How KelpDAO's Exploit Left Aave With $200M Bad Debt

Transcript

Hi everyone, welcome to another episode of Bits and Bips the interview. The show where we explore how crypto and macro are colliding one basis point at a time. My name is Steve Ehrlich. I'm the head of research at Sharplink and I'm also your host. >> [music] >> Uh we've got a lot to to talk about today. Um we're going to focus on the um the massive hack that impacted um KelpDAO and and Aave. But before we do, uh just a quick disclaimer, nothing that you're going to hear on today's show um should be considered financial or investment advice. For full disclosures, please see unchainedcrypto.com/bitsandbips.

And um finally, before before we we begin, let's just take a brief moment to hear from some of the sponsors who make the show possible. If you've been loving Bits and Bips, don't [music] forget that the show is transitioning to its own feeds on X, YouTube, and your favorite podcast player. If you're not already subscribed to [music] Bits and Bips on its own channels, go there now and hit that subscribe button so you can keep up with our twice weekly live streams and [music] macro meets crypto breakdowns. Bits and Bips will only be on the Unchained feed for a few more weeks. So [music] subscribe today to be ready for launch. You can get all the links at unchainedcrypto.com/bitsandbips. >> [music] >> All right, welcome back. So all of us have been sort of pondering the implications of the the massive hack that impacted uh KelpDAO and led to

almost uh $200 million in ETH being borrowed uh I guess like um being taken out of of Aave um over the past couple of days. It's led to uh a tidal wave of outflows. Uh over $8 billion worth of assets have left the the DeFi blue chip and the industry is really pondering the implications not only for how uh this is going to be resolved, who's going to be left holding the bag, but then also what does this mean for for DeFi in general, what does it mean to be trustless and decentralized, and how can sort of DeFi recover and um and move forward in a world where we're going to see more stratified institutions coming online, RWAs coming online, and expectations of just massive growth in the industry as

stable coins continue to to expand. A lot of a lot has been written about this hack and trying to distill exactly what happened. Blockworks Research published one of the better or one of the most rigorous postmortems that I've seen so far. Kind of goes beyond just the mechanics of what happened, but also raises some important questions and and answers some of the thoughts that are running through my mind and I'm sure plenty of yours as to how the industry, again, is going to move forward. And it also raises some hard questions that I think we're still waiting to get answers to. So, I'd like to bring in all three authors of that report so that we can explore it in more detail.

First, we have Luke Lesher, the head of research at Blockworks. So, welcome, Luke. We have Shandra Devans, research research analyst at Blockworks. And then last but not least, we have Carlos Gonzalez-Campo, another researcher at Blockworks. Hey, happy to be here. >> [snorts] >> So, so welcome, guys. Shandra, let's just start with you. I know our audience is pretty sophisticated. Many of them are on top of what's been going on, but for anyone who has not been up-to-date on the details, can you just briefly explain what happened, why it was such a sophisticated attack, and if there's any

sort of like last-minute updates or or news events that have transpired today, I'd love to get those as well. Yeah, sure. Thanks so much for the intro. Our research was primarily focused on how the exploit affected Aave but just for some context of what the exploit actually was, there was an attacker who signed a malicious cross-chain transaction exploiting layer zero's DVN network. And that malicious transaction claimed that 116,000 rETH was burned on the L2, which allowed that equivalent amount to be minted on the Ethereum mainnet.

The attacker then took those funds, split them into various wallets, deposited them into Aave V3 as collateral, and as an exit strategy borrowed a 93% LTV against that position, leaving the protocol with around 193 million in debt taken away. Um so, following this, there was a lot of people withdrawing funds from their wrapped ETH position. Uh so, we saw the market go into around 100% utilization. Borrowers could not take their collateral back anymore. And instead, as a kind of synthetic form of exit, they borrowed stablecoins against their wrapped ETH position. Uh so, now the situation that we have is there has

been ETH withdrawn from Aave. Uh wrapped ETH markets and stablecoin markets are at 100% utilization. And normally, the dynamics of a lending platform would be that with high utilization, you have higher interest rates to incentivize borrowers to step in and provide liquidity and for lenders to pay back their loan. But just due to the uncertainty of how this bad debt is going to be handled, we're kind of stuck at a standstill. Um 100% utilization, lenders cannot withdraw their funds. Yeah, it almost seems like one of those circumstances where um price like interest rates just didn't matter um because it it almost felt like a a binary outcome. Either you get your money out or you don't. You don't even care what what price.

