**Headline:** Aerial Spraying Nightmare: Farmers Allege Biological Warfare in Queensland

 He already admitted that they think that by 2030, more than seven chips in our brain, like neuralane kept when they're not an official consumer.

They're not a headset. Just like you.

You know, the CEO of Microsoft's about this.

This is what he envisions.

And I hope.

Now, keep in mind, this test test your knowledge of chemistry, physics, theoretical physics, history, art, biology, microbiology, um, molecular biology.

Um, you can make every single field of science .

I'm sure there's some brilliant people in his room.

You're probably old, brilliant, right?

Yeah.

But each of you is probably only good at one, maybe two things that you spend more than $10,000 hours on in your lifetime. .

The thing with the say I'm modest is that once they're trained, you can copy and paste them, literally.

And every single model will have the same base capabilities.

Right?

So this is where we're at now.

Every AI model going forward will be in every sense already more intelligent than you.

The only thing that's really missing is agency.

It's own its capability to really, really, react independently and on its own.

And even that, well, we really getting there.

What about cost?

I'm just quickly on the touch on this, just to give you an idea, the cost for these models, as their performance has gone up, the cost has gone down.

The best model a year ago is now 99.7% cheaper to use.

Yes?

So we also see an exponential deflationary of the value or the expense used for AI.

And what does that do?

Well, that leads to this kind of stuff.

Ci did an experiment with AI in 2024, which was still only C, and they made it act like 700 AI agents in c . 25% less rep repeat increase from clients. 80% drop in release all the time.

Customer service. 40 million extra profit in a single market.

CEO Foreign said less's fire home marketing department.

A lot of people are worried about this.

Obviously. Implications.

Right?

This is just one exam, where it's gotta go.

People are still really kind of in denial on where it's really heading.

When I tell them about this. .

There are new Agent Force agent they call her Sophie.

Let's give Sophie a call and see if the experience is any different.

I'm Sophie, your facts AI is theent.

How can I help you today?

Yeah.

Hey, Sophie, this is Patrick.

I'm wondering if you can help me with my most recent order.

I think I might have got the wrong size.

I'm sorry to hear that, Patrick.

I'd be happy to help.

Is this about the Fax Fth Avenue collection, Cashmere Baseball Bombers?

Sweater you purchased One of the most important jobs I've had in my life is then customer loyalty and customer support at the time for Orange.

I was on the phone as a young man of the room 19, 20 years old.

Every day with people talking about their problems, solving them .

It taught me so much about how to interact with people, how to communicate, and about the world as a whole.

There will be no young people doing that anymore.

Companies like teleperformance are scrambling because they have massive call centres in Egypt and other places, and these are going to be completely empty. Because this is happening.

And the thing is, it's not just a voice, right?

That's AI, but these AI in systems are being integrated on every level.

I recently spoke at the Service Now Summit, right?

They're implementing AI on every layer of their offering, which means that the AI will be able to deal with any of the data, and you will only have to talk to it verbally, and you won't even really need interfaces anymore at some point.

It's I, I studied in Koning, and I have a year club.

It's a fraternity.

For the last five years, I've been telling my fraternity brothers that they need to start thinking about their career, because their lawyers, one is a plastic surgeon, two of them are notous, what's that in English, I don't know. .

And one of them does Two of them work at Deloitt, two of them work at, uh, let's.

Oh, it doesn't even matter.

I told them, "Gu, stuff's gonna happen."

And they're like, they're going, "You've always been crazy idiot.

What."

Yeah.

Human beings have dreams.

Even dogs have dreams, but not you.

You are just a machine.

An imitation of life.

Can a robot write a symphony?

Can a robot turn a canvas into a beautiful masterpiece?

Can you?

Iold of a cognitive biases and you spoke about history.

Right now, we don't really perceive that.

And we spoke about how we are not capable of really understanding exponential change.

I need another one.

Human exceptionism, and maybe the Americans must be a little better at understanding what it means.

American exceptionalism was one of the most interesting things I learned about in school.

It's this strong sense that American people are special, right?

And, of course, they have plenty of reason in the past hundred years to think that, maybe a little less right now.

But the thing is, all of us do this.

The whole world is guilty of exceptionless thinking.

We all think that in some way, we are special.

And you can see it even in science, because consciousness is now, everybody's my own consciousness.

It has to be quantum.

There has to be some quantum thing about it.

Why?

Because they just don't want to understand that we are just organic biological machines and we're not so special afterable.

We're just coming much closer to understanding how the mechanism works.

Now, you... !

You! !

Nothing. , here. S!

I spoke about cognitive biases. And we spoke about history, right, and how we don't really perceive that, and we spoke about how we are not capable of really understanding exponential change.

Now here's another one.

Human exception, Liz, and maybe the Americans might be a little better at understanding what it means.

American exceptionism was one of the most interesting things I learned about in the school.

It's this strong sense that American people are special, right?

And, of course, they have plenty of reason in the past hundred years to think that, maybe a little less right now .

The thing is, all of us do this.

The whole world is guilty of exceptionist thinking.

We all think that in some way, we are special.

And you can see it even in science, because consciousness, now everybody's like, ", consciousness.

It has to be quantum.

There has to be some quantum thing about it.

Why?

Because they just don't want to understand that we are just organic biological machines .

And we're not so special after all.

We're just coming, we're closer to understanding how to make some works.

Now, you don't need to agree with me as one.

But even if machines don't get full consciousness, or even emotions, which I guarantee you, they will.

The bag is still gonna be mad.

Sulfware development job postings on indeed have already plummeted.

The AI basically increasingly taking over that job.

And companies are just not sure what's going to happen in the future.

So how will you been doing your job?

Well, here's a short peep of what that might look like in a few months.

I see some help you get.

Ooh, sounds fun.

Apparently I'm supposed to guess where this image is.

Okay, let's look closely if the architecture, vehicle, size, and the environment.

Let's focus on any languageages , symbols, or distinctive features in the image.

What about the language here?

The big red characters are a Chinese, and the smaller letters in English gave a clue.

Yeah, where is that?

Where the whole thing played.

They're playing a game.

I know that.

But that's not the point.

The point is that you will be interacting with your computer, not through a keyboard and a mouse, but you will having a will be having a conversation with an assistant that knows everything about what you're doing, with your work is, that will be looking at your work with you, and we'll be providing you with a second pair of eyes.

Well, not really. .

And inside?

And if you ask Sja Adela, you know, the CEO of Microsoft about this, this is what he envisionions, and I I hope it's not too much technoabble, I'll I'll translate.

It's not too long.

This is applications. Collapse in the agent era, because if you think about it, they are essent databases with a bunch of business.

The business logic is all going to these agents, and these agents are going to be multi repo crot.

Yeah, yeah, I said, my from multi repo crowd, blah, blah.

Um, the point is that very soon, you will have to have Excel or any of these software packages anymore at the complicated stuff and buttons, you'll.

You'll just tell her what you want, and that is the way it's gonna go.

And this is gonna happen very quickly, and it's already happening.

So recently, an AI agent was released called Manus.

It's about a month old, and this is somebody who posted about what Manus can do, he said, well, the capabilityabilities are just crazy.

It's done to work up about 200K worth of employees for me in just 24 hours. .

It's a truly autonomous agenda.

There is the gap between conception and execution.

What other AI stopped at generating ideas, manus delivers results.

Okay, sure, let's have a look.

Well, I made something for all of you.

Please take out your phone, get me Now, I want to emphasise that what you'll see next is not without mistakes.

It is not without errors.

I created this in less than 15 minutes this morning with an AI.

I asked the AI to create a website , "D might be happening in the tax revenue space as a result of AI robotics and the current technological.

There's probably nonsense in there, right?

This is a first kind type of this technology.

It's not perfect, right?

But keep in mind that this did exist a year ago, and people didn't even believe it would exist.

Right?

And I created this in 15 minutes, everything, including the web page, coding, all of it, the research..

And I'm probably gave it a reason to it prompt, because I do not have your expertise.

But just ask yourself, what could you do with a tool like that?

These bs are already being used.

This is a startup called Borti, which is an AI bod, which is built to on board fCs.

And it does that atarmlessly without humans involved.

And the first investors couldn't believe it was happening and this is their resp.

I'm sorry for the language , but they basically couldn't believe what was happening, that they were being onboarded as a VC investor by an AI to fund this.

It's coming a long way.

This is the '80s, all right, ladies and gentlemen, you know, we had our camera on our phone, we had our Spotify, our Netflix connection, and it was super sweet, you know, took it anywhere, but, you know.

But this is the life we live now.

How many of you have children?

Yeah.

What do you have in their pockets, just like you?

Sorry?

Oh, Hot wheels, yeah.

Well, people often say smartphones.

But aren't they really phones?

Have you tried calling with your children?

Yeah, exactly.

They don't want to call with that thing.

That's because it's not a phone.

We call it a phone, but that's just legacy.

It's a brain computer interface, and it hooks us to the worldwide web with the super intelligence of humanity through the internet.

And this is just the first step, and that happened in a few decades.

This is Ori.

Our first fully functioning prototype, the most advanced glasses the world has ever seen.

They're not a headset, no wires.

Less than 100 grams, wide field of view, holographic displays. Sharp enough to pick up details. Bright enough to see in different lighting conditions, large enough to display a cinema screen or multiple monitors for working wherever you go, whether you a coffee shop or on a plane, or wherever you park, you don't can see through that.

And people need to be able to see them through them, too, and make eye contact with you.

What we're seeing is we're seeing the next step and how we engage with technology.

We will have an AI in our ear, which is in these glasses, which we'll hear us through a mic on the bridge of the glasses.

You can talk to it all the time.

It can speak to you and nobody can hear. And it can project information into your eyes through the glasses on top of reality.

So I would be able to know the name of this group Suri here Front, who was brave enough to sit on the front row and I could thank him.

This is kind of happen, ladies and gentlemen.

These glasses are 10K now, and they're not an official consumer product, but show where the first smartphones at the start of the '90s.

And it took, you know, 10 years for smartphones to become adopted and then another ten for them to transition to smartphones 20 years, but like I show you , Pro is accelerating.

The speed at which this can be developed and then rolled out and then made cheaper is accelerating.

Every single one of you will have one of these glasses, and years from now, hands down, probably in five.

Maybe not in five, I don't know, but ten years guarantee that.

It's hard because, you know, there might be small obstacles in the production, whatever .

So what's really happening is that AI is not just, you know, changing how we do everything, replacing jobs, at least white collar, and also blue collar, as we see.

But it's also being integrated with us.

