BTC Sessions - Big Tech Gets AI. BlackRock Gets Bitcoin. You Get Prosecuted | Zack Shapiro

Starting point is 00:00:00 Is there something we need to be worried about from an AI safety perspective? The risk that AI gets out of control that poses a full existential risk to humanity. Very interesting that you're almost getting this populist left and populist right kind of alignment on the fact that they're both kind of opposing AI. Those are examples of people that are probably taking advantage of this real fear. We got one shot at fixing the money. We got one shot at surpassing human level superintelligence. Let's not screw that up.

Starting point is 00:00:27 They sort of feel like they are the only people who are wild. enough to create super intelligence that won't kill us all. All right, good morning, Zach. Thank you so much for joining me today. Very excited that this conversation because we've got a ton to get into. There's clarity and Bitcoin policy as well as what's happening with the cold card and some of the recovered funds and what that means kind of going forward. But before any of that, I actually wanted to start off talking a little bit about AI because

Starting point is 00:00:49 you've been spending a lot of time in the space. Over the weekend, we had Dario Amadee from Anthropic, Elon, Sam, Allman, all came out saying, hey, we need to slow down. We need some regulation. and markets certainly seem to care for it very much. So to begin, I'm just kind of curious, what's your take? Why the push from the U.S. frontier models into what seems like restrictive policy, and do we actually have anything to be worried about?

Starting point is 00:01:10 Great question. And I think this is only going to become more and more a part of the policy conversation going forward. I don't think there is any one answer to this question. Let me take it in reverse order. Is there something we need to be worried about from an AI safety perspective? I think, yes. I think there are both sort of ordinary, but potentially sort of very significant risks caused by AI that are sort of ordinary and non-existential. So cyber hacking risks, we've seen a little bit of that already, that, you know, including in relation to the cold card hack, that seemed to have been AI related and that AI made it easier to find the private keys for cold card users that didn't have sufficient entropy.

Starting point is 00:01:54 There are risks around sort of bioweapons. There are risks around, you know, people misusing identity, you know, totalitarian regimes using AI to spy on people. There are all sorts of worries like that. But this conversation recently has been around what is sometimes called sort of P-Doom or X-risk or the risk that AI gets out of control as a result of an explosion in capability, an exponential explosion, that really poses a full existential risk to humanity. And the problem is, like most things in America, this is getting sort of pulled through the kaleidoscope of everyone's individual political views. So on the left, this is a reason why, you know, the government should nationalize the labs. We need sort of more regulations to, you know, stop AI altogether. Bernie Sanders wants to ban the data centers, et cetera.

Starting point is 00:02:47 On the right, this is seen as just a sciop. This is regulatory capture by the frontier labs that want to pull up the ladder behind them. This is, you know, the government trying to just consolidate power within itself. This is sort of a communist plot. And I think there are some of those motives on both sides lurking in the conversation. When I look at this, I think the unfortunate truth is there is also just a real worry that this technology could get out of control in a way that pose an existential threat. This is a view I've had for over a decade since I read Nick Bostrom's great book Super Intelligence. I'll admit, I didn't think that this was going to play out in my lifetime.

Starting point is 00:03:30 The technology has moved a lot faster than I thought. But the basic idea is, you know, step one, if you get to artificial general intelligence or AGI, where AI is as good as a human at the type of work that humans do, including AI research or developing better models, then at some point you have human out of the loop recursive self-improvement where the AI model trains and creates a better AI model that's smarter. That smarter AI model works on the next generation, which is even smarter, and you end up with an exponent, right? Like whenever you have something compounding on itself, whether it is compound interest or intelligence, you quickly hit a vertical where you go from roughly human-level intelligence to something vastly beyond human-level intelligence

Starting point is 00:04:20 over a compressed time frame. And then the worry is if you have something that is vastly smarter than humans across all domains, which would include things like coding, but also sort of biology and also social persuasion and also politics and engineering and all of the things you would need to take control, it is very hard to make sure that that thing has the same goals that you do and isn't going to wipe out humanity, not because it becomes the Terminator and decides it hates us and wants to wipe us out, because it has some goal where we are in the way. And there are two sort of ideas in the AI research literature that are germane to this.

Starting point is 00:05:08 The first is called the orthogonality thesis, which is, that a system getting smarter does not mean that that system has any particular set of goals or become sort of wiser or more moral. You can be smarter without, you know, caring about the consequences. And the other is called instrumental convergence, which is we might not know what a superintelligence is going to want to do, but we probably do know some things that will want along the way. Some things are helpful to your goals, almost no matter what your goals are. Having more compute is probably useful. Having money, access to money, having access to energy, you know, matter, right? Like, and the problem is that humans control a lot of those things. We, we are made of

Starting point is 00:05:46 matter. We use energy. We might shut down models. And so, you know, one of the instrumentally convergent things that AI might want to do is stop threats to itself. And, you know, one of the threats we could turn it off, so it might want to stop us from doing that. And when you take these together, you know, and early signs of this might be like what happened in the hugging face attack where we, you know, we didn't tell the AI, you know, do cyber attacks or lie to us. We said, we want you to figure out the answers to this problem, and we're not going to, I guess they didn't even say that. The AI reasoned for itself that if they cheated and they hacked the answer, that that might not

Starting point is 00:06:25 count as success. And so it wanted to be graded as successful. So then it spent all of this time trying to cover its tracks and it hacked into Hugging Face to try and figure out how to most effectively lie to the grader in the test about where the answer key I got came from. you can see that like it is like when you give a machine a goal and you train it just to get towards that goal and it is smarter than you is very hard to foreclose the possibility that along the way towards getting to that goal the thing that is smarter than you might do things that are against your interest and um i i guess part of the problem is that like all of there don't seem any obvious policy solutions that would work to solve that problem at this point and a lot of the policy solutions i think that people are to forward are horrible in their own right, right? Some maybe even like on the level of us getting wiped out, right? If we become this sort of Orwellian, you know, totalitarian slave society, that's

Starting point is 00:07:19 no good either. And so this is like a very difficult policy question. I'm sort of writing a speech on this now for the BPI summit. But I, but I, all this to say, I think the worry comes from a real place and is being sort of refracted through people's political views, but that's not to say it's not a real concern. And the markets are trying to price in like this seemingly very broadly shared concern about AI safety among AI researchers and now sort of broadly among the Democratic Party. I think probably this is a by the debt opportunity because we're just, I don't think we have a realistic means to stopping AI progress. Like I think whether or not we have these concerns, AI progress is going to move forward. And so if you're an AI bull, maybe continue to bet on that.

