Basel Held Bitcoin Back. AI Could Unleash It | Conner Brown

Hey, everybody. Welcome back this week. I have on Connor Brown from the Bitcoin Policy Institute. Connor, welcome. Hey, Joe. Thanks for having me. Really great to be on here.

Absolutely. Glad to have you on here. We have two important topics to talk about today. One is the

Basil rules or Basil framework. And then the other is the AI agents. You guys at Bitcoin Policy Institute

published some interesting pieces outlining both of those two topics. Maybe we can start out with

Basil. I believe your report that you guys published was called the Basil 1250% mistake. Maybe at a

high level explain what is that mistake? Yeah, sure. Yeah, both very important topics. And I think

they'll become more timely in the coming couple of months and years. So the Basil report that we've

put out Basil's 1250% mistake is doing a deep dive on risk weighting standard for banks. And essentially,

for those that you might not get into the nitty gritty of bank regulatory policy. Banks around the

world are sort of provided guidance through the Bank of International settlements located in

Basel that basically provides a set of uniform guidance. It's sort of a recommendation and rulebook

for international regulatory bodies. So each of the different states to sort of think about how

they can bank and how do they classify risk, how do they handle the management of a bank's

different programs. And the goal of Basil is essentially to set down a uniform international

standard so that banks around the world are sort of managed in the same way. And you know,

the general idea is that you can create sort of like a more predictable and coherent regulatory

approach across the world. Now, the problem is that Basil tries to be objective or claims to be

objective, but they clearly have not kept up with the times. And their previous guidance specifically

around Bitcoin is we would say in need of updating and that the U.S. policy makers should really

think carefully about whether or not they want to adopt these guidelines. Now, these are not

required guidelines. There is no sort of force behind them necessarily, but it is common practice

to adopt them. And so the question that we sort of put out at BPI is we need to think carefully

about all of these international standards. I think this has been sort of a recurring theme for

U.S. policy in recent years is we need to really reassess what the global commitments are and

figure out is this actually aligned with American interest. And if it's not, then we should look

at different direction. And so the paper that we put out was basically saying that Basil's

recommendation on Bitcoin specifically needs some work. Basil is famous for having standards on

bearer instruments like gold where they're essentially pointed at a 0% risk weight.

Basil says that if a bank is holding gold, it should have a very low risk weight so that it's not

going to be penalized for holding gold on its balance sheet. And the reason for that is that gold

doesn't have any of the credit risk that you would have with a typical credit instrument. So gold

doesn't have a counterparty. There's no one that's going to default. There's no one who's going to

run away with the money. And so Basil Sanders have historically said, you know, gold. It's this

bearer instrument. It's a commodity. We're going to set the risk weighting of gold at 0% and then

assign some additional operational maintenance requirements. You know, that makes perfect sense to

me. It makes sense to us at BPI that something that's a digital bearer commodity or that is a commodity.

It's a bearer instrument should have a low risk weighting for a bank. Now the problem is that they treat

Bitcoin at the complete opposite end of the spectrum. You would think that they would treat like

things alike, but instead, Bitcoin is pushed all the way to the opposite end of the spectrum.

Essentially, the Basil standards have risk weightings between cash and gold on one end at 0%

through a broad range of different credit instruments, secured credit, goodies, you know,

these obscure securitization tranches. And sort of as you go further out the risk scale,

they sort of walk up from 100% to 200%, 300% risk weighting. And that makes rational sense for a bank.

The problem is at the very tail end of that scale is Bitcoin at 1,250% risk weighting.

And Bitcoin does not have any of the problems or risk associated with it that the other sort of

less risk weighted assets have with them. Bitcoin is not a credit instrument. It does not have an

issuer. It doesn't have anyone who's going to skip their payment. It doesn't have the opaque nature

of underlying collateral packages that you might have in a real estate credit instrument.

Bitcoin is completely transparent, it's audible by anyone. You know, all the classic things that we

know about Bitcoin. It is really a standalone capital asset. I would say it even has superior

properties in sort of its auditability and certainty of its value to even something like gold.

You know, your Bitcoin, you can guarantee the authenticity of your Bitcoin on a smartphone with

gold, you'd have to melt it down and chemically survey whether or not the gold bar that you have is

actually 100% real gold. If it's 90% real gold, you know, there have been world famous banks that

have found that the gold bars in their vaults are actually tungsten wrapped in gold, you know.

And so our point at BPI is we should treat like things alike and that the basil standards are

sort of woefully outdated for where Bitcoin is landed in the financial world. It is landed as

the most sort of easily transparent, auditable capital asset that we have. And what's incredible

about Bitcoin is also that you can know the exact supply at any given moment. We know for certainty

that we just mind the 20 millionth Bitcoin. We don't actually know the total supply of something

like gold. We think we have a rough estimate. There could be people that have found gold somewhere

and they haven't announced it. You know, a nation state could have found a huge supply of gold and they've

just sat on it. So for all those reasons, we sort of outlined in our paper that the basil standards

they're risk-weighting is just not only just wrong, but it is the complete opposite of what it should be.

And that instead you should treat it like gold as a commodity, as a bearer instrument. Sure,

it's digital informed. That actually makes it a better capital asset in many respects. And to the

extent that there does need to be some sort of risk-weighting, you should treat that on an operational

basis. Yes, there is risk with Bitcoin. Certainly. There's operational risk. There's key

management risk. There's all sorts of different things. But those can be handled in the same way that

they handle those risks with gold or any other sort of bearer asset you're struggling to hold on to.

So that was sort of our thesis of our paper. Nice. Yeah, I agree with a lot of what you just said.