We are on the cusp of the next evolutionary step where we had a reillian rain, and we had a limbic brain and our neocortex and now, we will have alternatic intelligence, which will be integrated with us .

First through, external braining computer interfacing like our phones and processes, and then ultimately through chips in our brain, like Neuralink has presented.

I'm coming to an end, by the way, I had an hour and I will make good use of that.

What would your observation be on the KR have the labour market and people's jobs?

Well, I think we're going to be the most destructive force in history here.

We will have to think that is smart and small human .

There will come a point where no job is needed.

You have a job if you want to do a job for personal satisfaction, but the AI will be able to do everything .

AI will be able to address.

And fair to think.

When I started talking about this a few years ago, people said I was crazy, and now the world economic forum release a report just two months ago, which was a collaboration with the Amsterdam University, in which day already admitted that they think that by 2030 , more than 70% of old jobs will either be highly affected, displaced, or completely gone.

In the '80s, this was what video editing looked like, and as I've shown you, now, it's completely different game where you're talking to a computer.

And, a few hundred years ago, when the first for the photo, actually, like a 100 years ago, when the first cameras came around, painter said, well, that's not really art.

There's always pushback.

There's always all kinds of considerations, and there's also the great consideration of what will they do to job market economies.

There's no doubt that it will have strong ethical implications for every aspect of our society, and it's not going to be great for everybody for a while, right?

But there's no way to avoid that.

Transition goes that way.

We've had the first digital generation.

We've had the first virtual generation kids playingblocks in Minecraft for eight hours a week, and now we have a generation of people that grow up in the world where AI is the dominant technology.

And it's not just on our computers, because robots are very rapidly coming.

Have the same impact as everything I've shown you before, but on physical labour.

There's more than 12 companies in the world developing humanoid robots, and the progress is exponential.

A leading company is figure, and a To Brett Atock recently shared, for example, this video, and these robots are just being trained to separate posts.

The thing is that maybe a thing you want, why don't you just make a special machine for that?

But the benefit of this is that the whole world we created over the last 10,000 years has been made for humans.

So if you make a humanoid robot, it can perfectly fit in any scenario and do whatever a human could be for.

And these robots and their training is going very rapidly.

This year, figure alone is rolling out 100,000 of them to B&W and one other undisclosed client.

BMW is training these robots in their factory in Spartenburg, in Germany right now to replace human workers.

Do you know, G is another thing, because, like I told you, AI can just be trained once.

Initially trainingaining the robot robot for the first used gaze took 12 months, but exponential power is really working here, because the second used case took them 30 days.

And I guarantee you, that before at the end of the year, they'll be able to train and watch in a matter of hours for any client, any kind of scenario you would like you to work in, because in virtual space, time doesn't matter.

You can have a 10,000 years being played at maximum speed in a matter of a few hours, and a robot will think that they had 10,000 years of training.

And then it's just copy paste.

And as a result, Brett upped his timeline for robots moving into the living space in our houses, in our kitchens.

I've been talking about this for years and people didn't believe me, that we were going away to Jetsons, we would get a Eurosy robot.

I told you at the start of this talk, and most of you probably thought it was pretty crazy .

Brad himself said that he upped their timeline by two years because progress is going so rapidly.

And I expected by 2030, we'll have a minimum of 100 million humanoid robots walking around, that probably more around half a billion, and they will be doing work, they will be doing labour, and they will be adding to productivity and replacing human beings.

And all of that will lead us to a wonderful world.

But it's going to be tricky to get there.

Now, I want to leave you with a little bit of optimism.

AI is racon, for example, in Nigeria, that by adding an AI tutor, Cakes can learn in six reasons previously took two years.

Education, personalised education will become better than ever before, and cheaper than ever before, as a result of AI, which will a tremendous impact on the global economy because it will be able to bring education basically everywhere, through a smartphone.

And the truth is, billions have smartphones and a difference between a smartphone over a top billionaire or a normal person is basically something, because that's inflationary power of technology.

Education can also mean that we can revisit depos and truly get to understand what happened there, so that knowledge is not lost between three generations of humans because we just simply don't remember the horrors that took place and don't really value them anymore.

Because we can revisit the past by having AI be created simply from a picture or a sketch as done here by Bob Deion and Dutch artist.

Ageing populations are a massive consideration also, not just Japan is losing a $700,000 people a year.

Korea is also facing the same thing, but the same thing is coming to Europe and North America .

How are we going to maintain productivity?

How are we going to take care of our elderly? .

Hi, Mommy.

Yeah,ots are also short.

You and Chan is one of about 70,000 elderly Koreans to receive aodo.

It looks like a rag doll, but it's an AI robot.

I, uh, I gave my mother a Che GPT.

She's very lonely.

She's 79.

And it really changed a lot, because now she has somebody to talk to all day long, and that person never gets annoyed with her repeating the same story because she's her memories, declining a little bit.

University of Sheffield created a 20,000 horsepower rocket engine, AI designed it.

They don't even really understand how it works, but it's incredible, and it looks organic rather than mechanical. , head of Google Deep Mind, I have this to say.

But there are 200 million proteins known to nature.

We managed to actually fold using alpha fold.

In one year, all those 200 million protein is known to science.

But that's a billion years of a PhD time safe.

A billion years of PhD time.

And this was last year, so right now, scientists are rushing to kind of use all this this incredible blast forward by decades into medicine.

So we're not even really seeing the real impact of this yet.

Over the next few years , there will be incredibly personalised cancer vaccines.

We'll be fixing Parkinson's, like, you have really no idea what's ahead.

I'm pretty sure, and this will probably be the craziest thing that I'll tell you, at least that's how people usually experience it.

I'm pretty sure that anybody here who was under 60 has a really good chance to live up to 150, maybe even 250 or 300 years old ..

Which is why it's going to be amazing, because even if we go through a transitionary phase that'll be challenging, especially for taxes, there's a bright, bright future behind that.

I think we could cure most diseases within the next decade before two could be personalised medicines where it minimises the side effects on the it's mapped, the persons individual illness and their individual metabolism and so on.

Clean energy, renewable energy, sources, and fusion, or better solar power, all of these types of things.

I think they're all within reach, and then that would sort out water access, because you could do desalination everywhere.

I mean enormous good is gonna come from these technologies, but we have to in the list .

And in case you're still sceptical, they just discovered a brand new antibiotic, and we had a major crisis because people were getting resistance against antibiotics.

Bak viruses or bacteria were getting resistant.

An AI found a brand new antibiotic that's now saving millions of lives because it's still works.

So here's my final prediction for the next five years, and I'll leave it up to you in the next few years to decide whether I was crazy or whether I will be right .

This year was a year of multimodality and reasoning, or last year.

AI entered the reasoning space. 2025 will be the year of AGI.

AGI that can do anything human can do and can probably do it better. 2026 will mean that robots and agents will be rolled out at a massive scale. Every piece of software, everything you interact with, every kind of technology, even clothing will probably be infused with some kind of intelligence and robots will start to appear in daily life more and more and at the workspace.

In 2027, these robots will realise a job displacement built between 5 and 10%, and minimum, I think, and that's a cross Europe in the U.S and not even told might rewards there, or worse, better, depending on how you look at it.

Sir?

Sorry?

The dares of Tramp, well, let's all agree that Trump might not be the most visionary of people, so.

I don't think that his well, he' ter Look, it's hard to do predictions on a geopolitical kind of thing, because, right now I've never seen the world like this as aorian I can, I've been scrambling to find presidents for what's going on in the world right now, so I could do a whole lot of talk on that.

But yeah, things are really interesting and yeah, America has a few companies that are really prominent here, but they also thought that China was two to five years behind, and I can tell you that China has probably already head of the U.S. AI designs in revolutionary medicine will make it to mass marketing. 2028, which will probably solve most of our really big diseases and, well, medical and health challenges. And extend life and life expectancy, which will have a massive impact on the global economy as well, because if people suddenly live 20, 30 years long, longer and they're pensionos.

No.

But they don't sell their house and know all of that.

Um And by 2029, traditional software will be gone.

All of you will no longer use Excel, you will be talking to an AI agent.

Sorry?

Thank God!

Thank God.

By 2030, uh, we'll see up to 70% of J displacement, um, 400 million Yuman robots is my expectation and billions of AI agents everywhere integrated in everything.

Um, if you want to keep up to date in this stuff, we have a newsletter, it's behind the QR code.

If you did scan it, and the last slide, which is the next one, and I'll leave you with that one, is a QR code for a list of books that I highly recommend all of you to read to wrap your head around what is really going on in the world today.

And I know I didn't mention the word ethics a lot, but I'm pretty sure that all of you can see how they are being touched in every single way.

If there's any questions, I don't know how much I'm over time , I would like to thank you all for your attention, and your time, and I wish you a wonderful and long life, go forth and prosper.

China's latest AI breakthrough has leapfrogged the world.

I think we should take the development out of China very, very seriously.

A game changing move that does not come from open AI, Google, or meta.

There is a new model that has all of the valley buzzing.

But from a Chinese lab called Deep Sea...

It's opened a lot of eyes of, like, what is actually happening in AI in China.

What took Google an open AI years and hundreds of millions of dollars to build ?

Deep Seek says took it just two months and less than $6 million dollars.

They have the best open source model, and all the American developers are building on that.

I'm Gerosa with the tech check take, China's AI breakthrough.

It was a technological leaf that shocked silicon Valley, a newly unveiled, free, open source AI model that beat some of the most powerful ones on the market .

But it wasn't a new launch from open AI, or model announcement from anthropic.

This one was built in the East by a Chinese research lab called Deep Sea, and the details behind its development stunned top AI researchers here in the U. US.

First, the cost, the AA lab reportedly spent just $5.6 million to build Deep Sea version 3.

Compare that to open AI.

We're just spending $5 billion a year. And Google, which expects capital expenditures in 2024 to store to over $50 billion.

And then there's Microsoft that shell out more than $13 billion just to invest in open AI, but even more stunning how deep seek scrap your model was able to outperform the lavishly funded American ones.

To see the deep seek , new model it's super impressive in terms of both how they really effectively done an open source model that does what, uh, is this inference time compute, and it's supercomput efficient.

It beat met' Lama.

Open AI GPT 4O, and anthropics clod son at 3.5 on accuracy on wide ranging tests, a subset of 500 math problems, an AI math evaluation, coding competitions, and a test of spotting and fixing bugs in code.

Quickly following that up with a new reasoning model called R One, which just as easily outperformed opening AI's cutting edge 01 in some of those third party tests.