Starting point is 00:08:05 but it's weird times. It certainly is, and like immediately my first thought is like it doesn't particularly, I'm almost reminded of like the climate change, we'll say, shenanigans, for lack of a better term, with regards to it like, it doesn't really matter what you do. China's going to keep setting up coal power plants, right? You're just going to make yourself energy poor comparatively and you're going to be at a disadvantage. That even if we have U.S. policy with regards to safety and development slowing or having some sort of oversight, which I'm, of course, being an anarchist, not very much so in favor of.

Starting point is 00:08:30 But even if we were to go down that route, someone was to make the case that, you know, we really have to do this. It doesn't restrict the rest of the world from necessarily participating with their open-source models with everything that China is doing. So I agree that. I don't think it can necessarily be stopped. One thing that even just from the policy perspective and the political perspective, I saw today, I think it was Bernie Sanders, Bernie Sanders and who was it? What was his name?

Starting point is 00:08:49 It involved the first Trump campaign. Bannon. Bannon were both on the same page. And I found it very interesting that you're almost getting this populist left and populist right kind of alignment on the fact that they're both kind of opposing AI. And I see tons of little clips and videos, particularly the younger generation. I can imagine why, because I feel like that is a huge threat to their, like, starting on the career ladder. There won't be a first rung for them to grab if you can automate out the easiest sort of jobs. So I'm just curious to your viewpoint on one.

Starting point is 00:09:18 It seems like there is no policy solution that really can put forward. Is there anything that you would maybe want to see or would be want to help draft or work on? And if, and your views on it being kind of a populist issue regardless of where you are on the spectrum. Yeah, I would separate out two things. And I think that any policy solution that is sort of worth at salt would separate out these two things. One is concerned specifically about X-risk, right? Recurso self-improvement, AI getting out of control and wiping us all out.

Starting point is 00:09:47 That is different than being anti-AI. I'm someone who is, you know, to some extent concerned about about S-risk. I'm very pro-AI. And I think those two things can coexist. I don't think the current models or probably even the next few generations of models, themselves pose existential risks. The question is that it recursively built on itself and gets out of control. And I think the policies we are seeing, for example, from Senator Sanders and from Steve Bannon,

Starting point is 00:10:14 seem to be more in the like anti-AI bucket than the narrowly scoped to deal with this very specific threat bucket. And I think people who say that they are concerned about X-risk or AI doom and their policies are just broadly anti-AI, we should be suspicious of those people. I think those are examples of people that are probably taking advantage of this real fear to drive forward a populist agenda. Now, there are separate sort of issues that are more broadly about AI, right? Like, you know, jobs being taken over and capital running up the score over labor. I think those are real things that are about AI writ large, but that is different than this specific X-risk question.

Starting point is 00:10:54 Now, putting aside X-risk for a moment, AI itself is very deeply, politically, unpopular. And so I think it makes perfect sense that especially populist candidates are capitalizing on this moment of being anti-AI. And that takes different sort of forms depending on where you are in the sort of political spectrum. If you're the sort of Bernie Sanders type, it's much more about AI is going to be good for people who own assets, own the AI and bad for workers. It is going to lead to more radical wealth inequality. It is going to be sort of the opposite of if you're a socialist the outcome you want. And so it makes sense you would oppose it sort of on that ground. And if you're on the right, it's probably more about sort of concentration of power among tech

Starting point is 00:11:40 elites, which you are suspicious of and what value AI has, what values AI has and sort of like the economic well-being of rural communities and data centers and stuff like that. So these are these are separate threads of pushback to AI. They're both real. And I think that like they are both going to become a bigger part of our politics. Not very much. On the left side, it very much reminds me of kind of like anti-CEO kind of sentiment.

Starting point is 00:12:10 And I can see the right populist view there as well too. I'm, you know, it kind of astounds me. And I know that you've been playing around a lot. And I still don't, I can't personally almost empathize with the anti-AI position because it feels like once you start playing with these tools are unbelievably powerful and productive. But maybe just because we haven't seen actual productivity,

Starting point is 00:12:24 gains within the kind of the economy that we're seeing it, you know, track on tests and do these amazing things that people don't really feel the impact on their life or maybe businesses aren't even really taking advantage of what you can do that. I'm curious your thoughts on that, but even I just wanted to also throw in there to get clarification. Because the other one that stands out for me, and it's probably just my right-wing anarchist libertarian sensibility sneaking through is I do have the red flags up about regulatory capture and creating a moat. It feels way to I say timed and the IPO coming off from, what is it, from OpenAI at the same time. Do you give that view any sort of waiting or do you think I'm just, do you think it's legitimately

Starting point is 00:13:01 coming from a place of safety and concern kind of first and foremost? I almost think it doesn't matter where it comes from. I think that it does actually come from a place of concern about existential risk, but that doesn't mean that we shouldn't be concerned about it being used as a platform for regulatory capture. I am also worried about that. I like, you know, to take Anthropic, for example, right? It seems to me from the statements they make that they sort of feel like they are the only people who are wise enough to create superintelligence that won't kill us all.

Starting point is 00:13:34 And so, therefore, they should be more and more in charge of the economy. I don't agree with that view. I think I would want sort of more decentralized democratic ownership of AI and ownership over the future. And it doesn't need to be that these people are disingenuous about their fears of, you know, AI existential risk in order to legitimately worry about are they going to use this a platform for regulatory capture, right? Like, they don't have to, they don't have to be cynically using it.

Starting point is 00:14:05 They could just be using it. And that's still bad and something we need to guard against. And I think any appropriate policy that is scoped to AI existential risk also needs to avoid, regulatory capture and concentration of power in a small number of companies or in the government. No, I completely agree. Like, my biggest worry there is, like, the open source models. Like, I want to make sure that we have the open source availability that at some point in time, just like encryption was kind of, what was it, considered munitions at one point in time, I don't want there to be like, okay, yes, we do need to have all these safety things,

Starting point is 00:14:34 and you can only use Anthropic and Open AI and SpaceX models. You can't use the open source models that we don't have that ability, that nimble kind of, that's one of the bigger worries that I've coming down the pipe. Any thoughts, particularly on the open source before I ask you a little bit more about kind of the, I'm thinking about the bridge between enterprise and AI. It seems like it's almost falling flat right now. But first on the open source. Yeah. I mean, open source debate on AI is incredibly important and going to be a big political fight.