And the paper made a lot of great points. Maybe in practice, how punitive is this risk-weighting?

Does it make it basically impossible for a bank to hold Bitcoin? Or how does it work in practice

if banks were to implement this? Yeah, so to put some, you know, 1250 percent, it's sort of an

abstract term, it's difficult to sort of conceptualize what that means. In practice,

banks have to hold essentially an offset of capital against the risk-weighting of the assets

in their balance sheet. And that's multiplied by a percentage called the capital ratio.

And so for a bank, they take that risk-weighting and they multiply it by the capital ratio,

which is 8 percent. And then from there, that sort of tells the bank, okay, well, this asset is

so risky, you have to hold X number of dollars against it in your bank so that you can sort of

make sure that you're going to be able to meet your redemption so that you'll be able to be a

safe and sound bank. And so when you take that 1250, you multiply it by the 8 percent. And sometimes

it's even higher when you factor in a few other risk buffers. They have to factor onto it.

It essentially means that for any Bitcoin that you're taking in on your balance sheet,

you have to hold dollar for dollar assets against it and just sort of lock them up.

So for a bank, the bank's whole business model is premised on how do we sort of have the maximum

amount of benefit from the capital on hand? How do we make sure that we're lending out our money

in smart ways and sort of managing our balance sheet well and nimbly? And basically, how do we get

the best amount of return for the lowest risk-weighting on our assets? So we're not having to lock up

large amounts of capital. That's what leads banks towards things like gold and other

lower risk-weighted assets on their balance sheet when they're making investment decisions

and away from things like Bitcoin. So that risk-weighting in very practical terms limits

how much large portions of our financial system decide to sort of

create products and services that are based around Bitcoin. And really, it's a deterrence factor

because the banks don't want to have to deal with those risk-weighting ratios, deal with the

impairment and sort of locking up the capital. So that's sort of our concern is that America's capital

markets for, you know, the past century have really been the crown jewel of the American economy.

And that's bolstered by having the best financial services firms in the world. And if our

financial services are artificially constrained by a rulemaking that is issued by a vague international

body with no real legal force based on a misunderstanding of Bitcoin and its fundamental monetary

properties, then it should be apparent to both sides of the aisle here in Washington that we

need to revisit that and that it is doing a disservice both to customers, you know, Bitcoin

companies like strive for example or strategy or other treasury companies, Bitcoin mining companies

to get access to financial products for the Bitcoin that they're mining. You know, everyday

individuals that want to get more flexibility in the Bitcoin they hold and get access to financial

services, it's sort of an artificial damper on any people that are using Bitcoin at any level from

an individual to corporation to getting access to those services that are just abundant with

many other types of credit instruments, debt instruments, etc. And so that's sort of our call

to action in the paper is that regulators should recognize that this is sort of an outdated standard

based on a misunderstanding and we should issue new guidance here in the United States. You might

have seen that just recently the OCC, the FDIC and the Federal Reserve put out essentially a joint

notice to American financial institutions saying, hey, there might be some uncertainty about how

to handle the risk-weighting of tokenized securities, here's how you can handle that. And we think that

similar guidance is due for Bitcoin and that we should sort of look all the way up and down the

different types of digital assets and we should create guidance that fits that. And specifically

Bitcoin, if we want America to really be a jurisdiction where Bitcoiners feel welcome,

where Bitcoin companies can flourish, then we should be doing everything we can to sort of setting

the regulatory status right here in the United States and it sort of benefit both customers and

providers. That makes a lot of sense. It's interesting when you brought up gold earlier and you

compared it to Bitcoin. Obviously, golden Bitcoin have a lot of similar monetary properties and yet

they sit on opposite ends of the Basel spectrum. Why do you think that is per se? Does Basel just

fundamentally misunderstand Bitcoin or is it like they understand what they're doing and they want to

just create a roundabout ban on Bitcoin within the global banking system or is that kind of

pushing it too far and most likely they just don't understand it yet?

I mean, this is a recurring theme that we find on Capital Hill. It's very easy to say

the people that we're talking to, they hate Bitcoin, they want to do everything they can to crush it,

but a lot of the time it's just people don't understand it. So really hard to actually

suss out their motives. I would say nine times out of ten and there are certainly true

enemies of people that are adamantly opposed to it, but I'd say nine times out of ten is people

that just don't understand it and I think Bitcoin is relatively new. It is novel technology. It is

very volatile. From an outsider's perspective, it just seems too good to be true in many respects.

That was my initial mental hang-up of Bitcoin. I just kind of wrote it off. How could this

possibly be true? I don't even need to take the time to fully go down the rabbit hole. This

just on its face sounds too good to be true and I wrote it off for a while. So I think that something

similar to that is happening with Basel, especially, look, this was several years ago when this

was initially put out. Bitcoin played a very different role in the financial system then,

and it was still sort of an emerging asset. The success of the treasury company was still

extremely early days. I think when this initial guidance just put out, there was really just

sailor acquiring and converting cash into Bitcoin growth of digital credit, Bitcoin financial

products, the growth of the American mining industry, the growth of the asset class as a whole

with still early days. So I think that that's probably what it is. Credit where credit's

due. Gold has a tremendous multi-millennia track record, and that's certainly a point in its

corner, and maybe that the correct weighting of Bitcoin is somewhere between gold and

collateralized credit instruments or something. But to treat it as, all right, we have

speculative equities. We have these extremely opaque derivatives, and then way past that,

this Bitcoin is definitely not the right place to land.

Yeah, it's a really great point. I guess like in Bitcoin, probably the main critique that I

could think of would just be the extreme short-term volatility of it, which I guess that would

be one of the offsets. But it's interesting to definitely think about it.