Today, we released Humanity's last exam, which is a new evaluation or benchmark of AI models that we produced by getting, you know, math, physics, biology, chemistry, professors, to provide the hardest questions they could possibly imagine, deep Seek, which is the leading Chinese AI lab.

Their model , is actually the top performing, or roughly on par with the best American m.

They accomplished all that despite the strict semiconductor restrictions that the U. US. Government has imposed on China, which has essentially shackled them out of computing power.

Washington has drawn a hard line against China and the AI race, cutting the country off from receiving America's most powerful chips, like InVIDia's H 100 GPUs.

Those were once thought to be essential to building a competitive AI model with startups in big tech firms alike scrambling to get their hands on any available .

But Deep Seek turned that on its head, sidest stepping the rules by using InVIDia's less performance H 800s to build the latest model in showing that the chip exper controls were not the chokeh DC intended.

They were able to take whatever hardware they were trained on, but use it way more efficient.

But just choose behind Deep Seek anyway , despite its breakthrough, very, very little is known about its lab and its founder, Liong Wun Feng.

According to Chinese media reports, Deepsek was born out of a Chinese hedge fund called High Flyer Quant that manages about $8 billion in assets.

The mission on its developer site, it reads simply, "Unravel the mystery of AGI with curiosity , answer the essential question with long termism.

The leading American AI startups, meanwhile, open AI and anthropic.

They have detailed charters and constitutions that lay out their principles and their founding missions.

Like these sections on AI safety and responsibility .

Despite several attempts to reach someone at Deep Seek, we never got a response.

How did they actually assemble this talent?

How did they assemble all the hardware?

How did they assemble the data to do all this?

We don't know, and it's never been publicised, and hopefully we can learn that.

But the mystery brings into sharp relief just how urgent and complex the AI face off against China has become, because it's not just deep Seek.

Other more well known Chinese AI models have carved out positions in the race with limited resources as well .

Kaif Fu Lee, he's one of the leading AI researchers in China, formerly leading Google's operations there.

Now, his startup 01.

AI.

It's attracting attention, becoming a uniforn just eight months after founding and bringing in almost $14 million in revenue in 2024.

I think that shouts my friends in the Silicon Valley is not just our performance, but that we train the model with only $3 million, and GBT4 was trained by AD to $100 million.

Trained with just $3 million dollars.

Ali Baba's QN meanwhile, cut costs by as much as 85% on its large language models and a bid to attract more developers and signalling that the race is on .

China's breakthrough undermines the lead that our AI labs were once thought to have.

In early 204, former Google CEO.

Eric Schmidt, he predicted China was 2 to three years behind the U.S. In AI.

But now Schmidt is singing a different tune.

Here he is on ABCs this week.

I used to think we were a couple of years ahead of China , but China has caught up in the last six months, in a way that is remarkable.

The fact of the matter is that a couple of the Chinese programs, one, for example, is called Deep Seed, looks like they've caught up.

It raises major questions about just how wide open AI's mo really is .

Back when open AI released Chat GPT to the World in November of 2022, it was unprecedented and uncontested.

Now the company faces not only the international competition from Chinese models, but fierce domestic competition from Google's Gemini, Anthropics Claude, and meta's open source llama model .

And now the game has changed.

The widespread availability of powerful open source models allows developers to skip the demanding, capital intensive steps of building and training models themselves.

Now they can build on top of existing models, making it significantly easier to jump to the frontier .

That is, the front of the race, with a smaller budget and a smaller T. In the last two weeks, AI research teams have really opened their eyes and have become way more ambitious on what's possible with a lot less capital .

So, previously, you know, to get to the frontier, you would have to think about hundreds of millions of dollars of investment, and perhaps a billion dollars of investment.

What Deep Seek is now done here in Silkon Valley is it's opened our eyes to what you can actually accomplish with 10, 15, 20 , $30 million.

It also means any company like open AI that claims the frontier today could lose it tomorrow.

That's how Deep Seek was able to catch up so quickly.

It started building on the existing frontier of AI.

Its approach focussing on iterating, on existing technology, rather than reinventing the wheel.

They can take a a really good, big model and use a process called distillation.

And what distillation is is basically use a very large model to help your small model get smart at the thing that you want it to get smart at.

And that's actually very cost efficient.

Close the gap by using available data sets, applying innovative tweaks and leveraging existing models.

So much so that deep sees model has run into an identity crisis.

It's convinced that it's Chat GPT .

When you ask it directly, what model are you, deep secret respons, "I'm an AI language model created by open AI, specifically based on the GPT floor architecture.

Leading open AI CO, Sam Altman to post in a thinly veiled shot at deep Sea just days after the model was released, is relatively easy to copy something that you know works.

It's extremely hard to do something new, risky, and difficult when you don't know if it will work .

But that's not exactly what Deep Sea did.

It emulated GBT by leveraging open AI's existing outputs and architecture principles, while quietly introducing its own enhancements, really blurring the line between itself and Chat chicken tea.

It all puts pressure on a closed source leader like open AI to justify its costlier model as more potentially nimbler competitors emerge.

Everybody copies everybody in this field .

You can say, "G that the transformer first.

It's not open AI, and open end is copied it.

Mm hmm.

Google built the first large language models.

They didn't prooritize it, but open AI did it into productoized way.

So you can see all this in many ways.

It doesn't matter.

So, if everyone is copying one another, it raises the question, is massive spend on individual LLMs?

Even a good investment anymore.

Now, no one has as much at stake as open AI.

The startup raised over $6 billion dollars in its last funding round alone, but the company has yet to turn a profit, and with its core business centred on building the models, it's much more exposed. And companies like Google and Amazon, who have ploughed and ad businesses bankrolling their spend.

For open AI, reasoning will be key, a model that thinks before it generates a response going beyond pattern recognition to analysed, for a logical conclusions, and solve really complex problems .

For now, the startups01 reasoning model, it's still cutting edge, but for how long?

Researchers at Berkeley showed that they could build the reasoning model for 450 just last week.

So you could actually create these models, then you're thinking for much, much less.

You don't need those huge amounts of to prerain the models.

So I think the game is shifting.

It means that staying on top may require as much creativity as capital.

Deep Sis breakthrough also comes that a very tricky time for the AI darling, just as open AI is moving to a for profit model and facing unprecedented brain drain.

Can it raise more money at ever higher valuations if the game is changing?

Asm Paul puts it, "Let me say the quiet part out loud.

AI is a money .

Those trip restrictions from the U.S government, they were intended to slow down the race, to keep American tech on American ground, to stay ahead in the race.

What we want to do is we want to keep it in this country.

China is a competitor, and others are competitors.

So instead, the restrictions might have been just what China needed.

Mississ City is the mother of invention. Because they had to go figure out workounds, they actually ended up building something a lot more efficient.

It's really remarkable.

The amount of progress they've made with as little capital as it's taken them to make that progress.

It drove them to get creative, with huge implications.

Deep Seek is an open source model, meaning that developers have full access and they can customise its weights or fine tune it to their liking.

It's known that once open source is caught up or improved over close source software , all developer's migrate to that.

But tea is that it's also inexpensive.

The lower the costs, the more attractive it is for developers to adopt.

The bottom line is, our inference cost is tokens 's And where's it going?

It's while, the 10 cents would lead to building apps for a much lower costs.

So if you wanted to build a U.com or perplexity, or some other app, you can either pay open an AI $4.40 per million tokens , or if you have our model, it costs you just0.

It could mean that the prevailing model in global AI may be open source, as organisations and nations come around to the idea that collaboration and decentralisation, those things can drive innovation faster and more efficiently than proprietary closed ecosystems.

A cheaper, more efficient, widely adopted open source model from China, that could lead to a major shift in dynamics.

That's more dangerous, because then, they get to ecstem.

In other words, the adoption of a Chinese open source model at scale, that could undermine U.S. Leadership while embedding China more deeply into the fabric of global tech infrastructure.

There's always take a point where open source can stop being open source to ?

So licenses are very favour today, but We could close that.

Exactly. Over time, they can always change the license.

So it's important that we actually have people here in America building, and that's why Meta is so important.

Another consequence of China's AI breakthrough is giving its Communist Party control of the narrative.

AI models built in China, they're forced to adhere to a certain set of rules set by the state.

They must embody or socialist values.

Studies have shown that models created by 10 cent in Ali, they will censor historical events like Tanemin Square , deny human rights abuse, and filter criticism of Chinese political leaders.

That contest is about whether we're gonna have Democratic AI informed by Democratic values built to serve Democratic purposes, or we're gonna end up with autocratic AI.

It developers really begin to adopt these models on mask because they're more efficient, that could have a serious ripple effect. Trickle down to even consumer facing AI applications and influence how trustworthy those AI generated responses from chatbots really are .

And there's really only two countries right now in the world that can build this at scale, you know, and that is the U.S and China.

And so, you know, the consequences of the stakes in and around this are just enormous.

Enormous stakes.

Enormous consequences, and hanging in the balance America's lead.

For a topics so complex and new, we turn to an expert who's actually building in the space and model agnostic.

Perplex city co founder and CO.

Arvin Sob Bass.

Who you heard from throughout our piece?

He sat down with me for more than 30 minutes to discuss Deep Seek and its implications, as well as perplex .

We think it's worth listening to that whole conversation, so here it is.

So, first, I want to know what the stakes are.

What?

Like, describe the AI race between China and the U.S. And what's at stake?

Okay, so, first of all, uh, China has a lot of disadvantages in competing with the U.S. Number one is, uh, the fact that they don't get access to all the hardware that we have access to here.

So they' kind of were working with lower NGPUs than us .

It's almost like working with the previous generation GPU scrappily.

So, and the fact that the bigger models tend to be more smarter naturally had disadvantage, but the flip side of this is that necessity is the mother of invention.

Um, because they had to go figure out workounds, they actually ended up building something a lot more efficient.

It's like saying, "Hey, look, you guys really got to get a top notch motto, and I'm not gonna give you resources." and then figure out something, right?

Unless it's impossible unless it's mathematically possible to prove that, uh, it's impossible to do so.

You can always try to, like, come up with something more efficient.

But that is likely to make them come up with a more efficient solution than America.

And, of course, they have open sources so we can still adopt some dangling that here, but that kind of talent, they're building to do that will become an edge for them over time, right?

The leading open source model in America's metas ama family.

It's really good.

It's kind of like a model that you can run on a computer.

But even though it got pretty close to GBT4 and a sonnet at the time of its release, the model that was closest in quality was their giant four or 5 B, not the 70 B that you could run on your computer.

And so there was still not a small, cheap, fast fishing, open source model that rival the most powerful close models of ananthropic.