Uh again, I know your report focused more on the Aave side of things, but I I do want to kind of get your thoughts and and um I and anyone else uh like Luke or Carlos like I'm not an expert in bridges. I'm not an especially technical person. But it seems like the the setup, and I know this has been a point of contention um um between uh Kelp and Layer Zero, it seems like it was a one-of-one type of of signature scheme that was necessary to create that that sort of fraudulent transaction. Um I know from what I've heard, it seems like Layer Zero says that's sort of like the default baseline, although they recommend um something a more sophisticated type of multi-sig.

Um KelpDAO obviously went with with baseline. It seems highly dangerous to put that much money um in the hands of essentially one one signature. Uh it's it's funny. I don't know if any of you guys have ever watched the movie of the show Seinfeld, and I know I'm a probably a bit older than you are, but there's one famous episode where George Costanza is trying to break up with his girlfriend, and even that required a two-of-two signature cuz he tried to he tried to break up with her and she said no. And they were joking that it's it's um it shouldn't be as complicated as launching a nuclear missile on a submarine where both people have to turn their keys. Um but it seems like it was much simpler here to I mean, not simple in the sense that it was simple to implant the fraudulent messages, but and

since you really only had to compromise one um I guess one signer. So I'm just curious your thoughts on that. Yeah, I'd say um there's certainly some fault to be shared across the board here. You know, Kelp with over a billion in client deposits um can do better than a one-of-one, and certainly we know at this point a one-of-one is pretty frowned upon. Um then of course Layer Zero uh who had their their off-chain RPC infrastructure compromised, um perhaps they shouldn't allow the one-of-one to be the out-of-the-box kit um for new deployers on the OFT

standard, and perhaps push back more. Um I think the numbers show close to 40% of those unique OFT deployments are on that that one of one. Really? So you're saying that 40% of people that use this infrastructure still have that one of one? Not of total capital, but of um of unique OFT deployments and a number of apps have since been upgraded as well following the exploit and >> Hopefully. we saw a lot of um unique uh asset issuers who run the layer zero FT, they had to freeze their infrastructure as well to evaluate any any compromises on their end. Um and then third, downstream um for Aave, when

listing the the RS ETH as collateral, um perhaps of course that the one of one uh signer on the OFT deployment, um that could have been flag flagged as a risk for the collateral listing and parameters adjusted accordingly um for that. Yeah, you would think it I mean I I know we're going to I think in this conversation we'll get to things like like circuit breakers or rate limiting, um but uh yeah, all those things um certainly could have could have helped. Um so Shondra, back back to you. Um want to focus on on Aave and and there's different levels of um I guess knowledge about how these DeFi platforms operate. I mean, to a certain baseline

like how they function, but then also how they're used. Uh and your report does break out um some delineations. I mean, Aave um sort of their uh implementation on on mainnet Aave version three, which is which is the main pool, the blue chip pool, that sort of like commingles all all deposits and that's part of what makes it a very powerful um platform for uh for DeFi participants to use, but then it also um kind of socializes this risk that um the extent of which was was unknown. Uh love for you to just explain the design choices that went into that, but uh yeah, let's just start with that. I mean why did they choose that setup? And then I have a follow-up.

Sure, so this kind of ties into modular versus monolithic lending. And it also ties into a very interesting point that we made about looping and its impact on the severity of the attack. So, with protocols like this, you have two sides, right? You have the lenders that are supplying their collateral, and that gives them borrow power to allow them to borrow assets against that collateral. But, in this [clears throat] collateral pool, assets are all pooled together, and there's no distinction between what assets are allowed to be borrowed against it or different rates of interest rate. So, normally under lending setup, you would expect if I were to deposit ETH in the protocol, you expect a very diverse set of lenders

using different collateral at different loan-to-value ratios to borrow against that, right? I might have one fund that has deposited Bitcoin borrowing my ETH. Some users depositing stable coins. But, when we look at the actual user behavior on these lending platforms, and we think why would someone borrow ETH as an asset, there's only one real reason, which is to go short that asset, right? To borrow it and then sell it on the market. So, we don't really see much organic behavior borrowing ETH. We saw We see a lot of people using it as collateral, depositing to borrow USDC, but we don't really see much organic borrow demand.