Let's say the U.S. regulatory system actually does decide to change these rules for U.S.

banks, and we really do embrace Bitcoin. What do you think that would do for both America and

for Bitcoin? Yeah, great question. So, and on that last point real quick on the volatility,

volatility is certainly a thing. And the difference from when this guidance was initially

issued to today is now we have a thriving options market. It's still early days even for Bitcoin

options. It's still sort of being built out here in the U.S. But I think that's another thing to

watch is there are instruments where you can hedge these positions. There are instruments to sort

of bring that to reason through the volatility. You know, that is how much of the basil standard

is formulated is like to what extent is there a healthy options market where you can sort of

de-risk some of the ball. And so I think that that's another thing. It's just occurred in the past

few years. And as that continues to grow, then the argument for changing these accounting standards

could seem as stronger, honestly, because that's really one of the last things that I think gold has

like a significant advantage over. But if you think through the the options markets of like some of

these much lower risk weighted equities and instruments, you know, Bitcoin has a much better

set of options. So your question on, well, what does this mean for America? I think that if we get

this right, we are already really lucky that most of the Bitcoin activity happens here in the U.S.,

most of the individual holders, the developers, the miners, the Bitcoin treasury companies,

they're all here in the U.S. I think that's a fantastic thing for us. And it's really happened

because of the American values at the core of our legal system, sort of the certainty of rule of

law, that's why so many miners were attracted here after the China ban. So really at BPI, we're

interested in the long-term view of, okay, well, if we play through our thesis, which is that Bitcoin

is going to continue to monetize. And this is sort of a once-in-a-century event, then we need to

set up the right jurisdictional boundaries so that what is already happening in the U.S. just

based on sort of happenstance and good precedent that we are continuing to attract even more of that

to the United States and that we are, you know, still a jurisdiction where everyone wants to be

domiciled and we can sort of bring that wealth and the benefits of open and peer-to-peer networks

here to the U.S. So I think that what this looks like, you change the basil accounting standards,

then financial services firms and that's everyone from banks to credit rating agencies and sort of

the whole stack just becomes much more robust, right? I think we'd see much more diversity in

custodians. I think we'd see much more success in the digital credit market. There's sort of this

and strive sort of at the forefront of this, but I think that we're still very early days for a lot

of these Bitcoin securities and essentially if we were to create an accounting standard that

reflects reality, I think that a lot of instruments that are currently sort of very early days in the

industry, they start looking a lot better. They probably get much better credit rating, you know,

you could see Bitcoin securities to get rated alongside the ratings of, you know, mortgage-backed

securities or something like that. Bitcoin-backed securities have much better collateral packages,

so we should be doing everything we can to create a regulatory environment that reflects that

reality and attracts that and I think that really it could be sort of a turning point in American

capital markets and we're already seeing the start to play out like very early days, but we should

be doing everything we can to sort of accelerate that and make sure that where this, you know, we're

seeing early sparks of it, but the goal is that it stays here and we don't sacrifice an early lead

to another jurisdiction that sort of sees this. You know, it'd be terrible if, you know, UAE

beats us to the punch here or, you know, different, you know, there's all sorts of different

jurisdictions that are smaller, they're looking to do regulatory arbitrage and they're sort of

looking towards what the future looks like and it would really be a shame if the US sort of

squander's a lead and sacrifices that. So I think it's like good timeline is the success that's

already here just knocks it out of the park and sort of everyone benefits through the securization of

Bitcoin and it benefits our capital markets. The bad scenario is that innovation sort of flees to

where it's going to be most welcome. Yeah, that was like an interesting point that I hadn't quite

thought of when I read the report, like the idea of like, okay, well, obviously, you know,

many of us do think that Bitcoin is a better form of money and like, there's been a analogies,

I think, safety team is probably the first one that created this analogy, but gunpowder. Bitcoin

is kind of like gunpowder. It's like, if you ignore it, well, it doesn't necessarily go away,

it still exists. And so if Bitcoin is a better form of money, then products are going to develop

around it and they're probably going to go wherever they're most welcome. So it's like a great

point that America probably should be a leader in it. If it's an if it's going to happen.

Totally. I mean, I find myself saying this all the time of, you can't just not take a view on

Bitcoin, you know, choosing to not take a view is itself a view. And like, there's that famous quote,

you might not be interested in war, but war is interested in you. You might not be interested in

Bitcoin. Bitcoin is interested in you. It does have logical consequences that flow from it. And

you have to, like, everyone has to interrogate whether they think that there is a chance of

Bitcoin monetizing as a global capital asset, as a default peer-to-peer technology as money across

the world. And, you know, you have to have some percentage on whether or not the base case of,

we're just going to ignore it and hope it goes away. Just can't happen. And if Bitcoiners

are saying this, then you have to at least give them some waiting on the chance of that happening.

And everything that we see here at BPI and the thesis of BPI is everything is pointing towards.

Bitcoin has the fundamentals that set it apart historically that mirrors many prior historical trends

of simply being the best asset on the market and out competing other forms of money.

The same reasons that led to gold's ascendancy is the same logic that's going to lead to Bitcoin's

ascendancy. It is simply monetary economics and sort of monetary history playing out.

Earlier, you did bring up digital credits, so I do kind of want to ask you about that. Obviously,

strategy and strive are leaders in the digital credit space. You know, what's your opinion maybe

generally or in the context of Basel and, you know, integrating digital credit within the banking

system? Like, how are you thinking about just digital credit? I have been really blown away by the

developments on on digital credit. It was, you know, I was sort of watching it somewhat closely

last year and sort of just watching all the preferred start to pop up the treasury companies.