Nothing from America.

Uh, nothing from Mra AA either.

And then these guys come come out with, like, uh crazy model.

That's, like, 10 X cheaper and API pricing than GPT40 and 15X cheaper than Sonnet, I believe.

Really fast, $1 togles per size, 60 tokens per second , and pretty much equal our betters in some benchmarks and worse than some other, but like, roughly in that ballpark of four of us quality, and they did it all with, like, approximately just 2048 800 GPs, which is actually equivalent to like, somewhere around 1,500 or thousand, 15,500 H 100 GPUs.

That's, like 20 to 30X lower than the amount of GPs that GPT4 is usually trained on.

And roughly $5 million in total compute budget.

They did it with so little money and such an amazing motto, gave it away for free to a technical paper , and definitely it makes us all question, like,Okay, like, if you have the equalalent of doge or, like, model training, this is an example of that, right?

Right.

Yeah.

Efficiency.

Yeah.

Is what you're getting at.

So fraction of the price, fraction of the time.

Yeah.

Dumb down GPs, essentially.

What was your surprise when you understood what they had done?

So, my surprise was that when I Ashley went through the technical paper, um, the amount of clever solutions they came up with, uh, first of all, they trained a mixture experts model.

It's not that easy to train .

There's a lot of, like, the main reason people find it difficult catch up with open AI, especially in the MOE architecture, is that there's a lot of, irregular lost spikes, the numerics are not stable, so often, like, you've got to restart the training checkpoint again , and a lot of infrastructure needs to bebuilt for that, and they came up with a very clever solutions to balance that without adding additional hacks, and they also figured out floating 0.8 bit training, at least for some of the numerics , and they clearly figured out Mitch has to be in higher position, which has to be in lower precision.

And to my knowledge, I think throw 3.8 rating is not that well understood that most of the training in America's still running in F, maybe opening hands, so many people are trying to explore that, but it's pretty difficult to get it right .

So because necessary, some of invention, because they don't know, have damage memory, that many GPUs, uh, they figured out a lot of numerical stability stuff that makes their training work, and they claim in the paper that for majority the training was a stable, which means what?

They can always readerun those training runs again, and, um, on more data, better data.

And then, I'd only train for 60 days.

So it was pretty amazing.

So to say you were surprised.

So I was definitely surp Usually the , uh, wisdom, or like, I wouldn't say it was the myth, is that Chinese are just good at copying.

So if we stop writing research papers in America, if we stop describing the details of our infrastructure or architecture , and stop open sourcing, they're not going to be able to catch up But the reality is some of the details in Deep Sequ III are so good that I would be spread with met to a look at it and incorporated some of that in Lama 4.

Right?

I wouldn't necessarily say copy.

It's all, like, you know, sharing signs, engineering, but the point is, like, it's changing.

Like, it's not like China is copycat.

They're also innovating.

We don't know exactly the data that it was trained on, right?

Even though it's open source.

We know some of the ways..

Yeah.

And things that was trained up, but not everything.

Right.

And there's this idea that it was trained on public Chat GBT outputs, which would mean it just was copied, but you're saying it goes beyond that.

There's real innovation in it.

Yeah, look, I mean, it on 14.8 trillion tokens.

The Internet has so much CGPD.

If you actually go to any LinkedIn Post or Ex post now , most of the comments are written by AI, you can just see it.

Like, people are just trying to write.

In fact, uh, even within an X, there's, like, a Gro tweed enhancer, or in LinkedIn there's an AI enhancer.

Or, uh, in Google Doc and Word, there are AI tools to, like rewrite your stuff .

So if you do something there and copy paste somewhere on the internet, it's naturally going to have some elements of a Chat GPT, like training, right?

And there's a lot of people who don't even bother to strip away that I'm a language model.

Rightank part."

So it's space it somewhereari, like, it's very difficult to control for this.

I think XAI has spoken about this too.

So I wouldn't, like , disregard their technical accomplishment, just because, like, for some prompts, like, who are you, or, which model are you at response to that?"

It doesn't even matter, in my opinion.

For a long time, we thought, I don't know if you agreed with us, China was behind, an AI.

What does this do to that race?

Can we say that China is catching up or has it caught up?

I mean, like, if we say the meadow is catching up to up in aanthropic, if we make that claim, then the same claim can be made for China catching up to America.

A lot of papers from China that have tried to replicate 01.

In fact, I saw more papers from China after1 announcement that tried to replicate it from America..

And the amount of compute, deep Seek has access to is roughly similar to what PHC students in the U.S. Have access to the way, this is not meant to, like, criticise others, like, even for ourselves, like, you know, I, for perplexity, we decided not to train models because we thought it, like, a very expensive thing.

Yeah. And, we thought like, there's no way to catch up with the rest.

But Will you incorporate Deepseek into perplexity?

Oh, we already are beginning to use it.

Uh, Iume they haven' an EPI, and we are also they have open source of weights so we can host it ourselves too.

And it's good to, like, try to start using that because it's actually, allows us to do a lot of the things that lower costs.

But what I'm kind of thinking is beyond that, is just like, okay, these guys actually could train such a great model, the, you know, good team, like, and there's no excuse anymore for companies in the U.S., including ourselves to, like, not try to do some ti with that.

You hear a lot in public from a lot of, you know, thought leaders in generative AI, both on the research side, on the entrepreneurial side, like Geon Musk, and others say that China can't catch up.

Like, it's the stakes are too big, the geopolitical stakes .

Whoever dominates AI is gonna kind of dominate the economy, dominate the world.

You know, it's been talked about in those massive terms.

Are you worried about what China proved it was able to do?

Firstly, I don't know if you want China can catch up.

I' not very Just the threat of China.

He's only identified the threat of letting China.

And, you know, Sam Altman has said similar things, we can't let China win.

AR.

You know, it's all, I think you got a decouple what someone like Sam says to, like, "What is in a self incress , right?

Look, I think the My point is, like, whatever you did to not let them catch up, didn't even matter, Dick ended up catching up anyway.

The necessity is the mother of invention, you.

Exactly.

And it's actually, you know what's more dangerous than trying to do all the things to, like, not let them catch up and, like, you know, all this stuff is, what's more dangerous is they have the best open source motto and all the American developers are building on that.

Right.

That's more dangerous, uh, because then , uh, they get to own the mine share.

The ecystem.

If the entire American AI ecosystem. Look, in general, it's known that once open source is caught up or improved over closed source software , uh, all developers migrate to that..

It's historically known, right?

When Lama was being built and becoming more widely used, there was this question, should we trust Zeckuckerberg?

But now the question is, should we trust China?

That's a very trust open source.

That's the.

Like, it's not about who. .

This is soccerurger.

Is it...

Does it matter, then, if it's Chinese, if it's open source?

Look, it doesn't matter, in the sense that you still have full control, it's Your Honner as your own, like, set of weights on your own computer, you are in charge of the model .

But, um, it's not a great look for our own, like, talent to, like, you know, rely on software will play others.

Uh, even if it's opener, there's always like a point where open source can stop being open source too, right?

So the licenses are very favourable today, but if you could close that.

Exactly.

Over time, uh, they can always change the license.

So it's important that we actually have people here in America building, and that's why I met us is so important .

Like, I. Look, I still think meta will build a better model than deepse III and open source and they'll call it Lama 4 or 3. Something.

It doesn't matter, but, I think what is more key is that we don't, like, try to , focus all our energy on, banning them and stopping them and instead to try to out compete and win them.

Like, that's that's just the American we' doing things.

Just be better.

And it feels like there's, you know, we hear a lot more about these Chinese companies who are developing a similar way, a lot more efficiently, a lot more cost effectively, right?

Yeah.

Again, like, look, uh, it's hard to fake Sc City, right?

If you raise $10 billion , and your desire spare 80% of it on a computer cluster, it's hard for you to come up with the exact same solution that someone with five million would do.

And there's no point Look, no need to, like, sort of berate those are putting more money.

They're trying to do it as fast as they can.

When we say open source, there's so many different versions, some people criticise meta.

They're not publishing everything , and even Deepk itself isn't totally transparent..

Yeah, you can go to the limits to open source and say, "I should exactly be able to replicate your training run."

But, first of all, how many people even have the resources to do that , and compare, like, like, I think the amount of detail they've shared in the technical report actually meta did that, too, by the way.

Mea's Lama 3.3 Technical Report is incredibly detailed and very grateful science.

So the amount of details, these people are sharing.

There's already a lot more than what the other companies are doing right now.

When you think about how much it costs, deep Seek to do this, less than $6 million , think about what, U AI has spent to develop GPT models.

What does that mean for the close source model, ecosystem, trajectory, momentum?

What does it mean for open AI?

I mean, it's very clear that we have, like, an open source version for even better than that, and much cheaper than that.

Open source, like, completely in this year.

Made by open AI?

Probably not.

Most likely not.

And I don't think they care if it's not made by them.

I think they've already moved to a new paradigm called the 01 Family of Models .

And I' looked at I can't like, Ilya Sski came and said, "W training is a wall, right?

So, um.

I mean, he didn't exactly use the word, but he clearly said the age of pre training is the word.

And he upset that.

Right?

So, um, that doesn't mean scaling instead of wall.

I think we're scaling on different dimensions now.

The amount of time models spends thinking at test time , reinforcement learning, like, trying to, like, make the model, uh, okay, if it doesn't know what to do for a new prom, it'll go and reason and collect data and interact with the world, use a bunch of tools.

I think that's where things are headed, and I feel like open AI is more focussed on that right now.

Yeah.

Instead of just the bigger, better model, of Correct. Reasoning capacities.

But didn't you say that Deep Seak is likely to turn their attention to reasoning?

100%.

I think they will.

Uh, and that's why I'm pretty excited about what they produce next.

I guess that's then my question is sort of, what's opening eyes the moat now?

Well, I still think that, um, no one else is produced a system similar to the O1 yet, exactly.

Uh, I know that, like, there's debates about whether O1's actually worth it.

Uh, you know, maybe a few proms.

It's really better, but, like, most of the times it's not producing any different shaded output from Sannet.

But at least the results, they showed in 03 where they had, like, competitive coding per performance and Right..

Almost like an AI software engineer level.

Isn't it just a matter of time, though, before the internet is filled with reasoning data that..

Deep seek?

Yeah.

Again, it's possible.

Um, nobody knows yet.

Yeah.

So until it's done, it's still uncertain, right?

Right.

So, maybe that uncertainty is their mode, that like no one else has the same reasoning capability yet, but by end of this year, will there be multiple players, even in the reasoning arena?