And an example we can give here is wrapped BTC, which is another market. That has a utilization rate of only around 4.8% at the moment, and it's extremely cheap to borrow, around 40 basis points per year. What we're seeing instead, that type of behavior, is people taking these pooled reserves, right? Because ETH can be borrowed, and there's no distinction between what collateral types are used, and we see a lot of leverage looping, which is activity of a user using an active staked version of ETH that generates yield to borrow a passive wrapped ETH version of it, converting that back into the active asset that generates yield, and looping that back and forth. So, that is basically the

primarily use case of using this wrapped ETH in the pool reserves. And when we looked a little bit deeper, we saw that around 98% of collateral backing these positions the wrapped ETH that was borrowed, was in these LSTs. Yeah. Yeah, it's it's interesting. I mean, the looping idea is something I've been familiar with for years. Anyone who actively follows crypto is is aware because it's just a very simple way to to get leverage. And and when things are are going well and the the numbers add up, it's it's a sort of a a market neutral way of of juicing returns. But it can be dangerous. I mean, it for I mean, for one thing uh I don't again, I don't know how much people understand the way these markets

work, but just because when you borrow money, my understanding from a platform like Aave, you borrow at a certain interest rate, but it's not fixed. It's variable and and it can change. People can get caught on the wrong side of it. And then it leads to rapid deleveraging and and real problems getting out of these leveraged positions. But then also, I wonder too, I mean, this looping do you think it constitutes any real economic activity or is it just a way of getting excess excess leverage? And to what extent is that sort of just what I'm looking for? Like war like incentives and and sort of risk risk guardrails.

I mean, I guess it would depend on your definition of real economic real economic activity. I would say it is real economic activity, right? Because without this looping demand, ETH would have no utilization. The looping traders are borrowing ETH, which increases the utilization. It also increases the interest rate for the lenders. So, when you deposit ETH onto the platform, I think you would have historically earned around 1.7% APY versus the example of Bitcoin that I gave where you earn 0.01 because no one's borrowing it. So, it is generating economic activity, a lot of demand, a lot of use from lending platforms is this looping.

Um the question is just does that kind of obscure the risk that you're taking on as a lender when you believe you're depositing into the diversified pool when in reality 98% of it is just running this trade. But, I would say that it does constitute real activity. Fair enough. And I know in your report you also pointed out that um the majority of of the wrapped ETH debtors are concentrated in like the top 10. Um did you have any sense of of who those those people are? Um and and I guess like yeah, do you do you know who they are?

Yeah, so we we did take those wallets and we just ran them through Arkham to see what they were labeled as. And these wallets are essentially vaults, right? If you think about what leverage looping is, you could do it manually where you take the LST, borrow ETH, convert that into LST and and do that loop again. But, a lot of people are just doing this automatically through vault programs. So, for example, the top um wallet is Etherfi's vault product, right? So, that basically just abstracts all the looping in the background and you basically present it to the user as high yield ETH product.

Mhm. Okay. And um I am curious too. I mean before we move on to um to Luke, uh you did contrast like this setup with with Morpho, which is which is a bit more a bit more segmented. Can you maybe expand on that a a little bit more and um sort of think about the design choices that might impact the way DeFi looks in the future. Yeah, exactly. I I think we'll get a little bit more into this in the end, but the meat of it is that you have a pooled option, which is Aave, where each asset can be borrowed, and you're not choosing what assets you can be lent again.

Right? So, if I'm depositing ETH in the pool, any user with a health factor who has deposited collateral can borrow that ETH. Uh I cannot say, for example, I'm not comfortable with a looper borrowing my ETH. I want to make that available only to someone with BTC or USDC. On Morpho, you have this extra layer of customizability, where you are able to choose what assets are allowed to borrow your collateral. But, the reason that we saw this do so well on Aave versus Morpho, like the looping trade has been very consistent on Aave, is just because when you have this one risk parameter that's shared to the entire pool, you're essentially forcing each ETH lender to opt into the system.