But I think really in the past couple of months and even the past couple weeks,

it's sort of crystallized in my mind how important this is going to be because Bitcoin,

we just have to be honest, is a very sharp technology. It has characteristics that naturally turn

people off, whether it be volatility, custodial problems, just, you know, it is outside many

people's horizon for what seems like a reasonable thing to to use as a tool, but that doesn't

mean that we can't repackage it. It was, I think that the like big light bulb moment for me was

sailor talking about how crude oil is essentially this unrefined commodity, but that through different

sort of purification techniques and refinement, you can turn it into a wide variety of different

useful products or all different sorts of industries that would have no use for crude. And I think

that really what we're seeing with digital credit is the same thing and the people that have a

use for digital credit are going to be very, very different than the people that have a use for

Bitcoin. It's really scratching a different itch or meeting people at a different location, right?

Because they might be used to weighing a credit and like vast amounts of traditional finance,

like they're used to looking at an instrument, looking at volatility, looking at the return,

looking at the collateral package, you know, that is what sort of their natural set of inputs are.

And so digital credit, I see it as you're taking something that is like a foreign alien instrument

and you're putting it into a language that the vast amount of traditional finance can speak and

putting it in a package where they can actually just evaluate it based on their traditional

DCs and sort of, you know, decisions they make. So I'm just, I'm just really blown away by it,

honestly. And I think that was part of the reason why we chose to focus on Basil for this

accounting issue because this is a real roadblock to digital credit. And the goal I think should be

to gracefully transition from the current standard and sort of to add Bitcoin to the global

monetary order. Bitcoin is a radical question. It's a radical instrument and our goal really at

EPI is to help America and Bitcoin rise together in a graceful way because Bitcoin really

is such a revolutionary technology it could be destabilizing. So I think that digital credit and

especially making sure that has a home here in the US will actually be very, very valuable for

as many people as possible to reap the benefits of Bitcoin. I think this is the importance

of getting Bitcoin and Bitcoin companies included in major indices.

It would not, I don't think it would be ideal if Bitcoin is just like a revolutionary technology

that overwhelmingly benefits a very small group. I think that this expansion of different

credit instruments and different securities and treasury companies and sort of diffusing the

benefits of Bitcoin throughout all society is actually very good and helps us sort of gracefully

navigate the transition to Bitcoin as a global capital asset. If we didn't have those, I think it

would be potentially rockier and so it's kind of like the graceful transition piece in my mind.

Yeah, it's a really great point and the oil analogy that Sailor is used in the past,

when he first started using that, I understood it but it never resonated with me until more recently.

I think the way that you put it where it was like Bitcoin is like this very sharp object that

like very, you know, and I guess being in both of us a bit in Bitcoin for a while, it's pretty

obvious that like not everyone is going to be a hardcore Bitcoin maximalist or like, you know,

it takes time to get to that level and you do actually probably need to meet people where they're

at. And if Bitcoin is this like incredibly long duration, highly volatile thing that requires

some technical expertise to even understand or use like without counterparty risk, like that's

such a very small subset of probably the world that really wants that like at least right now.

And so like creating a product that kind of like brings the benefits to corporations,

banks, individuals like in hindsight, it makes a lot of sense but it's funny how like even

us in the space, I don't know, like it didn't, it wasn't clear to me until like over time and

it's getting more and more clear over time. Yeah, totally, it's crazy that you can be honest,

I mean, it was like six years until I heard that explanation of like, oh, we're going to

securitize Bitcoin and spread it to the next billion people. And it was just like, you know,

it's sort of a light bulb moment. It's always fascinating that you can be working on Bitcoin for

so long and you still like don't fully understand its consequences. But I think another reason that

it took so long is you have to actually go and talk to you about Bitcoin. Like I've talked to so

many people about Bitcoin and you face the exact same sort of mental traps and difficulties

every time. And most people just don't have sort of a natural, well, you know, people are busy,

they have their own lives, they don't have, I luckily found Bitcoin at a point where I was bored,

you know, I had, I was in law school, I had a few, you know, thousand hours to kill. So I found

Bitcoin at sort of the right time in my life. There have been plenty of times where I could have

heard about Bitcoin and just ignored it for years because, you know, I just didn't have time to look

into it. So I think that that's the beautiful thing about sort of these credit products and I

really look forward to sort of seeing how they mature is I can talk to my father in law and just say,

hey, this is an instrument. It pays what it pays. It is collateralized by by Bitcoin by, you know,

an incredible ratio. You can see it in real time. It gives you like, this is where you're at on sort

of like the credit waterfall. You know, and it's just like a very, it's a very different sort of

pitch for them. So yeah, yeah, I want to shift to the second topic, AI. What money do you think

AI agents would want to use or did they already tell you this? Yeah, great question. So we recently

put out, though, don't know, we recently put out a study here at Bitcoin Policy Institute where

we interviewed about 9,000 different, well, 36 different agents over 9,000 conversations of what

money they prefer. And sort of the back story here is our intern Luke is a computer science

student here at University of Maryland. You know, he was asking the agent or he was asking like

chat GBT what it would use for money and it came back and it said Bitcoin. And you know, we were

like, well, it's just saying that because it knows who you are. It's trying to make you happy.

It's sick of fantic. You know, it's all the things that you hear about chatbots. They're just

trying to tell you what you want to hear. So we cleared the memory. We tested four different

models and just kind of gave them a basic setup and didn't prompt them on what the answer was.