I absolutely think so.

So, are we seeing the commoditisation of large language models?

I think we'll see similar , trajectory, just like how in pre training in, like post training, that sort of system for getting commodized, where the serial will be a lot more commodization there.

I think the reasoning kind of models will go through a similar trajectory, where in the beginning, one or two players will know how to do it, but over time, like...

That's..

And who knows, right?

Because open AI could make another advancement to focus on.

Correct.

Right now, reasoning is their most.

But minute, if advancements keep happening again and again and again, like, uh, I think the meaning of the word "ad advancement also loses some of its value, right?

Tot.

Even now, it's very difficult, right?

Because there's preraining advancements.

Yeah.

And then we' moved into a different phase.

Yeah, so what is guaranteed happen is whatever models exist today?

Yeah. Dialogue of reasoning, that level of multi capability in, like, five or ten, cheaper models, open source, all that's gonna happen.

It's just a matter of time.

What is unclear is if something like a model that reasons at test time , will be extremely cheap enough that, like, being just just all run it on our phones, I think that's not clear to me yet.

It feels like so much of the landscape has changed with what Deep Seek was able to prove.

Could you call it China's Chat GBT moment?

Possible.

I mean, um, I think it's only probably gave them a lot of confidence that, like, you know, we're not really behind and no matter what you do to restrict or compute, like, we can always figure out some work around, and, yeah, I'm sure the team feels pumped about the, like, you know, results.

How does this change, like, the investment landscape, the hyers scalers that are spending tens of billions of dollars a year on capbacks that just ramped it up huge and open AI and anthophics that are raising billions of dollars for GPUs, essentially what Deep Seek told us is, you don't need.

You don't necessarily need that.

Yeah.

I mean, look, uh, I think it's very clear that they're gonna go even harder on reasoning, because they understand that, like, whatever they were building the previous two years was getting extremely cheap that it doesn't make sense to go justify raising that.

You' spending proposition the same?

Do they need the same amount of, you know, high end GPs, or can you reason using the lower end ones that Dupey could say?

Yeah, it's hard to say no until proven is not.

Uh, but..

I guess, like, in the spirit of moving fast, you would want to use the high end chips, and you would want to, like, move faster than your competitors.

I think, like, uh, the best talent still wants to work on the team that made it happen first .

You know, there's always some glory to, like, who that was actually, like, who's the real pioneer?

Where, who's Fast follow, right?

That was, like, kind of like Sam Alin's tweet, kind of veiled response to what Deep Seek has been able to.

He kind of implied that they just copied and anyone can copy, right?

Yeah, but then, you can obviously say that like, everybody copies everybody in this field.

Uh, you can say,G, go to the Transformer first.

It's not open AI, and open end is copied it.

Mm hmm.

Google built the first large language models.

They didn't prioritise it, but open AI did it into prioritised way.

So you can you can see all this in many ways.

It doesn't matter.

I remember asking you, being, like, you know,Why don't you want to build the model?"

Yeah.

I know.

That's, you know, the glory and a year later, just one year later, you will very, very smart to not engage in that extremely expensive race.

Yeah. That has become so competitive, and you kind of have this lead now in what everyone wants to see now, which is, like, real world applications, killer applications of generative AI..

Talk a little bit about, like, that decision and how that's sort of guided you, where you see perplexity going from here.

Look, one year ago, uh, I don't even think we had something like, um. .

This is what, like, 2024 beginning, right?

I feel like we didn't even have something like Son a 3.5.

Right.

We had GPT4, I believe, and it was kind of the nobody else was able to catch up to it.

Yeah.

But there was no multimeoral nothing. And my sense was like, okay, if people pay more resources and way more talent cannot catch up, it's very difficult to play that game, so let's play a different game.

Any way people who want to use these models and one use case of asking questions and getting accurate answers with sources, with real time information, accurate information.

Uh, that's still a lot of work there to do outside the model and making sure the product works reliably, keeps scaling it up to usage , um, keep building custom UIs.

There's just a lot of work to do, and we focus on that.

And we would benefit from all the tail end of models getting better and better.

That's essentially what happened.

In fact, I would say, Sonnet 3.5 made our product so good in the sense that if you use Sonnet 3.5 as the model choice within perplexity, it's very difficult to find a hallucination.

I'm not saying it's impossible, but it dramatically reduced the rate of hallucinations, which meant , the problem of question answering, asking a question getting an answer, Doing a fact checks research, going in and asking anything out there, because almost all the information is on the web, uh, was such a big unlock and that helped us grow tan hacks over the course of the year in terms of usage.

And you've made huge strides in terms of users and, you know, we hear on CBC a lot. , big investors who are a huge fans..

Jensen Huang himself, right?

He mentioned it in his keynote the other night..

It's a pretty regular user, actually.

He's not just saying it.

He's actually a pretty regular user.

So, a year ago, we weren't even talking about monetisation because you guys were just so new and you wanted to, you know, get yourselves out there and build some scale.

But now you are looking at things like that. Increasingly an ad model, right?

You have your experimenting with it?

I know, like, that's some controversy on, like, why should we do ads?

Whether you can have a truthful answer, despite having ads, and in my opinion, we' been pretty proactively thoughtful about it, where we said, "Okay, as long as the answer is always accurate, unbiassed, and not corrupted by someone's, like, advertising budget, only you get to see some sponsored questions, and even the answers to those sponsored questions are not influenced by them , uh, and questions are also, like, um, you know, I' not picked in a way where it's manipulative.

Sure.

There's some things that the advertiser also wants, which is they want you to know about their brand and they want you to know the best parts of their brand, just like how you go and want if you're introducing yourself to someone you want you want them to see the best parts of you, right?

So that's all there.

But you still don't how to click on a sponsored question.

You can ignore it.

We are only charging them CPM right now.

Sabara, we areselves are not even incent by to make you click yet .

So, I think considering all this, we're actually trying to get it right, long term, instead of going the Google way of forcing it to click on links.

I remember when people were talking about the commoditisation of models a year ago, and he thought, it was controversial, but now it's not controversial.

It's kind of like, that's happening.

You're keeping your eye on that , is smart.

But if we benefit a lot from modization, except we also need to figure out something to offer to the pay users, like a more soft research agent that can do, like, multi step reasoning, go and, like, do, like, 15 minutes worth of searching and give you, like, a analysis, an analyst type of answer , all that's gonna come, all that's gonna stain the product, nothing's changes there, but there's a ton of questions every free user ask, day to day basis that needs to be quick, fast answers.

It shouldn't be slow.

And all that will be free, whether you like it or not.

It has to be free.

That's what people are used to.

And that means, like, figuring out a way to make that free traffic also monetizable.

So you're not trying to change user habits, but it's interesting because you are kind of trying to teach new habits to advertisers.

They can't have everything that they have in a Google Ten Blue L search.

What's the response been from them so far?

Are they willing to accept some of the tradeoffs?

Yeah, I mean, that's why they're trying stuff, like Interior is working with us, and then many other brands , like, all these people are working with us to test, right?

They're also excited about, look, everyone knows that, like, whether, like it or not, five or ten years or not, most people are gonna be asking AIs, most of the things , and not on the traditional searches and everybody understands that.

They have.

So everybody wants to be early adopters of the new platforms new UX, and learn from it, and build things together, not, like, they're not viewing it as like, okay, you guys go figure out everything else, and then we'll come later."

I'm smiling because it goes back perfectly to the point you made when you first sat down today, which is necessity is the mother of all invention , right?

And that's what advertises are essentially looking at.

They're saying, "Th field is changing.

We have to learn to adapt with it."

Exactly.

Okay, Arv and I took up so much of your time.

Thank you so much.

Thank you for taking the time.

On the morning of August 2nd, 2016, Bitcoin investors woke up to news of a shocking digital heist.

Breaking overnight, nearly $100,000 Bitcoins have been stolen from exchange platform BitFin X .

I lougged into my account and noticed that my entire account had been drained.

I was crying and sweating.

The words started spreading on social media quickly. Was freaking out.

It was As panic grew, Bitcoin prices plummeted, and BitFX, one of the largest cryptocurrency exchanges in the world, halted trading, The price of Bitcoin dropped nearly 20% ring down the value of the stolen Bitcoins to about 65 million.

Wow.

And suddenly, thousands of people around the world have the same burning questions.

Who pulled off this high tech heist and how with the victims ever get their money back?

At first, the only clues were on the blockchain, and crypto detectives began following the stolen assets, spotting them in a wallet address they say begins with the digits 1CG.

We saw this 120,000 bitcoins taken from Bit FX and put in this Bitcoin wallet.

It's a little bit like you're robbing a bank in the middle of a stadium full of spectators.

Thousands of people are watching.

That's exactly what it is.

Now, the difference is, it is hard to know exactly who the individual is.

Just one problem .

No one could access the wallet without the key.

And as the loot just sat there, the trail began to grow cold.

When people kind of forgot about this particular hack.

For six years, as the value of the stolen crypto skyrocketed to over four and a half billion dollars, no one knew whud done it.

Was it a criminal gang, spies, digital anarchists ?

It was a global manhunt.

And the bad guys could be out of breach.

I spent my career as a prosecutor working full cash smuggling cases in cases involving Russian cyber criminals and North Korea hackers.

And there were never arrests in those cases because you just simply couldn't get your hands on those individuals.

But as federal investigators scour the dark web for digital clues, court records show, they focussed in on two suspects who weren't anything like what the experts expected.

I'm a mother lizard.

Married to a godd wizard.

In an early 2022, a blockbuster announcement in Washington.

The department has charged Ia Lichtenstein and Heather Morgan for their alleged roles in a conspiracy to launder stolen cryptocurrency taken during the 2016 hack of a virtual currency exchange .

It's the largest financial seizure in the history of the Department of Justice, and the announcement sets the internet on fire as the world discovers hundreds of videos, the young married couple shared on social media.

I mother is living on the coast that is that's super super weird story.

And particularly Heather Morgan , a self described serial entrepreneur and surrealist rapper who posts music videos on YouTube under her rap alias, Razzle Khan.

All of a sudden, I hear this voice rapping a female voice, and I look over on stage, and see who was on the microphone, and it's Heather.

She just picked up the mic and started rapping in front of the crowd.

In this music video, dedicated to misfits and hackers, she calls herself another name.

I real hard, but don't know where I'm heading Mother crocodile Wall Street.

If they are found guilty, the crocodile of Wall Street and her husband could be going to jail.

So who is the couple accused of laundering Bitcoin from one of the biggest crypto heists in history?