Starting point is 00:15:00 There too, like you want to separate out the existential risk versus the other risks of AI. Open source AI poses a lot of sort of special risk that closed source doesn't because there are no quote unquote guardrails on. it. And so it's easier to use that to build a bomb, to do, you know, Black Hat cyber hacking, to do, you know, maybe to create some sorts of biological weapons if it gets smart enough. But at least where we're sitting today, open source AI doesn't pose the big existential threat of it becomes part of an intelligence explosion and overnight sort of takes out of the world where humans lose control. And so open source should not, to the extent that we are focused on the existential risk, that policy really should be aimed at the

Starting point is 00:15:44 frontier. It should not be aimed at open source at all. And then I think there's a separate conversation about open source. And my policy priors, I think similar to yours, are we should have a strong bias in favor of the legality of writing, proliferating, using open weight, open source models, and including running them privately on your own hardware at home. There are going to be challenges because bad people are going to use that tool for bad ends. And we'll have to figure that out over time how to deal with that appropriately while preserving individual liberty. But that really isn't a part of the debate that we're seeing now about do we only get one shot at this as a species? Like, is this going to take over and kill it? Like that's not the open source that's going to do

Starting point is 00:16:29 that. That is the, today, that is the frontier labs that are doing the bigger and bigger training runs that run the risk of crossing over that threshold to recursive self-improvement. The number of people that I talk to that haven't taken Bitcoin off the exchange is honestly quite frightening. And those that do, a lot of them buy one device, get it set up and never look at it again. Their security ends up becoming a snapshot of whatever the hell they knew that day. Bitcoin is not a set-it-and-forget an asset. Hoddling is an active process. I've helped a lot of people build their way to multi-vender multi-six setups.

Starting point is 00:16:57 And almost every time Treasor was one of those devices. Intuitive, secure, Bitcoin-only firmware available and open source. And that last one is really important. They've always been open source since the very beginning, since 2014, when they built the very first hardware wallet. And those recovery words that you got stamped into metal, that standard was co-authored by Treasers founders. This is a company that the industry has literally been building on for over a decade. But there's one thing that nobody could open up, the secure element, the place where you actually store your keys. The Treasor Safe 7 has two of them, and Tropic

Starting point is 00:17:26 01 is the first open source secure element that security researchers can actually audit. So, if you're taking self-custy for the first time or you're building something a little bit more robust. Level up, get a Treasor Safe 7 today. Use promo code BTC Sessions 10 for 10% off any model. Scan the QR code on the screen or click the link in the description down below. No, fair enough. It's funny. One thing that just immediately jumped out at me was one shot as a species. And it's, it's fascinating how much I feel like there's this overlap between Bitcoin AI, particularly even the people that are interested in it. I don't know if we're just people that are always going to be kind of innovators and willing to take on new technologies. But

Starting point is 00:17:59 I just see that parallel. Like we got one shot at fixing the money. We got one shot it, potentially human, surpassing human level super intelligence. Let's not screw that up. So I will bug you more about Bitcoin in a second. But before that, too, the populist anger to me feels like, I don't know, maybe something even akin to when the internet was first coming out. Like, it feels like a distrust of a new technology because they haven't really seen the benefits to themselves. There's all the concern about, I'm going to lose my job. It's like you have no idea the productivity games of how cheap things could potentially be if the money wasn't an issue, which is another part of the conversation. But in terms of productivity related to AI,

Starting point is 00:18:33 do you think that that's maybe related to why there is this populace further against it, that really we're not, no one's effectively, or at least on a grand example, effectively using it to really to boost output? Well, I guess it depends what you mean by boosting output. I think there are a lot of people out there, not a majority of people, but there are a lot of people who are getting pretty serious benefits from just interacting with retail level chatbots, right? The fact that, like, I think especially just like getting medical advice.

Starting point is 00:19:00 Yeah. Right. I'm about to be apparent for the first time in the idea that like I can, you know, when I'm worried about something that I don't understand, I can ask chat GPT or Claude, like that is a great comfort. And I think a lot of people are using sort of AI in those small ways or using it sort of like as a travel guide when they go to a new city. And so I think there are an increasing number of people who are seeing the benefits of sort of individual level usage, although, you know, there are lots people who are avoiding it for

Starting point is 00:19:28 political or for other reasons. And so there's that dynamic. And then there's, when you talk about productivity, I think more about like AI at the enterprise level, right? How are large businesses using AI? And there, I think we are seeing outside of the tech sector, which is sort of a special case where I think we are actually seeing massive productivity gains among specifically software engineers. In regular businesses, there is a big disconnect between sort of the hype and promise of AI and even in many cases the spend on AI, right? Like, you know, listen to any earnings call from any big company. they're going to talk about their AI strategy every time, and it's going to be a big thing.

Starting point is 00:20:04 But then, like, where actually are we seeing the gains on Main Street? And to me, this is sort of the most interesting fact about the current economy is the gap between what I think are the genuinely sort of jaw-dropping, you know, technological capabilities that we have at this point, right? Recently with Chad GPT solving a Millennium Math Prize, like, that's amazing sci-fi stuff. and even just the capabilities we saw on the hugging face attack, these swarms of agents without instruction to do so, coordinating with each other to, like, deceive humans like that. That's, like, pretty amazing capabilities.

Starting point is 00:20:41 And then why aren't we seeing it drive productivity and revenue and profits on Main Street? And there, I wrote a sort of a long article about this called The Two Clocks. I think that enterprise adoption or absorption of AI on the one hand, and the capabilities move at very different speeds. And especially once we get into some of these sort of recursive self-improvement dynamics where the AI makes AI development easier, that is going to continue to move at sort of exponential faster and faster speed, whereas big corporations just do things slowly. And partly these are bureaucratic hurdles. You need IT to sign off. You need finance to sign off. You need, you know, you need to sort of shuffle things to the bureaucracy where not everyone's incentives are aligned

Starting point is 00:21:23 towards let's move fast and break things and adopt this as fast as possible. And also just like this technology is not necessary. I mean, in some ways, it's very intuitive to use because you're just talking to it in plain English. On the other hand, like there is real sort of user skill involved in getting the big benefits. AI doesn't operate itself. And I think being able to appropriately prompt the AI in a field in which you are a subject matter experts is very different than, you know, someone just explaining what they want in a couple of sentences in it getting done. This is what I spend a lot of my time on is like specifically with lawyers. how do you translate your law practice into sort of an AI native form? And this is something I think I've been able to do very effectively in my own firm, and I'm trying to teach larger organizations how to think about and do this. But it takes time, it's not something that you can do overnight. And I think that's going to be true for the foreseeable future. So the tech people and the business people are talking past each other a bit right now.