Now, you have billions of dollars of ETH TVL that's compounded over the years, and the risk team has just made that available to be borrowed at a low interest rate for loopers. Right? So, you have a lot more liquidity and lower interest rates on Aave, which is why the trade has just um flourished over there. Yeah, and it it's interesting because it's always that debate between sort of like security and and then also, I mean, this isn't the same thing as cross-margining, but it's the same concept in in my mind. I mean, on my show last week, we were talking about how uh some people were sort of like leveraging some of the market neutral like basis trades on commodities on platforms like like Hyperliquid um because there were so many mismatches they could exploit that don't exist now

in, I guess, quote-unquote traditional crypto. But, they were limited in how far they could deploy because they couldn't cross-margin. And um they they couldn't pool their capital together. It's the same concept here. It's always that that give-and-take. All right, Luke. So, let's let's turn to you. I want to get into Aave's response to all this because they they acted, I guess, somewhat quickly. They were um they froze RS ETH reserves. They um they also sort of cut certain parameters to to keep things in place. I mean, what what did they what did they do? Um how quickly would you say that they acted?

What did the response time look like? Yeah, so there are definitely a number of different steps in their crisis response process. Uh the first one was freezing the RS ETH market, um which occurred 85 minutes after the exploit. Um all things considered, I I I think that's fair time for the freeze. Uh you need to coordinate, I believe, at a five of nine multi-sig. Um and also the the exploit path made it um I guess difficult to catch at at the mint event cuz the attacker didn't need to manipulate an oracle or an AMM at all. Like, they minted the unbacked RS

ETH through the Layer Zero uh bridge, but kept the Kelps LRT oracle on uh the native mint contract kept reporting the the pre-exploit rate. So, because of the exploit mint. I was just going to I mean, just for again, people that are not entirely up or haven't in been in the weeds on this. I mean, basically, what you're saying is that um Aave's systems were not breached in any way, and there there was no, I guess, realistic um expectation of Aave knowing that the RS ETH that was minted on mainnet um was uncollateralized.

Yeah, correct. The the exploiter was able to deposit that collateral and take out the loan against it within minutes. And by the time that loan is originated, the loss is pretty much already crystallized. So, freezing the market prevented new exposure, but of course that the amount of bad debt that's defining this event, they would need to respond within seconds to have been able to prevent that.

Yeah. And I mean I guess it's maybe it's a this is a silly question, but I mean new exposure as far as you can tell, um they were able to essentially borrow against all of the fraudulent um RSETH that they minted. I guess any new exposure would have come if they tried to do the whole exploit again and and had to mint more, correct? Correct. Yeah, and I believe the um the during the exploit, the attacker tried to run a subsequent exploit transaction that the the second one was blocked um by the DVN, but the first one went through successfully and and after that uh the exploit was good to go.

Okay. So, let's talk a um let's kind of expand and and and talk a little bit about the bigger picture here uh because there's a lot of moving parts. There's a lot of people that are at risk to to varying levels. Um I mean, right now Aave is uh I guess short almost $200 of of of wrapped ETH. And the real question is going to be who's going to who's going to end up holding the bag? Who's going to have to pay for it because it has to be made up somewhere. Um from what I can tell, kind of ironically, KelpDAO is sort of the kingmaker. They can sort of choose to socialize the losses across um all holders um across like um L2 um um, participants on Aave as well

as as well as mainnet ones. Or they can choose to sort of look, uh, localize or limit the losses to, uh, to the L2 and make all of the mainnet holders, um, whole. And, uh, it's kind of there's some really interesting dynamics here because on one hand, like maybe it would make sense, yes, let's make all of the mainnet people whole. They did not realize that they were taking on this risk and that's a separate question as to whether or not they should have been made aware and and by who. But, I know in your report and and just some of the other, um, reporting I've seen on this, if they choose if they go that route and make the the L2 holders take a bigger haircut, that could actually lead to

more damaging economics because of the unwinding of the looping trade that would happen because of the full utilization of those pools. So, could could you please just kind of lay out the scenario right now like where where we stand and the different trade-offs of these choices? Yeah, absolutely. So, so resolution of the liabilities across the board is still pretty indeterminate right now. Um, like external financing solutions can come into play. There's some chance that attacker funds can get, um, frozen or reclaimed and, uh, attributed to affected parties, um, perhaps like what we saw with the Arbitrum freeze.

Um, but right now Kelp kind of has this decision, um, that Llama Risk laid out in their their governance post on the matter where they can basically, uh, socialize the losses across all holders of rETH across all chains, um, or they could choose to just isolate the losses to the layer two instance. And what you see in scenario one is, um, if they socialize the losses across all REs ETH holders. That would look like about a a 17% markdown.

