Alex Blania on Worldcoin's US launch: 12M verified users, racing to build proof-of-human before AGI makes the internet untrustworthy

the stream how you doing what's going on tools for humanity founder of tools for humanity sorry worldcoin is one of the co and the world the orb There's a whole ontology of of technology here, but thank you for joining. How are you doing? Good. Thank you for having me. I'm just at the Andre Horitz LP day. Oh, cool.

Yeah, we just got off You just got off stage. I literally just got off stage. So, almost was a little too late, but um yeah, right now in Las Vegas. Uh first time, actually. First time in Las Vegas. I'm really excited. Wow. Narrative violation.

Somebody building in crypto that hasn't been to a crypto conference in Vegas. Um no, it's great to have you. We we've been wanting to have you on the show for a while. Glad we made it happen. Uh and it times nicely with your guys' entrance into the US market in a big way. Yeah.

I mean, maybe you should kick us off with kind of like an overview of where the company is, where you started. I know there was an international component. Now you're in the US. There's different products. There's um different ways to interact with like the overall ecosystem.

Kind of give us the state of the union on tools for humanity. Yeah. So um maybe uh starting right right from the beginning. So like a little over five years ago now uh five and a half years ago um Sam Alman from OpenAI and uh actually had part of the initial idea of this and we we met I was in AI before as well.

uh it was uh in theoretical physics and deep learning and so both of us believed that AI will continue to make a lot of progress and very likely we will see AGI within our lifetime which back then was not like a common thing to say. Yeah.

And um and of course like both of us believe that this is mostly going to be a tremendous force uh for good and it's going to drastically accelerate science and technology but also that we will need global scale infrastructure um to kind of really make this a great outcome for everyone.

And so we um started a company together called tools for humanity um that really had that as a mission is like build global scale infrastructure to help with that and um and then the actual kind of coming together of the company there were two main ideas.

One was around the fact that as AI continues to make progress, we will need uh some form of proof of human that is at internet scale. Otherwise, many of the things that we care about will start breaking down because um obviously like as a society, we form our opinions on the internet on our social networks.

We meet each other on platforms like dating apps. We entertain each other with uh games that that we play with each other through the internet.

And like the moment we have uh not only passed the touring test which I think we did in the last two years meaning now we have AI systems that we can talk to and we cannot tell anymore just by talking to them that in fact is is an AI but also that we are now about to enter of course like uh strong agentic AI systems real-time video generation all these kinds of things.

Um and so we will need a proof of human internet scale otherwise uh social networks many of the places where you meet online etc will start breaking down and so this was idea number one it's like how could you build such a system um and once you have built such a system you can use that to launch a digital token by giving ownership in it to eventually every human and as a result you would create what will turn out to be the largest real human network on the internet and so it was a very humble vision uh used to by by Sam.

Um and and so we really uh kind of we took it very seriously.

We thought about it for a long time and um back then really thought about how would you need to build such a proof of human um if you assume what we did uh which is that AI will become increasingly better and then many of the initial ideas you could have on how to build such a system like why don't we just use our online reputation so like our accounts that we used in the past we build some kind of reputation graph or what about if we just rely on government identity systems or what if we just use phone cameras to to get some kind of signal.

All of these we realized pretty quickly will either um just break down as a result of AI becoming more okay just like more powerful and better or it just doesn't have the properties that we think such a system will need which is it needs to be global and it needs to be fully anonymous and privacy preserving all of those things.

And so well back then we realized we will likely have to um build a physical hardware device that we have to distribute all over the world and at some point get uh almost anyone that uses the internet to verify with such a hardware device and that everything else will very likely break down.

And that was like a very a very very crazy thing to do uh 5 years ago. So like people mocked the company for a long time uh and that yeah it was like just a crazy thing to work on for quite a while. Um, and so essentially talk about uh just going deeper there. You guys uh went to market internationally first.

I'm sure you had a lot of temp temptation to really focus in San Francisco early on, right? This is a an area where there's an extreme density of people that understand this broader vision that you guys have.

So I'm curious like thinking about um you know the decisions that that went into that go to market and then also giving our audience a sense of the scale of the operation because I don't think people fully process just how large the operation is.