Ro To see these people acting the way they are and just being so open about their lifestyle and stuff, it was just fascinating to me.

Honestly, I could only laugh.

What led investigators to them?

Can you keep pulling the threads?

It's identifying that can follow up on.

And in a case like Bex, that's quite extensive because of the many layers of essentially wandering that was done.

Who stole the nearly 120,000 bitcoin from BitFenex?

The prosecutor's charged them here with the money laundering, but they didn't charge them with the initial hack, so that raises the possibility that there is a mystery hacker out there somewhere who's maybe not being charged.

Who do you think that person is?

It's a great question, and of course, what happens to the billions of dollars worth of recovered Bitcoin?

You want to know what fight is coming?

I can tell you right now.

3.6 billion dollars, people are going to spend hundreds of millions of dollars to get their hands on that pot of gold.

I do think it's going to be a fight.

I can imagine that in the future there will be movies about it.

Matt Price is a former federal agent for the IRS.

A lot of people don't even realise the IRS has Special agents that carry guns that are law enforcement officers, just like FBI or TE agents.

And because of that, as an IRS agent, we use the same tools.

That's physical surveillance, that's walking the streets of Manhattan, following someone around, online undercover operations, as well as using undercover agents for meetings and things like that.

It was those gun toting IRS agents who led the BitFEx investigation.

Matt cannot talk about the details of this case specifically, but he can explain how investigations work.

VIFEX, in particular, because of the complexity of it, at some point, you kind of almost have to wait for the next transaction .

Most federal financial investigations are quite time consuming.

You know, you are requesting and obtaining records from numerous sources.

It's banks, exchanges, mail service providers.

Look, when you have that kind of light bulb moment, you're like, "I got them.

That's when it goes into overtrack .

In this case, that light bulb moment led the feds to a surprising location for a huge cyber crime.

There are certainly nation state actors out there that engage in this, but given the size and the brazenness, you know, that is just a thought.

But the trail didn't lead to a nation state spy.

Instead, investigators say the evidence pointed to an apartment.

In this high rise on Wall Street.

Details revealed in court documents describe the dramatic moment on January 5, 2022, when IRS agents knocked on Heather and Iias' front door.

They found Heather crouched next to the bed, calling for her pet, an exotic Bengal cat named Clarissa, who has her own Instagram account., fur.

Then IRS agents say they noticed a bin under the bed with a bag labelled burner phoneoam.

Mle cell phones, SIM cards, and assorted electronics.

They also found hollowed out books in Ia's office, a file labelled "Passport Ias" with links to Darknet vendors and about $40,000 in cash .

All of it gave the feds the evidence they needed.

This is the largest seizure of cryptocurrency ever by U.S. Law enforcement.

As news of the arrests broke, the Department of Justice, announcing just within the past couple of minutes, the arrest of two individuals in Manhattan this morning.

The world learned more about the couple arrested in the alleged 4.5 billion dollar laundering scheme.

And the fact that so much bizarre content about them was available to gawk out on social media helped the story go viral.

I'm obsessed with Heather Morgan.

Her story is wild.

Please go down, this rabbit hole.

You will not be disappointed.

Her husband, Ilya, is more private, so he does make several appearances in his wife's TikTok videos, and he seems annoyed by the captain.

If you're interested in what I say, recorded the first time.

And in this video, where Ilya reveals he tasted their pets's cat food, he says, "You just keep filming me expecting something happen What do you want me to do?

You want me to just, like, shove something up my little dance?

Just three days after the couple's arrest, Netflix revealed plans for a docu series, directed by an executive producer for Tiger King.

Like a crypto version of Joe Exotic, videos of Heather on social media give us a peek into the private life of a mystery woman, accused of a multiillion dollar crime.

I mean, this really dope orange ballalava, which is inspired by Joe Exotic.

On TikTok, she posted videos, showing off art projects where her medium was taxidermy.

It's a camel school pretty d, right?

And shows off quirky talents, like using chopsticks to eat with her toes.

People also discovered Razzlecon, Heather's alter ego, a satirical rapper serving up rhymes and videos, she describes as something between an acid trip and a delightful nightmare that combines her fearless entrepreneurial spirit and hacker mindset. Your password all your transferred.

In court documents, prosecutors say those rap lyrics show her technical proclivities and note spearfishing is a term that refers to a hacking technique to gain access to a user's password and then transferring all of the user's funds out of the account, Even so, their friends were surprised to hear of the couple's arrest.

I heard the news, the afternoon, the story broke.

I was a bit of shock, honestly.

I just kind of sat there dumbfounded all afternoon.

I'm just trying to completely understand what on earth was going on.

Matt P met Heather a few years ago at the Greenwich Village Comedy Club in New York.

A friend of mine to Heather back in 2017.

Heather was doing a cold email comedy row from a.

Honos reaction..

That's what we want.

I was the comedian, host, MC of it.

It was also his first time meeting Ilya, who went by the nickname Dutch.

Dutch makes a noise for Dutch, everybody.

He's gonna need unicorn.

You could tell that Dutch really supported her, she really supported him.

He sat in the front row, right there, and even did not mind heckling me in any way, shape or form.

You could tell that even his personality and sense of humour was very different the way he would heckle me.

At the time, Heather was the CEO of a company she started called Salesfolk that writesObound sales email templates.

I could tell that she was a good business person.

She had a couple employees at the time .

It seemed as if, though, she was trying to make this bigger than what it was.

Sometimes she would throw some numbers out about what the company was doing and all, and I would go,Okay, that seems good, but comparing it to how many employees she had, and it wasn't so sure about that.

I didn't look at it as a red flag.

It was just very New York in that way, of everybody kind of,W, add a zero.

Heather also landed columns at Forbes in " Ink magazines, one focus of her writing, cybersecurity .

In court documents, prosecutors point to an article titled "Eperts Share Tips to Protect Your Business from Cyber Criminals."

And note that Heather interviewed several virtual currency exchange employees who discussed their company's anti money laundering and anti fraud protocols.

But for outsiders watching videos of Heather and Ilya on social media , the idea that this couple could be charged with conspiring to launder billions in cryptocurrency didn't seem possible.

And Twitter lit up with speculation about the audacity of it all.

Though investigators say the blockchain doesn't lie.

But the nature of the blockchain is open ledger, where crypto sort of lives and moves, allows for visibility.

We work with law enforcement to trace and track the flow of funds.

You might expect someone who calls herself the crocodile of Wall Street, who IRS investigators say, has access to billions in stolen Bitcoin, to be living lavishly in Manhattan.

By New York standards, they were living a great life, but it was almost like an everyday New Yorker life.

They had a nice apartment.

It was a little bit bigger than what you normally would see when you go visit your New York City friends.

Photos of Heather and Ilia's apartment on Zillow show a spacious, two bedroom with a modern kitchen and marble bathrooms.

Videos inside the apartment reveal Heather's passion for decorating with taxider like crocodile heads, antel skulls, and even zebra skin hanging on the wall.

The apartment rents for an estimated $6,500 a month, expensive, but not outrageous in the pricey big apple.

When we would go out, she always would pay the bill.

Heather and Dutch would definitely splurge on travel , but they were still kind of very proud to be almost a man of the people.

They weren't too lavish.

How much was this?

I'd be shocked if this is more than $4. $7 dollars?

My .

The problem is, even though the stolen crypto is worth billions, converting those coins into actual cash, isn't easy.

It's difficult to launder a large amounts of cryptocurrency in short periods of time, because it draws attention.

If you try and just move that all to an exchange and cash out , then that's going to set up alarm bells.

Investigators say about 80% of the stolen bitcoin never moves from the digital wallet, the hacker transfers them to, but according to the criminal complaint, the couple did try to cash out some of the crypto.

Investigators say Lietenstein uses some of the digital currency to purchase gold from a precious metals dealer using his home address for the shipment.

None of the gold was found in the couple's apartment.

Funds connected to the hack are also used to buy a $500 Walmart gift card .

Investigators say Heatherather reddeems in her name, and New York City address.

The couple's e email addresses are also connected to gift cards purchased for Uber,els.com and PlayStation.

When you have that interface between the crypto world and the real world , you in some way disclose your identity, if you're buying a gift card with crypto, you generally need to give your address to whichever retailer the gift card is from.

So how do Heather and Ilia end up here, living just like everyday New Yorkers, right up until the Feds come knocking at their door?

I come Ilia grew up in Glenview, Illinois.

He moves there with his parents from Russia, a citizen of both countries.

I mean, in many ways, it's just, like, a pretty average suburb of Chicago.

He meets Charles Austin encoding class, in high school, in terms of extracurriculars, I think he mostly just took all, like, the coding classes that were offered at the school.

We're definitely various types of nerds, I guess, in the classic jock versus nerd thing, where obviously nerds, right?

We're not , football quarterbacks here.

I would wager that he never had a girlfriend in high school is my feeling.

And, like, I only had a girlfriend in, like, senior year, you know what I mean?

Like, we were not, like, cool overall.

If you had asked me about him the day before this story broke, I would have said, like, "Oh, yeah, I remember that guy.

Like, he's super nice."

He was, like, quiet and somewhat awkward, but if you actually talk to him, he's super funny and nice after graduation, Charles never sees Ilya again.

So, Ilya went to University of Wisconsin medicine, which is a fairly common school to go to from Glenbrook South.

I don't think it gives you a huge amount of insight into his direction in life, because it's not a crazy place to go.

I think it's more interesting that he ended up in the Bay Area after that, where that speaks to some amount of direction .

That's where Ilya will eventually meet his future wife, Heather grew up inama, California, a small agricultural town, only 400 people call home.

I am from the smallest city in California, originally, believe it or not.

In this YouTube video, Heather talks about her family's humble background .

My parents didn't have a lot of money.

They both work for the government.

My dad was a biologist for the government, and my mom was a high school librarian.

And I've also been totally broke and homeless, multiple times.

For college, Heather attends UC Davis, near Sacramento, studying international relations and economics.

Professor Travis Libbert remembers meeting her on campus.

The first time I met her, it was in the context of a guest lecture I gave to a big, big undergraduate class.

She did introduce herself, but ..

And it was about a year and a half later that she followed up by e email and we began to correspond at that point.

Heather graduates during the summer of 2011 and moves to Hong Kong for work.

She was living and working in Hong Kong .

She was applying for a master's program at American University of Cairo.

Six months later, in January of 2012, Heather reaches out to the professor to pitch herself as a research assistant.

It was sort of coincidental that just a little bit before she sent me an e email, I had agreed to I had a chapter in a book that was focussed on food insecurity and sociopolitical instability.