Um and a huge share of that that collateral exists on Ethereum mainnet. And there's still some LUSD equity um that can offer a cushion to take those losses. Um but also a 17% markdown is much less than what you see in um the isolated L2 instance where certain markets could take a much more material markdown and um completely blow through any equity cushion and lenders would take a much bigger impairment.

What we also have here is uh the umbrella module, which uh is constructed to be junior capital um to take first loss in the in the event of um uh bad debt. And this is the first kind of I just want to I mean again just to clarify for readers uh and the umbrella um I uh I guess component is sort of Aave's rainy day fund. It it's supposed to be the backstop when when something like this happens. But um from what I can tell and I think what you're about to explain is that it's insufficient to cover the the debt in in this particular case.

Yeah, so I guess once once LUSD's equity cushion is exhausted um as as opposed to impairing lenders, the umbrella module exists to cover that bad debt. So, lenders can basically um stake to the umbrella module to receive an additional risk premium. That's historically been about 2 to 4% uh depending on the asset type. Um and they put up that capital to take loss to cover uh any bad debt. And this module has attracted about 370 million in deposits. It was sitting at its target of 250 million

going into the exploit. And this seemed like a really nice cushion in the context of the historical bad debt Ave has has realized, which was less than 5 million throughout its history. So, of course, the 250 million pile seems adequate. But what we see now is it's quite undercapitalized to cover the left tail scenario of a large exploit in an under underlying collateral listing.

And it's quite likely that in the event of scenario one and losses are socialized across all RS ETH holders, the the close to 50 million in the ETH umbrella module will likely get wiped out. That being said, the you would reduce the impairment on ETH lenders from about 1.5% down to down to 0.6% but still inadequate to cover all the potential losses that are in play right now.

And these losses come from the fact that sort of the looping positions that Ashandra pointed out, interest rates are surging, their positions are increasingly underwater. They have to kind of get out of these trades as quickly as possible to pay back their debts over and over again and there's just no way to do it. So, despite the fact that this number seems bigger than the total nominal size of the loss, the actual losses to traders that would have to be made up from this bad debt are will be orders of magnitude bigger than than the amount that's in this umbrella module.

Is that Am I interpreting that correctly? Yeah, not quite orders of of magnitude, um but cer- certainly insufficient on the margin. So, so most of this bad debt will come from the rETH markdown, um depending on which scenario uh Kelp chooses to go with, but back to Shonda's point about um the scale of the LST and LRT uh looping that's going on. Um historically those were positive carry trades uh call it there's 50 basis points, 1% positive carry, and traders would leverage that 5 to 10x,

whatever it might be, but following the exploit and the freeze of the rETH market, and utilization on ETH went to 100%. Uh interest rates on the borrow leg of that strategy went to 10%. So, what was positive carry went to perhaps -7% napkin math, and that's, you know, leveraged multiple times. So, there there was the risk that looper equity, uh even on positions that weren't exposed to rETH, um looper equity would get wiped out, health factors would crater, and more liquidation flow I would come to the ETH market. Yeah. It's um I I was I mean, everyone's probably thinking about

historical precedents um in crypto and otherwise. Uh it reminds me a lot of sort of like what happened with the the LTCM back in the '90s, I think when they were got caught on the wrong side of what was it like the Asian financial crisis or or something like that, cuz they just got they levered up on what was it Treasuries, and uh and eventually they uh just got caught on the wrong wrong side of it. Um okay. And and then one other component, again, I want you to to um to walk us through, because it kind of gets a little bit into the weeds, but it is relevant in terms of how um some traders are going to uh observe losses and try to get out of their positions is that depending on what happens there's going to be a haircut on essentially Aave's deposit tokens for for wrapped

ETH. A Aave and uh um and they have a deposit version. They have that for every single token that they offer but but the the Aave ones are the ones that are of note here. So can you just briefly explain um how those work and how they factor into all of this? Yeah, so the ETH deposit basically functions as a loan. Points in time it looks like a demand deposit you can take back share of your principal. You're being streamed the yield and it's considerably liquid. And the the value of that loan position would track near par value of one ETH. And you know, lenders they expect a degree of liquidity under normal market

conditions. But when utilization goes to 100% there's basically no available liquidity for lenders to withdraw. And while they were being compensated maybe 10% at the first signs of stress um but what governance did in the or sorry, what Llama Risk did in the um initial crisis response was they lowered that ceiling rate down to uh close close to 3%. So basically lenders in the ETH market were facing this risk of impairment um on loans that had gone completely illiquid with indefinite duration no

defined maturity and the risk premium they should be receiving under conditions like that they were not being compensated for um their yield had gone down to about 3%. So what what's happening now is you know, there's there's the risk to impairment on these loans from the bad debt and the rates aren't elevated enough to incentivize new suppliers to come into the market and harvest the risk premium. So, 100% utilization is is basically just a symptom that this market isn't clearing.