So um first to the to the first part of the question which is why did we not launch the United States much earlier?

Um that had a simple reason that because Sam is a co-founder uh everything we do has like a lot of visibility usually and uh in the last administration uh there was basically no regulatory clarity around how to behave uh as you build these crypto networks and kind of what are the things that are allowed which which are the things that are not allowed and SEC of course like had a posture of enforcement and so given the large visibility of the project.

We just took everything always through a very cautious lens and just decided that we will only enter the US once there is a clear path to regulatory clarity which I think is now the case. Um and so now we really excited to uh kind of make the servers available. We launched two weeks ago in the US.

Uh in the US itself it's still a meaning kind of pretty small uh operation. It's going to take us a couple months to really ramp this up. Um, and just to give you an overall sense, we're now at 20 25 26 million users in total in our app world app, which is essentially the initial client to make this network useful.

Uh, but kind of you have to think about that there's going to be many clients that will use this this network.

Um but but still and then we have 12 million uh people that actually are verified uh really all over the world from South Korea, Japan uh to um Europe to Latin America and and so yeah the company is now around 4 or 500 people.

also tools for humanity and it's a pretty it's a pretty complicated thing to manage because you have everything from on the ground operations um that are somewhat complicated and a lot of things can go wrong to uh of course hardware manufacturing and ramp up to uh a lot of software so it's it's been quite a quite a journey what what is your background have have you done crypto stuff before or hardware stuff before or managed huge like workforces before because you're doing a lot of stuff and I'm wondering like no I I was um I was I was I was literally a theoretical physicist.

So just I used to sit at a desk all day and do math and write code. That's fantastic. I I I did like I did decent amount of self-engineering just because like already back then the clusters got larger but yeah I didn't didn't do any any of that.

So talk about uh yeah talk about how basically talk about bootstrapping the network and the user base with the token right and and um because that's I don't I think I have a good sense for why you would do that but you know the the sort of boomer push back would be you know why does this need to be you know crypto and I think you guys have good reasons for that.

Yeah. Yeah. I think there's two reasons for that.

like first of all of course appreciating that crypto still has like a meaningfully bad reputation just given everything that happened but I think um there's like real technological reasons of why you need to do or why why I think it's best to build such a platform that way.

So one is as it was mentioned bootstrapping where uh all of these networks that we use of course they're absolutely meaningless if they're not at scale.

Um and then so like um that applies to social networks um also applies to financial networks and and so maybe one of the most notable uh examples of that was the launch of PayPal where um you were able to refer your friends.

Your friends had some amount of US dollars in their PayPal account and and and very famously um PayPal spent a lot of kind of a huge amount of the renter funding.

I was of course not there so I don't know how much exactly but a huge amount of the venture funding to get this network to critical scale with that referral program and that then led to network effects and that then led to this turning into among many other things of course uh into a successful platform and so um the same applies here it's just that these uh crypto networks are valued at billions of dollars if they're somehow working and so I think we can do uh something similar just at a much much larger scale and so we can use uh very likely billions of dollars to kind of really get this network to scale.

And so that means that everyone that actually joins this network gets a small piece of it uh through the Rokin token. Uh and so you actually receive ownership and in in the quite literal network.

Uh but also of course that is an incentive to verify early on where maybe the utility of the network is not fully there yet because it's subscale. And so that was that was that idea. And that the second idea is of course that like this is going to be one uh very expensive to build uh but then also extremely valuable.

It's going to be very valuable to many of the largest internet companies uh very likely if we succeed in our mission. And so we need a business model but not for us as a company but for the entire network. And so uh that is the second piece. I think these tokens actually allow you to build a sustainable network. Yeah.

Um yeah and I think now we're actually bridging that gap where it's actually becoming quite useful. We we now announced the partnership of Match which is the the company behind Tinder, Hinge, many of the dating apps to actually bring prov to uh to these dating platforms.

Uh partnership with Razer which is uh one of the largest hardware manufacturers for gaming hardware to turn that into the standard for for the gaming world. So I think like we we're actually getting there now pretty quickly. Uh but of course it's still early. Can you talk about the scale of bot activity online?

