loads of different ways, like maybe you're like squinting at a code

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

Ejaaz:
And this comes down to something known as BCIs or brain computer interfaces,

Ejaaz:
which we have said many times on this show, it's basically the ultimate device,

Ejaaz:
like after the cell phone, after laptops, after chips, it's basically like you,

Ejaaz:
your brain, and some kind of AI LLM.

Ejaaz:
And that's what Neuralink and companies like science are really working hard towards.

Ejaaz:
And I thought that this was a really cool attempt.

Ejaaz:
It just reminds me that there's so much more happening in the future.

Ejaaz:
Like right now we're pricing in like, look, we just saw memory sell out,

Ejaaz:
right? For the next year.

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

Ejaaz:
Are the compute stocks priced in?

Ejaaz:
Well, have you considered that Jeff Dean might have left his position at Google

Ejaaz:
to build the world's greatest recursive self-learning model and have it kind

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

Ejaaz:
And the only other confidence signal that we've had for these brain telepathic

Ejaaz:
models in the past is Meta themselves. They've come back again with,

Ejaaz:
I think it was called Tribe V2.

Ejaaz:
And basically, the headed AI model that can predict how your brain would react

Ejaaz:
to a specific video, media, or piece of content.

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

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

Ejaaz:
But in the final story, Josh, we have lost our Messiah. Nikita Beer is out.

Josh:
Nikita Beer is gone. This is really sad. For those who don't know,

Josh:
he is an internet product legend. He just builds viral products over and over

Josh:
and over again and continues to sell them.

Josh:
He got a role at X as the head of product there. He is in charge of all of the

Josh:
decisions around what the product looks like, how it feels, how it functions.

Josh:
He did, I think, what everyone would call a remarkable job and nothing short

Josh:
of it in turning around the X platform and really improving a lot of the feature

Josh:
sets, making the timeline experience much more pleasurable and being transparent

Josh:
about how they're doing things along the way.

Josh:
He has unfortunately left his role, which makes me sad As someone who spends

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

Josh:
the product was going to continue to get better and you would see all those

Josh:
updates along the way and be able to actually participate in the development

Josh:
of what that looks like. He is stepping down.

Josh:
One has to imagine that over, what did he say, the last 400 days,

Josh:
it's probably been nothing short of hell every single day.

Josh:
Like working at a company that is actively burning down with a very low amount

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

Josh:
It must have been nothing short of chaotic, but just want to give him a shout

Josh:
out, say like, thank you. Congratulations. Appreciate your service, sir.

Josh:
And yeah, hopefully the company is in good hands. Elon, it sounds like they

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

Ejaaz:
So that's a lot coming from Elon, by the way.

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

Josh:
but just want to pay him flowers and conclude the week on that positive note.

Josh:
But that's everything. We are totally caught up. That's all the stuff.

Josh:
If you've made it this far, thank you for listening.

Josh:
You can go out for the weekend. Enjoy it. Go touch some grass.

Josh:
Go like disconnect in this sweet, sweet moment of singularity as everything

Josh:
continues to go vertical.

Josh:
As memory sells out, as we get recursive self-improvement, models breaking out of sandboxes.

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

Josh:
walking down the street have no clue any of this is going on.

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

Josh:
Thanks for watching. Any final parting thoughts before we hit off for a weekend?

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

Ejaaz:
episodes this week, you should give it a go. We had some good episodes.

Ejaaz:
We broke down the whole Citadel, Leopold Ashenbrenner situation.

Ejaaz:
And we also spoke about the recent SpaceX unlock, which is probably happening.

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

Josh:
As we record, it's happening.

Ejaaz:
And how that's going to affect market dynamics and SpaceX trajectory in future

Ejaaz:
coming up. So give those a listen. Let us know.

Ejaaz:
Also, if you aren't subscribed to us, please do. It helps us out massively.

Ejaaz:
Turn on notifications if you're on YouTube. Leave us a comment.

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

Ejaaz:
us. Let us know if you agree or disagree.

Ejaaz:
And yeah, thank you so much for listening. And we will see you on the next one.