Okay. All right. Um All right. So, we've kind of set the table of of of what happened. Uh we're going to take a very quick break so we can hear from some some additional sponsors and then we're going to uh come back and talk about what's likely to happen next and what this means in the in broader terms for crypto and DeFi. If you've been loving Bits and Bobs, [music] don't forget that the show is transitioning to its own feeds on X, YouTube, and your favorite podcast player. If you're not already subscribed [music] to Bits and Bobs on its own channels, go there now and hit that subscribe button so you can keep up with our twice weekly live streams and macro meets crypto breakdowns. Bits and Bobs [music] will only be on the Unchained feed for a few more weeks. So, subscribe today to be ready for launch. You can

get all the links at unchainedcrypto.com/bitsandbobs. All right. Welcome back. Um so, Carlos, let's turn it to you. Um there's a lot of big questions that that need to be answered and we're going to get to that but but before we do, I mean, I'd like for you to just give me your sense of of it's probably hard to predict exactly what's going to happen but what are the different avenues available to a resolution? Um talked with Luke about I guess some big decisions that Celcius will have to make in terms of who's going to take the haircuts but from Obvious' point of view, I I would imagine they're talking to all their different partners. I mean, they're

looking to find capital backstops. Um maybe um maybe raise uh additional funds, uh bring in other market makers to try to help this those markets clear and and get them started again. Uh like I'm sure Aave's very busy, but if you were part of Aave, what would you be doing right now? Yeah, for sure. I think they would be talking to all the partners and all the parties involved uh in this incident to try to come to the best resolution that uh affects the least amount of parties or like gets I think their top priority right now is just uh have made the users whole on the on the lending protocol.

And I'm confident that they will be able to do that. However, it does set a precedent in the sense that uh what was the risk management going into this incident? And are you better off like having external risk graders or external risk um um contributors or just grading that risk yourself? And I think that's one of the questions that will be going into like in the coming months and weeks. Yeah, it's kind of interesting. I mean, we can't talk about this hack um without mentioning the Drift hack that that happened What was it? Like a a week or two ago? Um I know I think Tether stepped in to try to help recapitalize Drift um to a certain degree, and um

that led to some interesting dynamics vis-à-vis Circle, which we'll actually get to in a in a minute. But uh I am I'm I'm I'm wondering if there's going to be sort of a white knight in here to to save the day or um perhaps are they going to try to issue issue more tokens or change fee structures to um provide additional rewards to uh peo- people who stay with them? I mean, this goes all the way back to I think to the Bitfinex hack from like even 2016 where they essentially minted an IOU token that they bought back at a certain period of time um despite uh I think at which was a questionable legality for lack of better term. So, I mean, could you expand perhaps just a a little bit further and um the kind of what you think or Yeah,

please. >> Yeah, I don't I don't think we'll see a new like token to make like users whole. Perhaps Lucas better like suggestions on this, but I do not think we'll come to a resolution of that. Uh sort of like nature. I do think an external party may come in and make everyone whole, but I don't think we'll we'll see a Bifi next like resolution where you see a new token. Okay. Um and I mean this might seem like again a silly question, but do the three of you have any doubt that um that this wasn't Lazarus or is it pretty much a 100% certainty from your perspectives?

Okay. Uh I outsource my on-chain forensic work to a lot of people on the timeline and right now it seems like consensus is very much that our friends in North Korea. Yeah. Yeah, our friends in North Korea. Okay. So I mean what do you think this means in the bigger picture? I mean we have we have this, we have Drift. I mean over 500 million dollars has been stolen. Uh I would imagine they have armies of people kicking the tires on every smart contract there is out there. Um do you think this is the end or or do you think we're going to be be seeing more?