And I have to imagine a lot of people are interacting with bots daily that they don't necessarily realize are bots, right? There was a high-profile issue recently uh with Reddit where, you know, somebody was running a a scaled influence operation experiment, you know, leveraging, you know, leveraging the power of AI.

And for those that don't know, people were uh that the AI would effectively look back through somebody's comment history and deeply understand the person and then make arguments to them based on their sort of pre-existing beliefs, which is very powerful.

And and obviously it seems, you know, all the users were not aware that what was happening to them and I just have to imagine that's happening, you know, at a at a very wide scale. Yeah.

I so no one really knows of course the actual scale of the situation but I think the the paper that you outlined is from the University of Zurich.

Um so as as you described the University of Curric actually ran an experiment of changing public opinion u by using AIS to respond to just like a comment in subreddits and they actually have been quite successful in uh changing the opinion of these subreddits which is kind of crazy and so uh of course we have like a similar problem probably at meaningful scale on X uh and very likely now every other social network.

Um maybe uh so dating also already at a quite a meaningful scale like you have these now these these accounts that like have photo realistic images and start texting with you and everything.

But I think like maybe the more interesting part is that I think we're like at 1% or something of what I think will happen over the next 24 months. I think a a big trigger point will be once you have uh an open-source very competent agent.

Um because like that's yeah that's that's where like things will get quite exponential and we're like probably 90 more than 90% of almost all activity on the internet uh almost all content on the internet will be AI created uh sometime after.

So that will lead to the fact that I think for now or until quite recently by default we trusted everything that we have seen and we assumed that it's like authentic. I think that will completely flip by 180 degrees to by default we don't trust what we see.

We don't trust who we interact with if there is not a way to verify the claim. Um so it's going to be quite a profound shift. Uh what was your initial reaction to the to the Coinbase news which was this morning?

Obviously very unfortunate uh potentially leak of you know a bunch of uh personally identifiable information, home address addresses, balances, things like that. Is that the kind of thing that in the future could be avoided through utilizing your network or is that just sort of a reality of of you know finance today?

So my first reaction was that I think it was a very very strong uh reaction from Brian.

huge huge fan of Brian's of course but I think um very clear message uh very clear next steps uh and these uh the these things they are unfortunate uh but I think this is like a was somewhat at least from the outside of course I don't know the specifics but it seems like a relatively sophisticated social engineering attack um I don't think what we do will be able to help with something like with customer agent social engineering like I think that's a very specific problem um but I but I mean more on the KYC side.

Is there a world 10 years from now where where somebody is effectively able to KYC using the world network? Oh yeah, I think I think that will almost certainly happen.

Uh we Yeah, because the issue here was like, you know, somebody has their government ID tied to a home address tied to a balance and then that creates a risk vector. Uh which is why Coinbase is taking it so seriously. What about um kind of centaur applications of artificial intelligence?

So, a situation where someone's gaming and they're verifying with the orb or their eyeballs, but um they're still using AI.

I mean, we've already seen this with a real person who's on a dating app, but they're have a second app where they're taking the text, putting it in there, like the Riz GPT, uh and the the responses are still AI generated, but there's just one person involved.

Is that less of an issue this centaur model because there's still a human in the loop versus there's one person controlling or puppeteering like a million accounts?

Yeah, I think that I I think by by in a couple years I think almost everything we do on the internet we will do with help of an AI because it just make our writing better will make our thinking clearer and so I think that is actually not really a problem.

I think um in some sense the way we actually think about it is we think a lot about agents and agent delegation these days. Mhm.

And uh for example, I think one thing how this will be very useful is you will need to understand that behind an agent uh or an agent actually acts on behalf of a human um like a well one human and as you described I think the the the the important uh property of such a system is that you cannot deploy uh thousands or tens of thousands uh of these AIs.

um not to not allow you to interact with that. And then maybe actually the gaming use case, there's a cool example here where with Razer, they they build these gaming uh um webcams. Yeah. And um what so when you actually verify for for world, you receive a a package of uh data that is not stored anywhere centrally.

It's stored on your phone. And one of the things it has actually a signed face image. And so what Razer will be able to do, it actually will use that to ensure that you this is you're not a deep fake. You're actually the person you claim to be. And so you can authenticate a live video stream.