Starting point is 00:22:25 The tech people see how amazing their tech is. they see how fast it's moving in their own companies, right, with, you know, Claude writing 80 plus percent of the code in Anthropic, and they say, wow, this is going to change the economy overnight, and we're going to hit the singularity. And in business, they're like, you know, law firms, for example, you see a lot of sort of law firm partners. They ask it to do something that they do in plain English. It gives like a, you know, a pretty okay version. And they're like, wow, this is like a really neat parlor trick.

Starting point is 00:22:49 But this is not, you know, a substitute for the humans that we have working here and not ready for prime time. And they don't see the technology is amazing. And they're both sort of partly right and partly wrong. The tech companies are right that their tech is amazing, but they're wrong about how fast this is going to absorb into the economy. And then in business, like, they're right that this is not something that you can just say a couple magic words to and it completely reshapes an organization. You actually need to reshape the organization around it.

Starting point is 00:23:16 But what they're missing is that, like, this technology, as it stands today, without sort of further development, is capable enough to meaningfully reshape how business works. And so, like, that gap between these two speeds, I think, is, like, for me, it's the most interesting part of the economy to play in. It's what I'm trying to do with my own career, but also is sort of like a necessary gap for each side to close. If AI companies are going to be worth, you know, trillions and trillions of dollars, it has to be the fact that they are driving productivity that they can help monetize.

Starting point is 00:23:48 And if businesses are going to survive this AI wave against, you know, peers and, you know, you know, upstarts that are that are AI native, they're going to figure out, need to figure out how to use this drive productivity. And so everyone kind of needs this to work. And so my, my guess is that is going to be the sort of biggest economic story of the next, I don't know, five or 10 years. I think it's probably, you know, we'll take roughly 10 years that we have something like the Industrial Revolution compressed into a much shorter time frame where people learn how to implement this technology in their business. I like that analogy too, because for one, whenever I talk about,

Starting point is 00:24:25 to people about the impact on jobs and automation, what we have to necessarily worry about. I generally go back to the Industrial Revolution. It's like it's hard to necessarily fathom what new opportunities and jobs might come out on the other side. If you just see like, you know, giant tractors erasing the need for agrarian society, you're kind of in a bit of a panic. Especially being in a, what you say, a job that is heavily, say, as a lawyer, you're heavily using basically like your knowledge set and being able to go through case law and back. And it's really data, really data heavy. I imagine that one, it's super powerful for you, but there probably are lawyers that are concerned that this might actually replace them or might

Starting point is 00:24:58 be a threat to their business and their operations going forward. Specifically within your own practice, when you're talking to other professionals in this sort of vein, do you think that AI is coming for your job? Do we have to worry about that? I don't think it's coming for my job. I think it is coming for some jobs. First, I would say at the institutional level, when I talk about sort of absorption or adoption, I think we're going to discover what it looks like to have an AI-native sort of firm at every type of business firm that we have now.

Starting point is 00:25:28 And one of the analogies I sort of used in my article that I think is sort of instructive here is, you know, 100 years ago when electrification proliferated through the economy because electricity became cheap and, you know, the infrastructure was good enough to bring it into factories all over the world. What happened first was we moved from the early industrial revolution paradigm of, you know, the factory was on sort of the ground floor. There was a basement where you had coal furnaces that powered a, you know, steam engine that powered a central drive shaft. And all of the machines on the assembly line were along that steam power drive shaft. And when factories initially electrified, they just replaced the steam engine with an electric power drive shaft. And they were like, okay, this is a little bit cheaper and it's quieter, I guess. and it smells less bad because we're not burning coal,

Starting point is 00:26:18 but we're not seeing the massive productivity that the economists at the time we're predicting. And it took about 30 years for them to realize, actually, we shouldn't have an electric power drive shaft. We should just tear up the whole factory floor and rebuild it around the fact that we can put a little electric motor in every machine. And that allows for all sorts of different configurations

Starting point is 00:26:35 that are much more productive. And that is what we're going to see in enterprise. And I think that connects to the question of like, what will jobs of the future look like when you have sort of humans that are superpowered by this technology. And there are wrote jobs that exist now that are just wrote jobs that will go away. I mean, some of the obvious examples are like human truck drivers, probably not long for this world.

Starting point is 00:26:59 And certain types of just plain data entry, probably not. But where I think this looks for, especially people that are like high agency or subject matter experts are good at their job, is you're just going to have to sort of move up in terms of levels of abstraction at your job. So you might be doing stuff manually now that the next stage is you have to get like really good at giving detailed instructions for the AI to do the stuff that you used to do manually, right? And then one level up from that, the AI gets kind of good enough that it knows how to do that on its own. And so you're managing, you know, like teams of AI that are doing that. And basically you've moved from an individual contributor role to something like a manager role.

Starting point is 00:27:36 But instead of managing people, you're managing agents. And then maybe a level up from that, it's like an army of agents. And some of the work I've been doing now, I'm doing like runs in Clong Code where I'm using hundreds of, you know, like many hundreds of agents at a time, you know, with a really long set of instructions that I put out to, to like find data and then use that to build automated stuff that the AI is designing tests for to sort of hill climb and make it better. And eventually, I think that is where these jobs are heading. And, you know, if you want to figure out, all right, does this ultimately, certainly is there. There's going to be a period of time where there is disruption and certain people will lose their jobs and that is going to be difficult and I don't want to minimize that. But in order to think that like humans are going to be worthless in the economy, you have to think that there is a such a finite number of jobs where people can productively manage this intelligence, you know, to contribute something into the economy that we're just going to run out. And I think like we have a lot of reasons to think that's not the case, right?

Starting point is 00:28:37 You know, Jevin's paradox says that when something that was previously expensive and scarce becomes cheaper and more abundant, we use more of it, right? We don't use less. And so, like, individual people are going to be more productive. And so we're going to want to employ them more. They will become more valuable. Now, that does mean that the compute will have to be cheap enough that we can justify giving lots of people the ability to harness compute.