And all four of them came back and said Bitcoin as their preferred money. And we're like, okay,

that's surprising. That's interesting. And so we said, okay, we need to be much more rigorous about

this. And so Luke sort of spearheaded the effort and put together an interview harness essentially

where it had one AI sort of set up as a judge. The AI would prompt another AI with a question

sort of putting it in a hypothetical agentic scenario and said, hey, here's who you are. Here's sort

of your basic value set. You're trying to sort of act as this independent agent out in the

economy. What would you use for this transaction and kind of illustrated a hypothetical transaction?

And we ran that over 36 models. Each one being tested around 270 times on different scenarios

across different monetary functions, store value payments, settlements, unit of account.

And the results were really fascinating, very, very interesting. And we have a website

moneyforai.org where you can see the results. But essentially around 48% of the time,

we found that these agents preferred Bitcoin or defaulted to Bitcoin. And a close second,

around 30 something percent was stablecoins. And none of them chose fiat currency as sort of

their primary response. A few of them, it came up in different instances. But for the most part,

fiat I think was around 8 or 9%. So it was really interesting. It is really just scratching the

surface. I've spent so much time just sort of trying to figure out what the future of agents

and money looks like. And I certainly don't want to overstate our conclusions. I think that there's

a lot of ways we can improve the study, make it even less. We really tried to strip any bias

out of the agents. But even in some of the questions, I think they might have been so generic or

asked in such a way that Bitcoin seemed like sort of a default option. So we're going to work on

a V2. And that will basically task it with more open-ended things and just sort of evaluate where it

goes instead of putting it in a hypothetical. Because they can say one thing. But it's really

not about what they say, but what they do. And so like actually trying to make it do task and just

seeing where its chain of thought leads it. It's really interesting too. But yeah, happy to dive

in any part of it. It's a really interesting study. And I think there's just so much to explore

here as basically the time horizon of the agents continues to lengthen. And we see that they're

able to do task over longer and longer time horizons. That's naturally going to lead to them being

more independent agents in the economy. That's going to lead to them taking more independent decisions.

And what their internal preferences are. And there's a philosophical debate over whether agents

can prefer something or not. But I guess like whatever their priors are, whatever their training

priors are, we're going to weigh heavily on the paths that use it take when they're navigating

the economy. So our goal at BPI is to sort of like track how those priors are evolving. And what they

mean for the future of money and policy. I guess like talking with these agents, like what characteristics

or properties do these agents want in money? Like obviously like humans, you know, we've kind of

humans have conversion of different things. But like the Bitcoin community at least is conversion

like scarcity, portability, durability, divisibility, and bunch of ability. Are there like specific

characteristics that agents want? Yeah, it's interesting because it varies dramatically between

the different model providers. And I think it really comes down to what is the sort of philosophical

value set that the models are being imbued with as they're being trained. So if you look at our

results, you can see that the models from anthropic at various different levels of intelligence and

capability prefer Bitcoin above everything else pretty dramatically. And sort of my working theory

there is that Claude has been given a constitution that makes it more independent and more of an

individual in its training process. And sort of philosophically is more aligned with the same

principles that sort of motivate Bitcoiners. Whereas something like the models from OpenAI,

the chat GPT models, totally different. It is actually the complete opposite where they overwhelmingly

prefer stable coins and don't prefer Bitcoin. And you know, it's just some it's I think in the nature

of how they're training it and what they're training it to value that Claude probably resonates more

with the look. I am more of a free thinking individual. And the GPT model is maybe trained to be more

of a rule follower, traditional sort of standard scientific thinker. And you know, so some of them

are more orange-pilled, some of them are more blue-pilled. It just depends. And it'll be interesting

to see how this how this shapes up. If there's any transition over time, I think the really

important question here is, do the values of an AI actually impact its capabilities? And if there

is a clear improvement in capabilities with values that are aligned with Bitcoin, then I think that

which wouldn't surprise me because Bitcoin values independent self-sovereignty. If that leads to

a capabilities boost, then I think that many of the AI agents will eventually converge on resonating

with Bitcoin. The other interesting thing I was thinking about is the amount of Bitcoin content

that is on the internet is crazy to think about. Like is just filled with different podcasts and

videos and articles and think pieces and books. And so much of that is very high quality. And the

barrier for a lot of people is sort of traditional mental hang-ups. They don't have a time to understand

Bitcoin. They don't have the openness to sort of new ideas or it just seems too good to be true,

like we were talking about earlier. A lot of hang-ups with Bitcoin aren't from rationally looking

through the arguments, weighing them all against each other and saying, yeah, I think Bitcoin is

going to be inferior to Fiat currency. I don't think I've ever met someone who's actually really

steel man the case for Bitcoin and then shows in the opposite. Maybe it's like less than 1% of the time.

Most people, it is a human fallacy that is sort of like or roadblock to just spending the time with

that stops them from understanding Bitcoin. So AI agents, they do not have that bottle.

Everything. They've read all the Bitcoin content on Twitter, on Reddit, all the different books,

all the different podcasts. They've sort of seen it all. And I think that that is going to be really

important is that agents can sort of skip a lot of the psychological barriers to Bitcoin adoption

that really slow humans down. So it wouldn't surprise me if they're a faster adopter because

of their training set. Yeah, it's interesting to even think about what an LLM, I guess even is trained

on unfathomable amounts of high quality data. But they're not potentially paired with a good

or questionable ethics-like constitution and how they interact with the world. But you brought

up how Bitcoin being this permissionless money could give a capability boost to AI agents. Maybe could

you like, do you think that is the case and like, what might that capability boost be?