Uh I mean from their perspective hitting these two exploits back-to-back I mean and they're very different. Um they must know that everyone in crypto is now going to be testing all their systems and trying to figure out the best ways they can to to rate limit and and protect themselves. Even on the the assumption that perhaps I think best cyber security hygiene is to assume you will get hacked at some point. So you need to find a way to mitigate it. Um like is this the opening salvo or is this the end? What what do you guys think? Um, I think there are two things here, right? So, first is the risk management side on their RS ETH contagion. So, Aave scaled to $75 billion in deposits

essentially using the same unified pool architecture that was proposed by Compound back in 2018. Mhm. I would say that there is nothing inherently wrong with the unified pool model. Uh, the model has been incredibly successful. It bootstrapped deep liquidity. It enabled permissionless borrowing borrowing at scale. The problem is when protocols start moving down risk or enter more more complex collateral. Uh, so unified pools work best when the collateral set is restricted to a small number of high-quality continuously priced assets such as BTC, ETH, USDC, and USDT where the primary risk are essentially just volatility and liquidity that can be

modeled with uh, loan-to-values or liquidation thresholds, right? Uh, Aave Spark is actually a great example of this where they do have a unified pool model similarly to Aave, but they have very conservative approach to risk management and a tighter collateral set. So, they delisted RS ETH back in January and they have seen almost $2 billion in inflows since the Curve incident. So, on the one hand, you do have like the risk management that can prevent this sort of contagion. And to your specific question, are are we in a period where attack surface of DeFi has meaningfully like expanded? I do think that as systems become more composable and fragmented, the attack surface expands.

Uh, what's funny or like sad to me is that uh, I don't I I think we continue to see the same attack vectors that we saw 5 years ago. So, it's all bridges, it's always oracles, RPC infrastructure, risk-taking wrappers, etc. So, from that point of view, I think the industry can do a much better job at learning from past mistakes and actually taking security seriously. And the second factor is that I think AI likely accelerates both sides of this in the sense that attackers can automate vulnerability discovery and exploit simulation much faster. But, also defenders can harden systems faster and improve monitoring. So, in my opinion, the most competent teams will remain

security first and have largely avoided these issues. And an important nuance here, I think, is that Aave itself was not exploited. So, this was the contagion from the collateral impairment in one market, not a smart contract exploit on Aave itself. So, that's where I say that the distinction matters between the risk management itself and also the security on the other hand. Yeah, it's it's it's really kind of this conversation again to to expand it a little further out. I mean, in a different light in the world where we now have I guess it's called Mythos that that that new tool from Anthropic that they're only rolling out to a few selected trusted parties. Although, I guess there was a report in Bloomberg

that some unauthorized contractor has already gotten access to it. And and what does that mean for finding zero-day bugs and how quickly can they be stopped? But, it also was kind of fascinating to me that like um both of these hacks um I guess I mean the Drift hack was more almost like a very complex social engineering hack. I mean, this one perhaps came due to compromises over convenience versus security and and sort of on chain adoption. But, I mean, for the most part, I mean, when I look at cyber security events, not even just related to crypto, it really comes down a lot to hacking people. Sometimes it's easier to hack a person than than in system. And

like I'll hear all the time like I'll get questions even from my friends and family about all the cybersecurity stuff and I said like worry about like 2FA, your [clears throat] email account, and not falling for a phishing attack. Like don't don't worry about some zero-day threat that's going to wipe out your bank account. Like there's there's more acute things to to focus on. Uh and just related to Sparklend cuz you do point it out in the report. I mean, they made the decision to remove RSETH uh back in January. Do you think that they were lucky or or was there a real reason why why they did it? Because uh it it certainly seems to be exceedingly good timing.

Um it's probably a combination of both. They removed RSETH as part of a broader cleanup of low usage markets. Uh but in my opinion, even if RSETH had remained listed in the exchange in the in the lending protocol, Spark's losses would probably have been much smaller because of rate-limited supply caps and borrow caps. Uh so even if RSETH had remained on the platform, uh Spark has a mechanism to limit how quickly users can build exposure.

Uh so that have that would have prevented losses of the magnitude uh that we saw on Aave. Uh obviously, this is a hypothetical and we didn't see it play out, uh but that is my view of the risk mechanism. >> [snorts] >> Okay. Um so >> [sighs] >> yeah, I mean, we just have a couple minutes left. I I mean, I want to just kind of broaden it to to the three of you and um maybe you could just give me some some final thoughts because this is still a a dynamic uh situation and a lot still needs to be resolved.