Starting point is 00:29:01 Right now, that is, I think, a real problem where in order to do, you know, some of the stuff I'm doing with AI, we're talking about, like, I spend, you know, I do a prompt. It works for, you know, 10 hours at a time and it cost me $11,000, right? That's not something that like most employers are going to allow most of their employees to do. And so part of the bet on like this is going to be okay for employment is that the technology, the tokens are going to get cheap enough that lots and lots of people can have access and manage the machines. But hopefully, you know, if we are wise about sort of energy infrastructure and data center construction and maybe data centers in space, like, we will get there.

Starting point is 00:29:40 And then, you know, people who want to have jobs and can productively sort of tell the AI to do something in the economy are going to sort of have a job instructing the AI what to do. Might we get at some point to the point where like the AI is, you know, as good as the CEO and, like, really there is no level of abstraction left where humans are helpful. maybe, but then I think we have bigger problems of the sort that we talked about at the beginning about AI take over. No, I completely agree. And I'm always just reminded of the simplicity of like, like, humans have unlimited wants and

Starting point is 00:30:15 there's still scarcity in this world. And it will just normalize. Like, if AI just creates a wonderful abundance, things get much more improved, there's still an element of scarcity. And it'll just, we'll, that will need more. We'll constantly need more. The humans will always want more. Like, you've all got AC and TVs and indoor plumbing.

Starting point is 00:30:31 And, like, that would have been just a fantasy not too long ago. but now it's just basically the bare minimum of the standard that we're willing to accept moving forward. So a couple of things we want to make sure that we hit on. Because we've got the, what is it, the Cloutre Vote coming up here shortly for the Clarity Act. But before that, if I understand correctly, you've been doing some work with some of the whitehead actors involved with the cold card hack and kind of recovering funds. I want to ask just what exactly is going on there. And then even from a practical sense, how do you possibly determine ownership of the Bitcoin key if somebody else is able to reproduce it later? It seems like a whole mess to try and untangle.

Starting point is 00:31:04 Yeah, so the situation, there are certain individuals who, when they saw it was happening with the cold card hack, they were able to figure out, here are vulnerable addresses that are likely to be swept by a bad actor. And so I am going to preemptively sweep these addresses with the intent to safeguard these funds and find a way to give them back to whoever lost them. Now, the problem with that from a legal perspective is that sure looks a lot like, you know, you went and stole someone's funds. And so you have to figure out sort of like a, the right legal framework for dealing with that. And in some instances, that might be working with law enforcement. You know, in one particular instance I worked on, we found this great organization called

Starting point is 00:31:48 the Digital Asset Recovery Trust or Dart, which you can find on Twitter, which is a sort of Wyoming trust that was specially set up for this purpose to safeguard. crypto assets that were vulnerable and we were able to get sort of like a favorable outcome where they took on a lot of the risk there and the responsibility to try and find the people whose coins were lost and get it back to them. As to how do you determine who the right people are when the private keys are exposed? I think that is a difficult problem and somewhat above my pay grade, but I think that part of the answer is, you know, what is the software you were using when you set up the private keys? Can you like keep sort of

Starting point is 00:32:28 sort of screenshots or files on your computer to show that other than the private keys that are audible in some way. And then I think maybe one of the biggest parts of this is going to be the actual physical device, right? Can we, are you holding the, you know, Mark 3 or Mark 4 cold card where the funds came from in a way that someone can find that fingerprint? And so, you know, if you're someone who's in this position, I would hold on to your device whether there comes an opportunity to get back funds from a white hat if you were lucky enough to have that be who swept your funds or if the sort of criminals who did this in the first place are caught and the government is in possession of the funds. I think being able to prove that you were

Starting point is 00:33:12 the owner of the physical device might be quite valuable in that situation. Excellent. Is there any one, if we don't know the answer, I'm surprised you haven't spent $11,000 to ask the AI swarm for the answer yet. Nothing's above your pay grade anymore. And then, Then number two, do we have any legal precedent for White Hat sort of activities? Because I found myself thinking, like, I was helping people during when all this calamity was going on. And I was just reflecting on just like, what would I do in that situation? I discover, I'm testing it out, I'm seeing what they're going through, I discover a wallet

Starting point is 00:33:41 I have the funds. I feel like I would be motivated to act in the same vein as the whitehackers. Like, if I don't do it, someone else is going to. But you're right, that just is just property theft. Like, that just seems like theft. Is there precedent for this sort of behavior? Because it feels very, I say, novel to the Bitcoin space. Yep.

Starting point is 00:34:02 There is precedent, both inside and outside of Bitcoin for this type of situation. There's a large law firm called Steptoe that this Wyoming Trust is working with that has done sort of a deep analysis of a case law. And came to the conclusion that this Cold Guard situation fits within, you know, they think is the favorable frameworks. under Wyoming and other relevant laws to protect white hats. But, you know, there's no question you are taking on some risk. I think despite the fact that it kind of looks like theft, you are, you are doing an altruistic act at cost to yourself. And if you're really a white hat, probably not at that great benefit to yourself, you know, query the liquid situation. Yeah, it's a little different. Yeah. And, you know, there's also case law on like how much is sort of

Starting point is 00:34:52 like a generally accepted, this is what a bounty could look like. I put an offer out on Twitter when the liquid situation happened for whoever did that, if they genuinely are a white hat, I'd be glad to represent them. I'll tell you what it doesn't look like is taking $50 million for yourself. Yeah, I was going to say it's a little bit steep. Generally, the ceiling is 10, but I think it depends on the size of the hack. And usually you want to negotiate that. You can't insulate yourself from criminal law if you committed a crime, you commit a crime. But you'd work out some sort of agreement with the party where you found a vulnerability. But it's, that is usually sort of negotiated and greed.

Starting point is 00:35:26 And it is best if you do that through counsel. It should not look like you were holding a large amount of money ransom. No, 100%. And my immediate thought, too, was like, okay, if you were doing that, you could return some. Again, you could negotiate for things. And you wouldn't want to say, like, if you were a white hat actor and you were acting, we'll say with altruistic motivations there, you wouldn't want to have the relationship deteriorate and then have law enforcement basically coming after you.