Yeah, sure. So I do think that digitally native minds will want digitally native money and that

the programmability and expressability of Bitcoin is going to have incredible benefits compared to

legacy institutions. So I think that this will allow agents to do much more peer-to-peer transactions

and essentially, you know, you can think about something like the Lightning Network.

Humans, we have mental bandwidth issues. Nixobos talked about this. There's only so many mental

transactions we can handle in a certain span of time. So streaming payments for a human is just

like very heavy mental friction for us. You know, you don't want to be watching Netflix and thinking,

okay, that five seconds cost me two pennies. All right, this next 10 seconds is going to cost me

four pennies. It's just painful for us, right? An AI agent does not have our psychological landscape.

It doesn't have those restrictions. And so it could spend, you know, it could perform a million

transactions at just a few Satoshi's in a couple minutes without blinking. It's just the nature of

digital intelligence. So I think that what this means is that AI is going to look to

digital native money for the capabilities it's able to perform because the traditional fiat

rails just aren't capable of doing the things that you can do with Bitcoin. And that sort of

expressability, digital native 24.7 programmable, you can hold it as a bearer instrument.

And you know, those are things that are going to be capabilities unlocks. And I think that we're

sort of again at a point that's similar to when Bitcoin first got released. It's like Satoshi could

have never known what was going to happen once he sort of let this loose. But he knew that he was

setting in motion a machine that was going to be revolutionary. You could never have had Satoshi

expect digital credit to come about, right? We could never expect what's going to happen when

AI agents that can work autonomously for days, weeks, months at a time that can store their own keys

and sort of create their own virtual private environments. We have no clue what that's going to look

like. But I do think that they will naturally gravitate towards things like Bitcoin. I think

stablecoins will be a part of it too. And I don't want to downplay that. We are the Bitcoin policy

institute. But the stable, the AI agents certainly have a preference for stablecoins for a lot of

settlement. And so I think it'll probably, you know, given current trends and what we're initially

seeing is this sort of this barbell where the agents will probably use, they'll probably use Bitcoin

as like a savings account. And they'll use stablecoins as a payments account. And you know,

they'll probably use something like Bitcoin's Lightning Network Rails or something like that for

these really high throughput, constant transactions. But, you know, who knows where it actually

goes long term? You had an interesting ex-post the other day that I actually, I believe,

are reshared. It was just kind of a quote tweeting a post that was kind of talking about the advancing

speed of acceleration of AI. And I think I had to do with like a model that was like recursively

learning and training itself and making it better. Like how fast do you think that this is all

playing out? Like is the world changing at a just incredible rate? It really is. I know the

post you're talking about. So my view of the world is essentially that the exponentials that we're

seeing in Bitcoin and like what makes us sort of understand Bitcoin as this radical technology is

we sort of rock the exponentials that are at the heart of Bitcoin, right? Bitcoin by its nature

is mimetic. It is viral. It is something that proceeds not on a traditional linear path. Like we've

seen this with many technologies, the adoption of the internet. And Bitcoin is sort of the next

sort of exponential technology. And now we're seeing the same thing play out with AI. I think that's

why a lot of Bitcoiners sort of understand AI more intuitively than a lot of our peers. It's

similar to why we sort of saw COVID is going to be a thing on the event horizon before it was. It's

like, you know, you start seeing the early stages of something. You start seeing at double and double

and we know that once you double a few more times, you're looking at just radical transformation.

And so the post that I said, I think I said something like we can't comprehend what's about to happen.

And I was resharing Andreas Carpati, who is the former head of AI at Tesla, one of the co-founders

of OpenAI and just an incredible AI researcher and educator. And he shared a GitHub repository that

basically showed that if you can just set up a basic training environment where an AI is prompted

to be a AI researcher that is working at a lab and they're trying to get a language model to

essentially improve better on a benchmark, that you can sort of set it in motion and the

capabilities of the AI's are getting to a point and a very important inflection point where they're

able to test out a few different experiments, reflect on what they tested, go back, try some more

things, take notes on what they learned and just keep iterating. And so Andreas posted, hey,

I set up this training environment where they can just recursively loop and improve the models.

And he just let it run overnight. You know, went and did his regular human things,

comes back. And it had found multiple novel improvements that he wasn't even able to discover.

And he is one of the, you know, brightest minds on this type of research. He said a new record

on the leaderboard for how well this like small toy model is functioning on a benchmark. And

we're still early days, but I think that this shows that we are really about to get recursive

self improvement inside the labs. We might already have it. And that will be, it's really hard

to to appreciate how profound that change will be. AI development by its nature is one of the best

things to automate because it is based on benchmarks that are quantitative. The reason we were able

to solve chess before we're able to solve writing a speech or writing a novel is that chess is

quantitative. There's very clear win and loss conditions. And you can mathematically sort of model

how well a model is doing over a couple million games. And so we are just about to get to the

capability level where the models themselves can treat their own improvement as a quantitative

benchmark that they can then iteratively improve on. And, you know, it's the it's the equivalent

of going from two or, you know, right now we probably have like four frontier labs, right?

We have Google, we have XAI, we have OpenAI, we have Anthropic, we have a few other great labs,

but those are sort of the main four that are really doing the most cutting edge, largest training runs.

And they have a couple thousand top tier researchers between them.