And I think the same will happen on video calls. Um speaking of hardware, uh we were just talking about Apple uh getting into brain computer interfaces. Neurolink is also in the game.

Uh, is there a situation where authentication via BCI is as effective, more effective, something that you're partnered with or developing, or is that just on a different timeline from what you see the roll out of super intelligence or AGI uh being? I well so personally I'm very interested in BCIs.

Um, but I I think this is very much on a different timeline.

Uh so I think like to yeah deployment of invasive uh or even non-invasive PPCIs I think it's like still certainly five plus years away probably 10 15 sure um so I don't we will figure it out once it gets there I would say yeah yeah um what about um I'm I'm super interested in understanding how the decisions that you're making allow us to read into your AGI timeline Sam's AGI timeline lines and what the future will look like.

There's this interesting world where the the kind of the revealing feels like you see a future where there are tons of digital AIs in the world, but maybe there's not this super intelligence that can reconstitute matter at a atomic level instantaneously.

the gray goo scenario because in that scenario, even in the good ending, you would imagine that a super intelligence would be able to create a human eyeball, grow it in a lab, do something to create a a a fake version of the human head, right? Maybe scan me one way or another and then recreate the eyeball.

And so, uh, much like you could potentially fake a fingerprint reader, uh, with enough time and effort and detail in how you're constructing that matter, you would think that you'd be able to do that if you had unlimited resources and super intelligence and all these different things.

But that doesn't seem to be the timeline that you're operating against. So, can you tell me about uh like how your far future vision for super intelligence plays out a little bit? Um, so, well, I I think about these things quite a bit.

I think um so this this kind of also like Ray Coursewhile vision of like eventually we'll have AI that creates nanobots that will be able to reassemble matter etc.

I think that's um that is very likely to happen but I think as a society I think we'll like there there's going to be a lot of other things that will happen before we get there and um so I guess my internal mental model is that we are I think still at least 10 years away from from from that kind of uh level of intelligence or at least from us allowing that level of intelligence to interact with the real world.

Mhm. Um and then actually if you well if you if you then think about the fundamentals of kind of physics in such a scenario is uh that of course what we what we do to to measure that humanness is we use we fundamentally use photons of different wavelengths. So near infrared uh visible light etc.

And so I think like before you would actually create an artificial human such an AI would be able to maybe just emit um multisspectral wavelengths uh to fool such a system and so like well but I think then we will also use such a AI to build a countermeasure and build a more powerful system.

So I think that is in my mind still far out in the future. I I I think like all the medical advancements and um scientific advancements from AGI. I think is that generally your your mental model for how all of this progresses is just like good guy with AGI beats bad guy with AGI.

It's kind of an economic or energy warfare. Whoever has the more compute the more and as long as the the the good people have more resources they will be able to build stronger defenses. And so the net outcome is good.

Um maybe that's like one way to frame it, but like generally I'm uh I think I'm way more optimistic than than kind of most people or like many people are in in the tech world.

I think like the default outcome is that um these AI systems would just be incredible tools for us and I think they will be widely accessible and I think the the frontier labs will make these accessible to kind of a wide population for quite a long time.

Um like at some point you might get kind of this true super intelligence territory where maybe the government steps in steps in I could definitely see that happen. Yeah. Um and then it becomes like a geopolitical uh situation um which is going to become like very hard to predict.

Um but I think until then I think these AI systems will be relatively widely available and open source will not lag too much behind.

Um, so yeah, it seems like most of your model is like general optimism about how AGI and AI progresses, but there are pitfalls that need to be avoided and there's uh problems that will need to be solved in the interim and and worldcoin is one instantiation of that one problem to be solved.

Are there other problems that you're thinking about that maybe you're not working but other founders or other entrepreneurs or other companies might be thinking about as we progress into the AGI future? Well, so first of all maybe the framing about world I think you can definitely do that.

Uh but I think the other framing is as a result of the technology I think we will create one of or if not the largest real human network on the internet which I think will create a level of value. So like it like it's not just preventing the bad, it's also creating the good that I think uh really matters here.

Um you know other problems for other there's like there's like so much it's like what what what uh you know I feel like the main you know there's this um the world is undoubtedly controversial, right? You have people that are super excited about the potential of the technology and understand the importance.