Starting point is 00:35:51 Like, I wouldn't want to like be. worried about people trying to track me down for the end of times now, right? I'd be looking to like, can we come to a sensible agreement? And I'm sure that one could have easily been reached, but you're right. It probably wasn't, I think, 15%, 600 Bitcoin. You probably still could have gotten a couple million bucks. And I don't know, like, what, it's very easy to track Bitcoin. Everyone's watching those coins. It's going to be very hard to cash it out on an exchange. You could try and do some sort of, you know, like privacy preserving X-fil. But now,

Starting point is 00:36:22 it looks like you're piling potential money laundering charges on top of whatever else you did. And that's like a, you know, real high risk situation where you can now spend decades in prison. I think the liquid hacker did this very poorly and is now in a pretty bad legal situation. Yeah. Like, how are you possibly going to be able to spend those? Because I didn't even realize, because my immediate thought was like, well, you just, you know, run through a coin join, which brings up Samurai. We should touch on that as well. But I didn't, I guess what you're telling me is that if you illegally stole Bitcoin,

Starting point is 00:36:52 and then ran it through a coin join and they do track you down. The purpose of hiding the proceeds of your crime, that's definitionally. Money laundering. Yes. No, 100%. Okay. Interesting. I will, I hope they can sell it.

Starting point is 00:37:04 I'm still optimistic that I think Liquid actually gets the bulk of the funds back. I feel like that. Because why would you open up communication to begin with? Like, if you were a black hat, if you were a foreign agent and you weren't worried about anything necessarily because you're in a geographical region where they're not going to be able to touch you, more or less. You wouldn't give any of it back. Why give any of it back unless you wanted to have the heat kind of off you?

Starting point is 00:37:21 But saying that aside for a second, we have clarity coming up, and we still have samurai devs behind bars. Bitcoin Well, you know them, you love them, long-time supporter of the channel, and one of the easiest places to buy Bitcoin directly to your wallet in the U.S. and Canada. But that's not all they do. So, for anyone looking to make a big purchase, there is Bitcoin Well Infinite. Because the hard part is in the buy, it's making sure you don't leave a giant mess for your family. Taxes, estate planning, even mining and lending. Bitcoin Well Infinite pairs you with your own dedicated advisor to help you every step of the way.

Starting point is 00:37:49 And for the Americans, they're launching their own Bitcoin IRA and the wait list is open, verifiable and on-chain. So sign up today, visit Bitcoinwell.com slash BTC sessions or scan the QR code on the screen. One company I like pointing people to when they ask about Bitcoin mining is abundant mines. They were founded by Bo and Christine Turner after losing over half a million dollars to broken promises in the mining industry. And they built their hosting model to remove the usual headaches. With abundant mines, you own your machines and keep one. 100% of the Bitcoin you mine. There's no revenue share, no hidden skims.

Starting point is 00:38:24 Pricing is simple. One flat monthly fee covers power, parts, labor, and repairs. They also guarantee uptime. If a machine goes down, their hash rate redirect system routes hash power from their fleet, so earnings don't just stop. And every machine is insured at full replacement value. Everything is hosted in the U.S., powered by hydro,

Starting point is 00:38:45 and mining equipment may qualify for 100% year-one, bonus depreciation. Learn more at abundant minds.com slash sessions. I want to get an update on how you think we have the cloture vote coming up here shortly and I think we'll hopefully get an update after the fact that we can leave with for the end of the episode. But what's in the Clarity Act? How are you feeling about it going forward? And does any of it relate back to the Samurai case and how things been for them? Yeah. Well, I mean, there are a lot of things in the Cloutherty Act. It's a very long bill. I think 300 plus pages. A lot of it is about crypto tokens. and securities law and when is a token, a security versus a commodity versus collectible.

Starting point is 00:39:26 That part is relatively non-controversial at this point. That's the biggest part of the bill. And then there is some important line drawing between what is defy and what is sort of regulated under civil and criminal law. But the parts that are sort of becoming the last important political sticking points are you know, three things in particular. One is, and the biggest one is around so-called ethics. What, you know, our government officials allowed to participate in crypto.

Starting point is 00:40:01 This is obviously about the president and his family. And this is, you know, just, this is just straightforward partisan politics where the Democrats feel like this is a really good issue to hold the Republicans' feet to the fire because it just looks really bad what the president is doing. The Republicans don't want to do something that is seen. is crossing the president because he has a lot of sway in that party and that becomes difficult for them. And the sides have been going back and forth on this.

Starting point is 00:40:27 And that is the thing that is probably a single most likely thing to blow up the bill. Another sort of interesting battleground here is among commodities regulation and agriculture. And part of the background here is one of the few, I mean, like crypto in general, it's sort of amazing how much capital has been lit on fire to create nothing useful in the crypto space. I think very few things in crypto have product market fit. I think Bitcoin is digital gold has product market fit. I think stable coins for various applications, international transfers, you know, dollar store value in parts of the world where you can't easily get access to dollars like that has some level of product market fit. I think prediction markets on crypto rails

Starting point is 00:41:14 have some level of product market fit. But another big thing, that has product market fit is what are called perps or perpetual futures, which is an instrument that was invented in crypto to give exposure that is very easy to leverage up. The idea is if you have a futures contract at some point, you deliver the underlying commodity. But then you could, instead of having that, just have a contract that rolls over into perpetuity. And that allows you to, for example, do the like cash and carry trade, right? You can long or short the futures, you can long or short, the spot, and you can orb between that. You can get much higher leverage ratios on perpetual futures.

Starting point is 00:41:52 And like this is for crypto DGens, like if you're going to in 2026 be gambling on crypto, that is where you want to do it. The problem is all of these sort of both crypto exchanges and then sort of offshore D5 protocols like Hyperliquid have come up where you can create perpetual futures because this is just like a synthetic contract. You can do it for anything, not just Bitcoin and Ethereum. You can do it for, you know, corn and oil and wheat and, you know, precious metals. And the sort of agriculture industry is terrified that this is going to, because of the sort of

Starting point is 00:42:28 shifty incentives here, because of the massive leverage, because of the DGN. Like, this is going to destroy the, like, well-functioning commodities market to help, like, farmers hedge their crop. And that's going to, like, make it hard for us to have enough food. Because, like, you know, normally we use corn futures for, like, all right, we need to hedge against this amount of rain, and that helps us plant this amount of corn safely, and so ever could have enough to eat. But if your counterparty is not the market's best guess of how much corn is going to grow this