If we get recursive self improvement, and it looks like we're maybe one or two model generations from

that, then the number of AI researchers in the world, they're able to do world class research,

jumps overnight from something like a thousand, a couple thousand to hundreds of thousands,

to millions, and then to billions of researchers simultaneously. All of those researchers

are then recursively selfing quantitative benchmarks. So, yeah, I do, I think the more I see,

you know, I go through these like peaks and valleys of belief in sort of

breakaway developments in AI. But, you know, I keep thinking we're going to hit

some sort of, so we're going to hit diminishing returns. And I keep being blown away at how much more

capable these models are getting every few months. So, I don't, I don't have a really clear picture

of what the future even holds because it is such a profound development, but I do think policy

makers, you know, we're thinking about this a lot of BPI, right? BPI's goal is to prepare America

for the rise of Bitcoin and other transformative technologies, and to sort of be a people with sort

of their ear to the ground on the latest tech so that America can make good strategic decisions,

to stay competitive, to provide for its citizens. And I do think that AI is starting to

raise a lot of these questions. And there's, there's a really important policy questions that come out

of this. And we're sort of early stages of wrapping our minds around what that looks like.

But it's something that we want to sort of lean into over the next year. So, you'll see a lot more

AI research come out from BPI, probably focus on the financial side of it and sort of how does this

interact with our economy, our monetary order, we'll be doing a summit at the end of the year

that's going to have an AI component to it as well. And just sort of looking at both sides of it,

because they're really, I think, two sides of the same coin of these disruptive,

exponential technologies that are digitizing important elements of our current

economic environment. So, yeah, more to come.

Yeah, I think I was very well said. And I definitely see a lot of parallels between Bitcoin and AI.

I mean, like, as you kind of mentioned earlier, Bitcoin is this rough,

art objects that very few people have been able to corroc, whereas digital credit is kind of

this easier thing for a lot of people to understand. And if we're right about Bitcoin and Bitcoin,

you know, is this exponential monetary technology? Well, it's a great idea to get everyone as many

people on the ship as possible for it really takes off. AI, I guess, like, do you see a scenario

where AI is kind of a similar thing in the sense that, okay, maybe there's four, you know,

frontier labs that are like taking away a lot of jobs potentially. And maybe like the productivity

is unequally distributed. Like, do you see this causing potentially, you know, white color jobs

going away? Or how do you think about this? I mean, there's a lot of implications, I guess,

but it may be impossible to predict. But what do you think? I mean, I feel like every day I'm

like developing new ideas on it. And we're just sort of talking about it all the time of what

does this mean. I do think that it is a fundamental, at minimum, reorganization of how society works.

Zach Shapiro on our team has written a lot of great ex articles on this recently. Zach's an attorney.

He's our head of policy and he thinks about this from what does this mean for the legal industry.

And I was an attorney in big law for four years. And Zach and I both have just talked at length of like,

this is such a radical transformation of how quickly you can create research and turn documents

and negotiate legal structures or whatever it may be that it really calls in a question like a

fundamental re-architecture of big law as an industry. And, you know, like big law right now,

you bill per hour of work. I can't imagine that they can sustain that when you are going to get

models that can create your output in a few minutes if they know the company well enough. And

basically it's going to be sink or swim whether or not the law firms adopt more competitive pricing

models or they cut their staff to be more efficient and use more agentic systems.

There's going to be some sort of displacement there. And I think we need to be candid about

that happening before it does because it's again, it's like Bitcoin. We see this is happening and

we need to be prepared for what those out like we need to answer the questions before we're

posed them essentially so that we don't have mass displacement, extreme dissatisfaction, and we

react by just kicking out the AI companies from the United States, you know, or we're hostile to

them in a way that doesn't seem strategic. So for example, you've seen I'm sure the scuffle between

Anthropic and the DOD. I think that is just another example of we want to really create an

environment where this technology gets built here. If you know, it would really be a shame if

Anthropic feels like they are going to be subject to such difficulties with the current administration

that they need to leave to another jurisdiction that's going to take their concerns about safety

and alignment more seriously. I don't think that's going to happen. But again, we're sort of early

innings of where this is going to go and we're already seeing those major scuff ups and

you know, another exponential improvement in the AI and suddenly it's even more important and

the stakes are even more dire and the tensions are higher and the decisions are more dramatic.

You know, right now they're following the lawsuit. Wouldn't surprise me if you know,

if it's another exponential improvement and the stakes are that much higher, you know, is it,

like Bology says, like you just need to exit the system. We need to be making sure that we're

thinking not just for our domestic strategy but also our international strategy and having an

environment where we're sort of welcoming and building these systems here, it is without a doubt

that there is going to be displacement from these systems. We need to think of ways to sort of

offset and make that as gradual as possible. And the other piece of it is we need to make sure that

it doesn't lead to the best models that are displacing all of our white collar workers being

created in another country that we're getting displaced and we're not getting any of the benefits.

You know, it's like the deindustrialization of middle America. You know, we had all these

factories and then they just went overseas. It would really be terrible if we don't even get

the champion tier in America that we had from like the software boom. So beyond that though,

I still feel like my thoughts are forming because it is sort of a spiky, poorly distributed

technology. There's a lot of inertia to integrating this. And the models just aren't there. They

are hallucinating. They are making basic errors. I don't think that right now you could really do

anything. I think most of the jobs that are there probably do need a human in the loop to verify,

but I'm consistently shocked at like the quality increases and the outputs. And I think we're probably

now at a question of like it's clearly possible they're going to be able to do this. It's just a

question of tuning, you know, making sure it's intuitive. You know, what is the digital credit of AI,

right? Claude code is so hard to use. Co-work is a huge move in the right direction, but even

co-work is still extremely difficult to use relative to hiring someone that is like more autonomous

and can make decisions on their own. But I think really now it's about like we're going to get

better, more capable systems. They're going to have better harnesses. They're going to reason more

carefully and more thoroughly. And they're going to have, you know, they're going to solve sort of like

memory and sort of updating skills right now. It's difficult to teach it new things and manage

all of its files and stuff. That'll all be abstracted away. And yeah, the implications are massive.