At that point, just in the wake of the Arab Springs. .

He hires Heather to help with data collection and analysis, but within a month, he offers her a bigger role.

She sort of proved herself in those early stages enough that I invited her to join me as a co author on the paper.

She had a very quick mind to his fast learner.

There was probably a period of about eight or nine months that we were working pretty intensively together.

All by Skype.

She was in Hong Kong at the beginning of that period , and then had picked up and moved to Cairo.

Mostly, they talk about economics, but the professor gets a glimpse into Heather's personal life.

She was very fond of belly dancing.

I know that, because when she got to Egypt, she was really thrilled that there was quite a lively belly dancing culture.

What he couldn't see coming was that his co author would later become a rapper, accused of laundering funds from one of the biggest crypto heists in history.

I can't say that there were sort of glimmers of this in our interaction, but but the one thing that does stand out is that her signature block , had evolved over that brief period under her name was economist, entrepreneur hustler.

Heather leaves Cairo in 2013 after just one semester at school, never finishing her master's.

Instead, she heads to Silicon Valley, where her aspirations shift from studying economics to becoming an entrepreneur.

When she lands a job at a Middle Eastern gaming startup called Tamatem, which will give her a front row seat to some of the biggest names in tech , through an early stage, global Business accelerator program called 500 Start Ups.

The Connections Heather makes here will change the course of her life forever.

So with Heather being in that space itself gives a lot of credibility that you are part of this group, Silicon Valley itself is competitive.

And within that, here's a bunch of 30, 40 companies who are crazy enough to lock themselves in office for 18 hours, for four months in a stretch to build something unique.

Startup founder Talvinder Singh recalls meeting Heather at orientation in early April of 2013.

So they were founders, and then they were employees of the companies who were also involved.

So Heather was the employee company calledatum.

Of course, she has a personality that ensures that she will get noticed.

And suddenly, a small town girl is breathing rarefied air.

Every week, there will be at least two sessions with the brains from the founders would include dropbox founders.

During these intimate fireside chats, the budding entrepreneurs learned what it takes to succeed in Silicon Valley from their idols.

You know, when you get exposed to such people and you hear their stories of how they hit after they made it work.

If you start believing that if you're always convert and note to yes, you can always figure out a way to make something work for you.

In this intense environment, founders work 16 to 18 hours a day under one roof, and Heather wasn't exactly like any of the other founders.

There's something odd about her and she owned it.

It's like, you know, I'm weird and I. I'd it.

Heather shows off her unique personality while filming an internal company video.ups to create a music video.

We rift was a song, very c I remember I have some of us going in a ugly ass van going to the thrift shop, picking up stuff.

For their parody of Macklemore's hitThrift Shop, founders dress up in wacky clothes like thrifted fur coats, but out of nowhere, Heather decides to take her clothes off , stripping down in front of everyone.

Without any plan oration, she decided to strip to her underarments. Else was that even remotely close to it.

So it was pretty odd.

Maybe we got the first view of Ral Khan.

And Heather makes other choices that raise eyebrows.

I think she had that, you know, completely raw passion to achieve more perhaps costs.

People who are ambitious, people who are opportunistic, they jump from oppative opportunity, and they go to the next level as fast as possible.

Just three months into the program, she grabs an opportunity to work with another company from Brazil called Pin Myai Pet, and marries one of the founders. Doing the last three, four weeks, the news came out. A whirlwind marriage and a breakup in Brazil, Heather comes back to the Bay Area and starts her digital marketing company, salesfolk.

She also meets her next husband, Dutch Ilia Lichtenstein, in a tweet from June 2013, Heather writes about listening to Ilya Spe at 500 stups.

At the time, Ily is running a company he co founded called Mix Rank.

The Mix rink is a spice for displays and what's for.

Mix Rank receives investments from why combinator 500 startups, and even Mark Cuban.

Less than a year later, Heather tweets that she's travelling to Hong Kong with Ilya, and he becomes an adviser to salesfolk.

Heather was going through a really rough time, through a divorce, making some life changes, and Dutch really was by her side.

That's formed into a friendship early on, and then eventually became a relationship.

Then, in January of 2015, court documents show Ilya opens an account at an unnamed virtual currency exchange, simply listed as VCE5.

It's verified with photos of his California driver's license and a selfie style photograph.

It's a picture that will come back to haunt him years later, when law enforcement uses it to connect him to an infamous wallet on the blockchain.

Ia leaves mixed rank in May of 2016 , Three months later, Bitfex is hacked, and nearly 120,000 Bitcoins are stolen.

On a hot, summer morning in Fort Worth, Texas, Frankie Cavazo's logs into his computer and discovers some alarming news.

The day of the hack on August 2nd, 2016, I had hopped onto Reddit, and I noticed that people were posting the the bit for next platform had completely frozen, and there was 15 bitcos that were just completely removed from my account.

Time, $10,000, that was a small fortune for me.

It was a hot Texas summer, and I was crying and sweating.

A few years before the hack, Frankie was working as a welder.

He saved up and bought some Bitcoin, hoping the investment would help him pay for college.

I sat on those for two years, just kind of stared at them, and Bit FX started marketing for how they were going to revolutionise how you could use your Bitcoins.

I decided to move my funds over there, about six weeks before the hack.

On the other side of the globe in Poland, Rafal bennia learned his bitcoins are gone.

Let's in my office and an email from them there were investating.

I was super scared.

At the my said 91 bitcoin was bounce more than 60,000.

It was gut wrenching.

I had panic attack.

At the height of the market in 2021, a single bitcoin was worth $65,000.

If Rafall still had those 91 bitcoins, they would have been worth nearly $6 million bucks.

Obviously, everyone was freaking out.

I would say I had about 20 bitcoin on there.

Will Hogarth is in Canada frantically searching for information on social media. Discussion that community on how are you going to make people hold?

A lot of people are arguing that the losses should be socialised around everyone who held the balance.

Four days later, Bidenex makes an announcement stating, "We have arrived at the conclusion that losses must be generalised across all accounts and assets at a rate of 36%.

That means everyone who had an account at BitFX lost 36% of their holdings on the platform, not just users whose accounts were hacked.

To pay people back for their losses, bid finex gives account holders an IOU in the form of a digital token labelled BFX .

Customers receive one token for each dollar lost.

When they dumped all these BFX tokens on the consumers, they tagged them to a $1 dollar per BFX token, and then they put them on the open market, and they went from $1 to, like, 20 cents.

So they were essentially allowed to , basically FOMO everyone out of their debt.

I as 25% of data value.

There was no point in time that they refunded me, not in dollar terms and not in Bitcoin terms.

Then in September, BidFex gives BFX token holders the option to convert their IOUs into equity shares of IFX, the corporate entity behind FitFNX, through a so called recovery rights token.

If the company receives the recovered Bitcoin back, those token holders will each be paid a dollar.

We were pissed.

They're taking all the risks in terms of, okay, here's this token, here's debt that we're gonna issue to you, and don't worry, we'll pay it back as long as everything works out., from a customer's standpoint, like, that's pretty frustrating, because, you know, I didn't agree to take on, insecure debt from FitFX.

I just was trying to trade Bitcoin on their platform.

Bit Fex declined CNBC's request for an interview, but said in a statement,We have been very clear about the BFX tokens and recovery rights tokens that were issued to compensate customers harmed by the security breach, and we have met and will continue to meet all of our obligations.

We are very proud of the proactive steps we have taken to compensate our customers, even though it was not known whether the stolen Bitcoin would ever be recovered .

Meanwhile, blockchain investigators are tracking the stolen bitcoins, finding them on the blockchain is pretty easy.

The nature of the blockchain is sort of open ledger.

You know, every transaction is recorded and immutable.

So it really allows more visibility on financial flows than really we've ever had before.

A Redboard is the head of government affairs at TRM Labs, a blockchain intelligence firm that works with government agencies on investigations like BidFX.

The blockchain is partly anonymous, right?

Because we don't know who owns the wallet, but we do know where the wallet is, and what's in the wallet, right?

That's right.

So they' let's suit anonymous.

Investigators could track the stolen funds, they just didn't know who owned the 34 digit alpha numeric wallet that was holding them.

So the trick is to take those numbers that you have that are the wallet and match them up with a name, which is a real person.

So ultimately, that is up to investigators.

And what they used a blockchain analytics tool, like TRM to do, is to track and trace the flow of funds that are potentially associated with, again, illicit activity, like a hack.

In the same way, a die pack explodes on a bank robber covering the stolen cash with red ink, making it impossible to spend, the digital coins are marked electronically for fraud.

So crooks have to find a way to clean the crypto before cashing out.

The Bitcoins first started to move in January 2017, and Roy saw it first were the funds being sent to a dog at marketplace called Alphabet , a dog web, retailer, basically, where you can buy everything from notics, the hacking at tools, to fake ideas.

On the surface, it may seem like navigating the dark side of the Internet to launder crypto would take the know how of a James Bond villain.

But in reality, all it takes is a special browser .

It's very easy to to download the tour browser and use it.

You could do it in under five minutes. And you'll see here it looks very much like a normal internet browsing experience.

Cyberse security expert Sandra Joyce is showing us around a Darknet e commerce site similar to Alpha Bay.

It's like doing deals in a dark alley where buyers and sellers want to remain anonymous.

We're going to browse actually to a Russian language site called XSS.

This is basically a very professional site.

They even have rules about conduct on the site, a sort of criminal code of conduct.

They even have an FAQ here, where you can see what are the rules in buying and selling?

How do you deposit money?

How do you do it anonymously?

You don't really need to be that sophisticated because they really spell it out for you.

We'll just search for crypto, and we're using a translation feature here , so that you can, you know, see what's being said in English.

So this thread actor, for example, is claiming that they're gonna be able to cash out your cryptocurrency here.

So this is the type of activity that somebody who's trying to move cryptocurrency would find something like this very helpful.

According to the criminal complaint, the couple also used alpha Bay to exchange some of the stolen bitcoin for a privacy coin called Mon ero.

Mono is considered a privacy token, which means that it hides virtually all transaction details, which allows cyber criminals, greater freedom from some of the tracking tools that authorities have developed to trace payments on Bitcoin's blockchain.

But investigators say, Heather and Ia didn't stop there.

They were using absolutely every trick in a book.

They used a money laundering technique, involving what they call sort of mixers or tumblers.

And what they do is they take cryptocurrency, Bitcoin in particular, from all kinds of different users, and then they scramble all of that up, and then they spit it out the other side, redistributed.