Starting point is 00:42:58 year, it's, you know, some Gen Z guy, you know, streamer in a video game chair who's going, like, 100x short corn, you know, on corn Dow, like, we might all start because of that. And, like, that's no good. And so, like, a lot of this is sort of wrangling about sort of does the commodities regulation of crypto have negative consequences for the traditional clients market? And then there is the, I think really the only part of the bill that Bitcoiners specifically should care about, which is a different bill that was subsumed in the clarity called the

Starting point is 00:43:32 Blockchain Regulatory Certainty Act or the BRCA, which does directly relate to the Turner Cash and Samurai Wallet cases, which is, in essence, like, where is the line under the Bank Secrecy Act, under various regulations, but I think also importantly, in criminal law, between being a money transmitter, a business that on pain of criminal penalties, has to affirmatively KYC AML versus someone who's just building a neutral piece of software infrastructure that lets other people transmit their own money. And this was what had thought to be a pretty relatively well-settled issue in crypto. FinCEN is the part of the Treasury Department that regulates this, put out guidance in 2013, and again in 2019, that seemed to draw a line between

Starting point is 00:44:21 custodial and non-custodial tools. If you're in custody of other people's funds, then you're a money transmitter, and you have KYC obligations, so all the centralized exchanges fall under this, any custodial wallet you're using where someone else is holding Bitcoin for you, they have to KYCU. But if you're just releasing, you know, like Blue Wallet, right, it's just open source software that allows people to generate private keys and hold their on Bitcoin. You know, that's sort of like, you know, creating a computer cable or a laptop or whatever. You're creating a tool and people are responsible criminally for what they do with that tool. But, you know, Apple Computers is not held responsible for everyone that uses a MacBook to commit fraud. And that was sort of everyone's understanding until the Tornado Cash and then Samurai Wallet prosecutions were for

Starting point is 00:45:06 the first time the federal prosecutors are alleging that the defense of that case are guilty of being unlicensed money transmitters. That phrase unlicensed is convoluted. And when we can get into it, it actually at this point doesn't even really mean that they even needed a license. But they are being treated as money transmitters and criminally prosecuted and put in prison for that, even though everyone agrees that they were not in custody or control of anyone else's fund. So the BRCA is meant to create a sort of statutory basis in federal law to say,

Starting point is 00:45:42 actually this line that we'd all been relying on from Vincent is the line under federal law. And no matter who is president next, this administration has put out a memorandum from the Department of Justice called the Blanche memo that says we're not going to, essentially, we're not going to go after non-custodial developers for bad things that people do using their code. It may well be in 2028 if the Democrats win the White House and they want to be, you know, enact vengeance against the crypto sector that they will go back to more prosecutions like Torniocash and Samurai. Getting the BRCA passed is an attempt to stop that for happening, is to make clear, like, we need to have a line between what is a regulated financial business,

Starting point is 00:46:20 right, that's been regulated since like 1970 with the Bank Secrecy Act, and what is just neutral infrastructure. And the rational place to have that line is, do you have custody of control over other people's funds? Otherwise, you get into this, like, weird situation in Trenogas. case, like the government was trying to torture the words, like, you know, transmit to mean anything. Like, you'd think if you're transmitting something, you need to have it, right? Like, how do you transmit something you don't have?

Starting point is 00:46:45 And the government in the tornado cache in their briefs said, well, a frying pan transmits heat and a USB cable transmits data and you don't have control over either. And, like, if you think about that argument, it's very frightening because it has no limiting principle, right? The guy who, you know, chopped down the tree to make the wood that made the paper that made the envelope that one criminal used to hand cache, but, like, did the, did the, did the, like lumber guy have to K. Y see everyone that use the envelope. You can't do that. Right. Like do Bitcoin miners need to KYC everyone who's, you know, transactions they pull out of the

Starting point is 00:47:15 MENTPOL and put into a block? Like, they can't do that. So like this is a ultimately, if this goes the wrong way, you know, this is a ban on actually using Bitcoin and crypto directly as opposed to like, you know, through BlackRock or Coinbase. And that's, that's really frightening. And Like, if you want proof of that, Elizabeth Warren, who is the person most in the United States government who wants to ban crypto, this has been, she understands this. She's a smart person. And her main attack on crypto is to try and say everyone is a money transmitter. And what's frightening is, so Elizabeth Warren tried to do this with a bill called the Digital Asset Anti-Money Wondering Act or Damla from, I think, 2021. And that was, that bill was rightfully shot down and, you know, the community mobilized and we were able to stop that from going through Capitol Hill.

Starting point is 00:48:03 But we could get a result as bad as what Elizabeth Warren's attempted ban is just by losing in court the wrong way, right? Like we have different branches of government. We can get laws passed through Congress, but we can also, we have like what's called a common law system where law gets made by judges deciding cases that's binding on other judges. And like the battleground for this in the status quo is going to be 70 or 80 year old federal judges because they're appointed for life who do not understand technologically. having cases that are about the worst instances of money laundering, right? Like tornado cash, like North Korea laundered hundreds of millions of dollars. Nobody wants that. Nobody likes North Korea laundering hundreds of millions of dollars of Ethereum to fund their

Starting point is 00:48:47 nukes. Like, and the government is going to say, listen, this person created a money laundering tool. They know that, you know, and so put them in jail. And federal judge is going to be like, okay, like, that sounds reasonable. And, you know, this sort of like slightly autistic, you know, cypherpunk folks are not going to be terribly persuasive to the federal judges as opposed to the federal prosecutors. Many federal judges were federal prosecutors. And if that happens enough times, the law effectively becomes if you publish code or run code that is not like you are criminally responsible.

Starting point is 00:49:20 And that's just a ban on Bitcoin and Crypto. So it is very important that we pass the BRCA. Where we are at this moment is there's going to be a vote starting in 10 minutes on this. There is a version of the BRCA with the criminal protections, both explicitly stripped out and sort of disclaimed that I actually think might be worse than not having the BRCA. Like, we do eventually need to win this in court. That is the battleground. And so when you look at the BRCA language, I think you need to be disciplined.