And then there's maybe people that are excited and understand the importance, but they're still wary. Yep. Uh and then there's like the whole other side of people that are just like, "No, you're never going to scan. You know, you're never going to get my eyeball. Don't even, you know, keep I don't even have pre-check.

" Yeah. Yeah. Yeah. Those those types. Uh uh but uh but but what do you say to the people kind of in the middle that are kind of like um interested but kind of wary or maybe on the fence, you know, when you're talking to them one-on-one?

Um you know, I I I think like we have very very high conviction around both the importance of this technology and how the technology is built.

um like I think it has all the properties that we will need as a society for such a technology in in that sense that it's anonymous, it is scalable, it can be inclusive like all of those things I think will end up mattering a lot like maybe much more than is kind of visceral uh at that point in time but I think in the next 12 24 months I think it will become relatively obvious and so I think um it's going to be interesting where like I think different parts of the world will will end up with different trade-offs like maybe some parts of the world will just require government IDs on pretty much every part of the internet and I think that might end up being a a really bad a bad situation.

Um and so and that's bad because somebody would just effectively sell their ID to be rented out for like ultimately uh it will be really hard to build that as a notraceable system uh by by the government and and and and I think like it's of course something we really believe in here in the US is like the freedom. Yeah.

So the government would track you on every single place you go on the internet, every web page. They could in theory they could. I'm not saying they would. And I think it really depends on which part of the world you're in. Sure. Uh but I think that's just not a great outcome.

And so I think like if we do like if we do everything in the right way over the next couple years, I I hope that our brand will become something like the signal messenger. We're like I think totally um this is like the the anonymous high trust uh solution that you can use and other people will have other solutions.

I think there will be a lot of competition around that space because I think it will become a relatively big one. Um so yeah if if others will want to use other options that is that is totally great. I think we will win by building the best product and reaching the most scale. Jordy great place to end.

Yeah this is fantastic. Thank you so much. Uh very excited to see the roll out. Uh I actually did I'm just remembering I had one last question.

Uh are you far as far enough along as you would like to be or do you wish you could just sort of like pause time and you know scale the world coin network you know just for another year to to sort of catch up because I because again when when you talk about these timelines around 12 months 24 months you guys are really uh racing against the clock uh from what it sounds like for a world in which we have these sort of agentic reasoning systems that will really make the internet a much more wild uh place.

Well, I think that that is definitely one perspective in my mind is like I think we're definitely at in in a race against time.

And what what is also what is interesting what really only happened in the last six months is that uh CEOs of many of the very large companies that you know um with large products um well either we we talked to them or they reached out and uh the problem is like currently we cannot serve that scale.

So there's like there's like a lot of pull um for the product to be available uh and we we just really need to get there. we need to roll this out faster and be able to serve these big platforms.

Um, so that I think that is like a extreme level of urgency for one, but then the other side is it's hard to describe how crazy it feels to be even here. Uh, from the beginning of the company like for like the first two to three years, uh, well, people basically just made fun of this. Mhm.

Um and and so like yeah, it's certainly personally the journey of a lifetime and it and and like when when I started working this like I I gave it a very low probability of success and I did so for the first three years or something.

It's like yeah still very very long to go and fast to scale but on the other hand it feels pretty wild to just be here right now. Amazing on the social network thing. Sorry last question. We're like all right thanks Alex. Thanks for coming on.

Uh I mean uh when you say large scale network valuable are we talking Visa network or are we talking Instagram? You're gonna you're gonna see Yeah. Yeah. I mean I I'm I'm interested to see the the business model, the pricing model, are the big platforms going to be paying in World Coin? Are they be holding?

Are we going to have a national reserve of this stuff? I don't know. We'll have to try We'll have to have you back on when there's a big announcement. Looking forward to that. Thanks for having me. Yeah. Fantastic. Talk to you soon. Uh anyway, it was good. We broke because we got to talk about eight sleep.

They got a 5year warranty, 30 night risk-free trial, free returns, free shipping. Pod five ultra. Pod five Ultra. I think you beat me. I got an 88. I think you got a 92, right? I think the 18 must have gone in the app and and uh whatever John gets. Just give Jordy a couple extra points. Four more. I got a 92. 92.

Let's hear for two hours of deep