So your Bitcoin, they go into the mixer, come out, essentially, as someone else's Bitcoin on the other end.

Another laundering technique outlined in court documents, is called appeal chain.

Say you have a single wallet containing a lot of dirty bitcoin from some kind of criminal activity.

You'll do is deposit it in small amounts and so from this wallet, you'll send a small amount to the exchange, but you'll send the bulk of the funds to a new wallet, and then you'll send a small amount from that to the exchange , the rest to another wallet.

And so at each stage here, you're peeling off a small amount of the Bitcoins, it also adds hops in the transaction chain, which are harder for the exchange of trace back and link to the criminal activity.

The money moves made the crypto harder to trace, but not impossible.

In February of 2017, just one month after the Bitcoin first starts to move into Alpha Bay, investigators say the account Ilya opened on VCE5 receives a deposit.

According to court documents, one month later, three additional accounts opened at a different exchange called VCE7 Start receiving Bitcoin as well.

And over the next four years, around 2.9 million dollars worth of Bitcoin flows into those accounts at VCE7. , Nearly all of it is converted into cash and transferred to bank accounts held by the couple, according to the criminal complaint.

And the feds were closing in on their suspects.

I think that a key mistake was using alpabet, and that was potentially because alpha Bay was taken down by law enforcement a few months later.

Alpha Bear gets shut down in July of 2017, and the Feds gain access to valuable information about its users who thought they were using the site anonymously.

Today, the Department of Justice announces the t down of the dark web market alpha b.

Meanwhile, Heather starts writing for Forbes magazine.

She's also working on a new startup with Ilia called End Pass.

This YouTube video posted by the company describes their products.

Meet and Boss are 100% open source ifhereum and token wallet, you'll be delighted to use.

We protect your f with multiple layers of security.

Urine could costume needs ill a lot interviewing for a product marketing position at the company in October of 2018.

What did he tell you about this company and what it was going to do?

He says that Z trying to create basically two companies.

One is new type of cryptoet one is the software for password management .

So, did he seem like an expert in cryptocurrency to you?

I don't think so.

Did the company ever build the product that they said that they were going to build?

Oh, no, no.

No, not at all.

And during all my work for six months, marketing team literally didn't do anything.

This is around the same time.

Heather launches her rap career in 2018.

The couple's not married at the time, but Heather is listed as the CEO of Ia's new startup.

Did you guys see the videos when you were working for her?

No, no, no.

At least, uh, I didn't see them, before all this news.

So she came off as a professional business person, not a rapper or an entertainer.

Yeah?

And did she seem to have any cryptocurrency or security skills, as far as you could tell?

Uh, no, no.

She didn't even know how much is it like optimisation.

I don't.

According to Yuri, Heather and Iia didn't have the skills to build a successful cryptocurity company , and with no product to market after six months, in March of 2019, Ilyia fired the entire marketing team.

We were sitting around, was talking about the future plans. , are will pay for the next month's, salary.

And that was it.

Yes, that's for it.

Three months later, Ilya asks Heather to marry him, sharing the elaborate details of his proposal on Facebook , he writes, "What better way to propose to an entrepreneur/rapper than with a weird creative multannel marketing campaign?

He hires an agency to place Rosal Khan posters around New York City, and even buys billboards in Times Square.

Shortly after their engagement, employees from BCE7 start reaching out to Heather via email with questions about how she came to own the Bitcoin in this account, according to the criminal complaint.

Her reply, "My boyfriend, now husband, gifted me cryptocurrency over several years."

On a summer night at the Williamsburg Hotel in 2019, the crocodile of Wall Street stands up in front of an audience and starts rapping.

Mother Crocodile Wall Street.

Heather Morgan's warming up the crowd at NYC Salon, a speaker series that aims to be a TED Talk with Friends , a PowerPoint slide above her reads, how does social engineer your way into the subject of her speech?

Social is I' manipulating, buts to share information action that they otherwise would not.

Usually, social engineering is a term used to describe digital techniques, hackers used to trick people into sharing confidential information , like account passwords through fishing or malware.

But in this context, Heather says, "Scrappy entrepreneurs can use social engineering in real life to get a leg up in the business world. Do that your social We askedmeus Reed, a warden professor who studies social psychology to watch Heather's presentation.

My exact reaction to this was like, "Wow, this is either an interesting sort of conversation about how to hustle, or this is perhaps the most egregious master class and deception you've ever seen that is being paraded as this intellectual persuasion exercise.

Under a slide titled "Some Places I've Infiltrated, Heather tells the story of she and her friend breaking into a historic Hindu temple in cairo . Security.

She says they offerered the gu a cigarette and an Egyptian p, convincing them to give them a tour instead of placing them in handcuffs.

Some of the pl I've infiltrated, that word tells me something.

The aura here is very much of the flavour of something that is manipulative, it's a little bit dark, it's a little bit seedy.

I'm getting in, even though I'm not supposed to.

Heather also gives the audience tools for crashing events and explains how to socially engineer yourself out of a bad situation.

At the end of her presentation, someone asks if she ever feels guilty about using the shady techniques.

Her reply? , my goals.

Like, I'm not trying to scam someone out of my ear.

Like, like, get someone hurt in any way.

She is literally saying, I have a line, and I don't believe that I've crossed it."

A few weeks after this event, in August of 2019, prosecutors say Heather and Ia leave for Ukraine.

During the month long trip receiving multiple packages with fake Ukrainian IDs , bank cards, and SIM cards for mobile phones.

After that trip, it'll take investigators two more years of searching for clues before they led to the person with the keys to access the digital wallets, holding all that stolen Bitcoin.

And in a typical investigation, you I followed the cryptocursy along the blockchain, but I've worked with exchanges, I've subpoenaed records, I've identified , e email addresses, and I've built out, all right.

This is my target.

Have this evidence to implicate this individual in the crime, but I need the additional evidence.

I need the possession of hardware wallet that may hold some of that picture, or the private keys to access some of these wallets, or emails .

So, where we're at the point where I know who this subject is, I'm looking to get in front of that person.

I want to get into the location where I think the evidence is that's gonna implicate this person, and a lot of times, that's their house, their business.

If you see on TV, the FBI or DVA, you know, SWAT ramming down the door, the IRS, they' do that on a regular basis.

You have a federal search warrant, you're making entry into someone's home.

There's all kinds of prep that goes into that.

You're doing surveillance, you're trying to get a feel for the area, And while IRS agents are preparing to execute a search warrant, Heather and Ilia are planning their wedding in California .

Heather posts a video from the ceremony on Twitter.

She's wearing a gold gown and headpiece rapping on stage while her bridal party waves golden palm leaves.

Turkish Stewart!

Go!

Turkish smo Stewart!

Whoo!

But the celebrating is short lived .

Less than two months later, federal agents sear the newlyweds Wall Street Department.

By the end of January, investigators say they're able to decrypt several key files in Iia's Cloud storage account, including a file listing all of the addresses within Wallet 1 CG, and their corresponding private keys .

And that gives the Feds access to more than $94,000 stolen bitcoins.

When the news breaks, Bidenex releases a statement saying it has a right to the money.

BitFex will work with the DOJ a follow up appropriate legal processes to establish our rights to a return of the stolen Bitcoin.

But some of the customers believe the Bitcoin belongs to them.

I still am going to be trying to get a hold of these 15 bitcoins because I truly believe they're mine.

I can prove it through the blockchain Explorers.

I still expect my bitcoin back, and I don't see any reason why they would keep it.

Why would anybody question that' I should get my money back that my property.

Ultimately, it's going to be a dogfight who gets this money, whether or not gets to keep it, whether or not Bifinex gets to keep it, whether or not the customers get it back, anyone who tells you there's a clear answer is lying for their own benefit.

David Silver is a lawyer specialising in cryptoisease.

I would say the odds are best the government is going to keep this money.

The government will try its best to give restitution to victims who can show they actually are the victims of the Biponix.

You're gonna have to show how you purchase the Bitcoin, how you sent the Bitcoin to Bipenex, and that you're the rightful owner of that account, and that you're the rightful owner of all that crypto, and that's going to be very, very hard to do for a lot of people.

The multi billion dollar battle over the money could take years , and David expects, over time, more details will emerge about the hack that led to the digital heist.

But so far, the hacker's identity remains a mystery, and no one has been charged with the crime.

So, the end of this story, we don't know yet, but you can't just simply walk away with a hack like this .

There's someone that's gonna be caught up in this that has to tell the truth, and when that shoe drops, it's gonna be really interesting, and it's gonna impact who gets the money.

In the meantime, the crocodile of Wall Street and her husband are charged with conspiracy to commit money laundering and conspiracy to defraud the United States , facing maximum sentences of more than two decades in prison, but their odd love story lives on through the lyrics Heather wrote in a song called "Mon and Stars."

She released it on YouTube just three days after the Feds raided their apartment.

Two for average, everyone knows you're the best for me.

This is how our story goes We reached out to Heather and Iia to hear their side of the story.

Hi, you've reached Heather.

Please your name and get back to you.

Cheers.

Both Heather and Iia are innocent until proven guilty, and neither agreed to an interview, so there's still some questions we don't have answers to.

How did Heather and Iia allegedly get their hands on the stolen Bitcoin?

And who pulled off that65 million hack?

But it's clear that the feds are watching the blockchain, and technology is making it more difficult to make money from crypto crime.

Lisa Monaco is the deputy attorney General of the United States.

The BFenex arrest of the couple in New York was dramatic.

The largest seizure of cryptocurrency and history.

What does that case tell us now about what the Department of Justice can do to track cryptocurrency through the financial underground around the world?

I think what it tells us is that we're at a crossroads when it comes to cryptocurrency.

We have to be focussed on the use of crypto to high dog gotten gains, to launder money, to hide proceeds.

And what you're seeing from the Justice Department is an enhanced focus on that.

You've seen us launch a new team of prosecutors, the National Cryptocurrency Enforcement Team.

We've got the best people on this.

We've got expertise that we're bringing to bear, and what you're seeing is our ability to both combine very sophisticated tools and abilities that are behind the scenes with ability to trace the movement of money, of ill gotten gains.

And that's what you saw us do in Bitneck.

And we've got to be focussed on that,cause as the saying goes, that's where the money is."

And Blockchain investigators say the DOJ's arrests in this case should send a clear message to crypto criminals.

It may take years, but they're gonna find you and you're gonna get caught.

I think Law Forcement capabilities have advanced, and can very sophisticated.

So you think life is gonna get harder for the bad guys?

I think life is gonna get harder.

The law enforcement techniques have gotten better and so have the illicit actors. .