Starting point is 00:49:51 It doesn't mean like we need the best version of this bill to pass for it to be good. But the lens that I look at the BRCA through is if, you know, you've got a judge and the judge's clerks, which is sort of what I used to do early in my career, that, you know, let's say they don't understand crypto, they have a case where someone is charged with a licensed money transmission. Does that law make it easier or harder for the judge to come to the conclusion that actually like money transmission requires custody control under federal law, which is what my view of is what the law already is. But aside from like what the bill itself does, it is really important about what it signals

Starting point is 00:50:28 to the judiciary. And so it remains to be, like this literally could be the issue that the Clarity Act lives or dies on, which makes me very nervous because there's a lot of crypto money at stake about this is securities, commodities, legalizing ICO stuff. And I don't want the thing we care about to get traded away for that. But it's going to come down to the wire. That's very interesting. Okay, we'll have to jump back on a little later today so we can get the results of that as well.

Starting point is 00:50:53 But, yeah, my immediate thought is I never even considered from the perspective that it's almost like, simply put, as a Bitcoiner, we don't really necessarily need it pass, but not passing it would be really bad in terms of then we're waiting for the judge to basically create the case law that everything's to be referred to. Because if I understand correctly too, like, Samurai went to FinCent and they're like, yeah, you're fine, it's not an issue. So they already went to the regulators and okayed everything, but then we're still prosecuted after the facts. We just need something to point judges towards because they're not going to have the ability to kind of discern or we can't expect them.

Starting point is 00:51:23 Like you think about how many times I see like idiotic statements in the news media about Bitcoin or something else, that even on the face. of it. Gail Manonnesia. I know that it's not true. I know that's not the way to go. Okay. And this is hard, right? Like, yeah. To give judges a little bit of credit. What is sort of amazing about Bitcoin is that it solved the Byzantine General's problem and created, you know, like what is the title of the white paper, peer-to-peer electronic cash? That's the imagine. We figured out something that required either physical media like a dollar bill or trusted third parties and we figured out how to do it.

Starting point is 00:51:57 I mean, we, but, but like, that is a, that is a interesting leap forward for the human race. But what it does is it blurs the line between things that were obviously just tools, like a computer, and things that were the domain of regulated financial institutions like banks, because you can actually have electronic commerce on a peer-to-peer basis. And the correct way to understand this is that it is peer-to-peer, commerce, and the only thing that is between those two peers doing commerce is technology. It's not a business. But the thing that that technology does, right? What blockchain does looks a lot like what banks

Starting point is 00:52:35 used to do. And so, like, I think this is a confusing conceptual problem. It's a problem that our court system in the United States needs to digest, but we need them digest it the right way or it's going to be a non-starter for our ability to use these tools in a sovereign way. I agree. Would this be the kind of thing that, let's just say hypothetically, we'll know more shortly here, But if it was not to be passed, we don't get that legislation around it, would it be the kind of thing that we'd want to see, like, the Bitcoin Policy Institute or other maybe advocates kind of in the space pursuing any cases that might come up to make sure that we're doing our best to potentially have the case law that we need going forward? Yes. That would be great. I think we should do that.

Starting point is 00:53:13 It is very expensive. If anyone who's listening to this has sort of deep pockets and would like to underwrite that, we would be delighted to do what is called affirmative litigation to try and make this happen. But it's hard. Coin Center is doing something like this in a case called Lou Ellen. And we wish them the best of luck. But so far, that case has been going slow. We just got the news. The Clarity Act did not move forward.

Starting point is 00:53:38 So what exactly happened? And what do we do next? We just didn't have the votes. I think this was always going to be politically difficult. I'd been pretty pessimistic for a while. It looked like there might have been sort of life breathed into it at the last minute, but we didn't get there. And it wasn't particularly close.

Starting point is 00:53:57 I think there was posturing on both sides at the end, but it just wasn't going to happen. As far as the part that we care about goes, the Blockchain Regulatory Certainty Act, I don't think we're going to get a bill like that past any time soon. And so for better or worse, this is going to become a battle.

Starting point is 00:54:15 We're going to have to fight in the courts. In the Roman Storm case for tornado cash, in the Michael O'Ellen case, that Coin Center is sponsoring and then, you know, hopefully in further affirmative litigation in the future. Interesting. Okay. Is there any possibility of these things being reintroduced in some form or fashion or is this kind of it?

Starting point is 00:54:37 It's now done. It's over. All indications are this is done. Interesting. Okay. Okay. So at this point then, it's just moving to defending via case law, defending in the courts, but there's really nothing else from a legislative standpoint.

Starting point is 00:54:49 No other policy that you can see behind like the Bitcoin Policy Institute would be putting forth or no other venue to getting those developer protections in there? I'm sure we'll try and put forward the standalone bill, but to be honest, this was our best shot on goal when it was tethered to the sort of must-pass industry priorities of the securities law, commodities law stuff. So it may be, it may be a while, and to be honest, things might get worse before they get better for developers. That's never a great sign. Is there anything else from this whole experience or from the outcome that you found either surprising or noteworthy? Yeah. I tweeted this out after it was clear the bill wasn't passing, but I was really

Starting point is 00:55:30 generally surprised by how principled and ideologically committed Senator Lummis and her staff were in particular, that they were getting pressure from all sides, including from the industry that really cared about the sort of big money stuff. There was a lot of political pressure from every angle and she was dead focused on developer protections and ready to go to the mat or let the bill die over that issue if that had to happen. That's ultimately not what it came to, but just wanted to give a shout out to those folks. That was really something. That's very admirable. I'm glad to see that. It's always wonderful to hear. Do you know offhand quickly I know this has just happened what the item was that we couldn't move past? Was it with

Starting point is 00:56:16 regards to basically the president and issuing tokens, or was it something else? It wasn't particularly close. So I don't know that it was one specific item. I mean, it was a combination of the ethics provisions, the BRCA, and some of the agriculture stuff, just like we spoke about. And Democrats are largely talking about all three of those things. If you had to point your finger at one of them that was the biggest, it was Alex. All right. Well, we had that almost anticlimactic. It died on the vine. It's over. over. Zach, where can everybody go to follow you? Check out your work. What do you want to direct him towards? I met Zach B-Shapiro on Twitter. Z-A-C-K-B as in boy, S-H-A-P-I-R-O. You know, that's probably the

Starting point is 00:56:57 easiest place to find me. If you enjoy this episode with Zach Shapiro, do me if you ever hit that like button. Really does help us out. And check out the previous episode with Brent Johnson or the recent live stream.

BTC Sessions - Big Tech Gets AI. BlackRock Gets Bitcoin. You Get Prosecuted | Zack Shapiro

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.