And and this is happening during um some some really important times for crypto. So uh here here I guess are two quick questions I have. Um one, what does this mean for what decentralization means uh when it comes to crypto? Uh and we talked about the our Arbitrum decision to freeze I think it was like 30 30 million ETH on um on their network. Uh Luke we discussed Aave stepping in uh a security council and I know it was like a five of nine multi-sig decision but but there still was a like a security council that was able to make decisions that affected everyone in a quote-unquote decentralized uh environment. So like what and this is all happening again and going back to

the Drift attack where Circle came under um massive criticism for deciding not to freeze the um the USDC that Lazarus Lazarus Group took from from Drift. So what what does I'd like maybe maybe Luke I'll I'll go to you for this. Like what does that mean for what does decentralization mean to you in light of all of this? Yeah, I think um we're we're kind of coming to the realization as as an industry that like cer- certain functions particularly like risk management and crisis response should be perhaps responsibilities that are gatekept to a select few of highly

competent and responsible parties that are also easily identifiable and then in turn liable um in the event of an exploit or a mishap like we see like we've seen recently. So there there are certain things in in Aave's governance that of course require uh DAO votes and whatnot. Um but at large I think the the surface area of parties involved is going to continue to contract closer towards centralization and and more trusted intermediaries with perhaps more

admin control on on particular parameters as opposed to multiple parties having responsibility at the full life cycle of how an asset might be used on chain. So yeah, increasingly looks um more like a a CBDC approach with with trusted intermediaries that are responsible and liable. Yeah, and that leads to this meta question all the time like is decentralization centralization binary or is it a spectrum? I know a lot of people will discuss it as a as a spectrum as long as certain core components are met but even that can can be up for debate and

and another just adding in another variable to all this is negotiations for the Clarity Act because a lot of it's going to turn on how DeFi is is being treated and the SEC is doing its best right now under Paul Atkins to sort of exempt DeFi issuers, DeFi creators from having to register as as as brokers and and and sort of go through all of that red tape that like the Charles Schwabs of the world have to go through but instances like this happen and you see like a group of people known or unknown making these unilateral decisions that can make it can make it more difficult I think in some degrees to to accept that these are truly

decentralized platforms or just requires some more sophistication in delineating between the two. Just one one last question again because a lot of people listening here I mean they're probably Ave customers they probably or own the tokens or are engaged in DeFi somehow. Again, none of this is financial or or investment advice but what should people do? They they just saw Drift, they just saw Ave, they have tokens that perhaps are are at risk. Markets are freezing up. What are one or two things that people should do to try and protect themselves?

I'll leave it to whichever one of you would like to share. I mean, I guess this this enters into an existential question for DFI in the sense of how can you justify depositing into DFI protocols when sometimes the rate are not even competitive with treasury yields yet you're on the risk of your principal going to zero overnight. So that's one part of the equation. I think the other is that we'll increasingly see vertically integrated platforms that manage risk end to end as we saw in Luke's Luke's call out. I think Maple is one example that comes to mind where they on they distribute the products, they

underwrite the risk and they have a more traditional credit structure when they have legal enforceability and there's a clear accountability, right? So in that case, you don't have the situation as in the Cal battle drama where everyone is just pointing fingers at each other but rather you have a clear accountability and legal recourse path that you can take for resolution. I think we'll see a push into that from one from from the institutional side of things. I think from the retail side of things, if you still prioritize permissionless access, you'll see the most battle tested just seeing flows.

One example that comes to mind is Sky which is perhaps one of the protocols that deserves a bit more recognition and it's oddly under discussed or under covered but it has been live like longer than any other DFI protocol like in existence today and had a situation like this. Yeah. So I think what what we'll see is just a concentration towards the battle tested platforms. Actually, I think my very first story I wrote for Forbes back in 2017 was on Maker and sort of breaking down injuries and who are what's his decision to invest I don't know at the time maybe like $16 million in a in a stable coin protocol and I had to explain what that what that

was and uh um why it made sense to invest in something with the word stable in it in in a world of crypto which was so volatile. Um we'll have to leave it there. Um thanks to Luke, Carlos, and Shonda for walking us through the Ave math and and basically um where things stand. Um that's it for Bits N' Bips the interview but stay right here because Lars is up next with Miguel Morel of Arkham to trace exactly where the stolen RS ETH has moved and whether the contagion is actually [music] contained. And then after that excuse me, she's going to talk with Griff Green of the Arbitrum Security Council on its decision to freeze $70 million in stolen funds. So, um stick around.

>> [music]