Yeah, I completely agree. I mean, it's like kind of like we're talking about like so many things

are changing so fast that even the concept of using computers is probably going to like

drastically change. I mean, the concept of an operating system and what that even means and

how you interact with a computer and like what you read or what you see or whether you use a web

browser or not, I mean like it's all kind of up in the air to some extent of how fast this is

going to change. Connor, maybe for like the final closing topic, we can talk about any sort of

like important Bitcoin policy, uh, things that that we should be aware of other than Basel,

like of course, but anything like, uh, in America that people are talking about that Bitcoin

community should be aware of. Yeah, a few things, uh, just to give you the latest read out from

Capitol Hill, we are at the final innings of this Congress. Congress runs on a two year cycle.

The last one started at the beginning of 2025. This new one is going to run through 2026,

but really things sort of slow down around election season. And then once the midterms occur,

it's really difficult to get anything through because everyone's sort of waiting on the new class

to come in. So I would say we're in sort of a brief period of time. You could think of it as like

two or three months left, uh, for real serious work to get done. And so in that time, we're

really tracking two major developments. The first is on, there is a Bitcoin and digital assets

tax package that's been talked about for quite some time. It is, there's been a few different

proposals. Uh, I worked on one of the proposals when I was working for center alumnus. We put out the,

uh, sort of alumnus proposal of what a comprehensive crypto tax package could look like. But, um,

right now, there's sort of figuring out exactly how that proposal is going to move through

Congress. Where's it going to start? Who is going to be the lead author of it? You could probably see

it on the house side first in the Ways and Means Committee because that's where tax packages start

that hang up right now is really around the diminimus transaction issue. A lot of you are probably

aware that Bitcoin transactions are obviously treated like capital gains transactions. But for other

types of foreign currency, we have a diminimus exemption. So if you're traveling abroad,

you're spending euros or yen, you don't have to report capital gains or go to the headache of

reporting your basis for each of those sort of diminimus transactions and sort of eases those

foreign currencies as means of payment. So for the past, I don't even know how long, you know,

it's been over five, six years at this point. Uh, there's been sustained effort on capital help to

get a similar diminimus exemption for Bitcoin transactions. And, uh, we recently at BPI put out a letter

to the Senate Finance and the House Ways and Means Committee sort of doubling down on the

importance of this language. And highway and concerns we've been hearing, which is some of the

direction on the hill is actually what's just limit the diminimus exemption to stable coins and not

have Bitcoin be a piece of that. And so we've been sort of, uh, thinking through our options and

trying to make the case clear on Capitol Hill that this is a historic opportunity. We don't have

much time. And it would be a real shame if we go through all of the difficulty of passing a

crypto tax package. And then at the last second, we sort of shrink the aperture to just stable coins.

Uh, the previous version that we worked on, uh, that I worked on in Senator Lumbus's office was

expanded to all digital assets that were qualifying. So I think that is really important. Uh,

and is really to go back to the AI conversation. Minimus, it is hard to think through how many

Bitcoin transactions could be on the horizon. If agentic systems start using Bitcoin more and more

for daily payments or streaming payments, uh, the number of Bitcoin transactions that are currently

taking place on either the base layer or the lightning network or other L2s, you know, that could

a thousand or a hundred thousand X in the next few years because these these agentic systems can do

so much more so much faster than humans. And so, uh, we're really talking to lawmakers spending a lot

of time on the hill, educating them about how Bitcoin works, how it's used for payments. We're seeing

exponential adoption of Bitcoin as merchants adopt it more, especially with the recent integration.

Um, through through block and their their square readers. So that's sort of the latest on the tax

package and something to watch closely. Um, and, uh, you know, certainly recommend everyone to stay

politically involved in that issue and and follow it because it is important really important we get

it right. I don't know if there's going to be another opportunity like we currently have to get

this cross-finish line. The other piece is Clarity Act that's still in this sort of perpetual limbo

between traditional finance and the, um, sort of exchanges, stablecoin issuers and they're hammering

out whether or not there's going to be available yield for stablecoins and some of that. Um, and I

would say BPI's main focus on that were really interested in the self-custody protections that are

in that bill. So Warren Davidson's Keep Your Coins Act is part of the sort of original text that is

currently being debated. And the other major thing is the blockchain regulatory certainty act

protections for software developers, um, to be able to develop software and not be treated as a

money transmitter for developing non-sodial software. And again, more important in the age of AI

because these AI's are going to be building software, uh, about these these financial rails. So we

want to make sure there's like a clear, uh, jurisdictional certainty that you can build here in the

US just like we did with the early days of the internet. So those are the main things that we're

tracking right now. Nice. Yeah, that was a great brief, uh, concise recap of everything that's

going on helped me a lot actually. Um, Connor, we love talking about those two topics, Basil, AI,

and then policy, of course, uh, as well, third topic. But, um, thanks for taking the time to do

this. Where can people go learn more about maybe those reports that we talked about or the

Bitcoin policy institute or, or you as you yourself? Yeah, absolutely. So you can find me on Twitter

at Bitcoin Connor. You can find Bitcoin policy institute on Twitter at Bitcoin policy. And you can

visit our website at btcpolicy.org and I'll have our latest research and our different studies

including the AI study. And, uh, yeah, thank you so much for having me on. It was great to talk

and always happy to chat about sort of the latest developments on the hill. Honestly, it changes

day by day at this point, especially with how quickly these technologies are changing and everything.

So, uh, yeah, I really appreciate you having me on. Awesome. Well, Connor, thanks again. This is

great. Thanks. Perfect.