Facts and Figures about Web3 Security: Where Do We Stand?  Davide Carbon

founder of two uh blockchain companies and one is a unisco is escro as a service Bas it on Smart contracts and tomato is a token automation platform for corporates and then I'm I'm also the main blogger of smart contract tips which is an Italian blog uh focused on best practices in web three so you can find me also on LinkedIn so me years of experience too many years of experience in it and uh okay let's skip all this boring personal stuff and let's start on the on the topic uh web three probably we are still in a in a phase of migration from Web Two to web three and uh the let's say people is not you know there is not a mass adoption and this is not a secret and is very difficult for people to understand how to take this web three and to take the responsibility to become the real manager of their own assets instead of being let's say um uh using uh Services which are holding our assets and when I say assets not only money or you know tokens but also anything also identity content everything is managed by Web Two uh companies and then one conjecture that I like to do is uh this that probably is very difficult to to have something which is self custodial easy to use and also secure and uh is a conjecture so I don't have any proof and maybe you guys that you work in abstract accounts maybe you saying okay but we can break this uh trilemma uh well I'm pretty bullish with uh abstract accounts probably is too early to say that really we solve everything and the M adoption is just around the corner then uh one slide from vanina was web 3 is not safe right I started to work with smart contracts more or less in uh uh 2016 and uh was the it was the time for the notorious Haw of the The Da Haw probably the the mother of all the Haws and uh uh in in this essay the author David gar um found that at the time the smart contract code was really bad a lot of bucks every thousand lines of code uh more than Commercial Code and of course if compare it with uh commercial non Financial code and if compare with the Microsoft code or NASA code was a complete disaster so where do we stand today okay if we if we look back 2016 the D hack the vulnerable ability that was exploited was the reentrancy so basically you have a smart contract the smart contract is connecting to another smart contract to make something and this one is able to recall again your smart contract if there is some bag in your logic of course the external smart contract can siphon your funds and this are very known vulnerability you know but still in 2024 uh there are more than 1 million USD that were hacked because of this vulnerability in protocols so that means that yes we are getting better there is a lot of uh you know information uh sharing of knowledge everything but still reany is still uh affecting our protocols in the F and this is not very good you know then if we if we look at the you know uh hacks over time uh you see that probably we are getting better that 2022 was the anus rilis with more than uh 3 billion hacket uh and 23 was much better but as said by vanina 24 is you know is replicating the same figures of 23 so um another point to see is that we have more than 8 billion hackets out of a total locked value of 80 billion in the F which is more than 10% is okay is not acceptable what if you connect your bank and you put you know your savings and okay 10% is is gone is not acceptable for Mass adoption to have these numbers of course then if we look at the the size of ax which is an angle uh we can see that uh the the medium the medium size is around 5 six million except 2023 where there were a lot of small acts I don't know which is the reason for that but now in 2024 we are again having more or less the same situation of 4 million as a medium size for uh ax so and which are the top vectors with top vulnerabilities exploited but basically here unfortunately is not very clear for the resolution but so trust me the first one on top is private key hack and not only is the first with um number of diverse and unknown methods but it is also in the fifth position for fishing and social engineering so what is the Insight here that yes maybe we are talking about defi we are talking about the centralization but most of the hacks are caused by private key so is basically a side of centralization because that means that people in charge with or system in charge of managing funds have been hacked not protocol themselves because we are talking about private keys and imagine if Bitcoin Satoshi Nakamoto was hacked the private key yes probably would be like uh uh Financial crash for a while because there is 1 million Bitcoins associated with his private key but not at protocol level okay it would be recovered in here instead we have protocols which are completely depleted by the the loss of private keys and if we look at this diagram not only in 2024 buts uh all time the situation is not very different private key compromisation is still the first important cause of uh hack and money loss so when I talk of web 3 security to let's say normis or it managers not in web 3 in the web 3 sector I need to explain the difference between a smart contract uh let's say ha why smart contract different in terms of security with respect to a uh common uh it asset well in principle it is very interesting to consider that if you have uh things like data corruption in a normal traditional software the the time that you know the loss is proportional to the time that you need to recover the data assume that that you have a backup of course uh in smart contracts the the data corruption is the loss in the sense that if your smart contract is messing up with the numbers and is a token contract then okay the money is gone period another Point interesting is the ransomware of course in in the in a traditional software ransomware the loss is based on the pay to recover and Recovery is not even guaranteed in smart contracts probably is not applicable that way but the hacker can be convinced to return the money and to deduct like a bounty and this is kind something that happened in the past okay so going back to the idea of audits that were very well introduced by vanina something which is very important to understand is that audited doesn't equal secure because has the master of informatics here Dix said once with program testing and verification you can check you can prove the existence of bugs but not the absence of bugs and this very important and by the way if you look at the news you you will find a lot you would would find a lot of cases where a well audited protocol was hacked and a well audited by topnotch firm security firms was haed so it's not so uh let's say it's not enough and also because there is also a misunderstanding sometimes in audits who is really reading the audits here whoever read a audit in this room I did couple of times also I wrote a couple of times but the point is the the scope the scope is very important this is a case exemplary case where the auditor say okay yeah there is this function which is actionable by the company but it is not in the scope of our audit so who know maybe and this is an audit and uh of course it's completely out of the scope of the Auditors and in this case it was also AED then back bounties back bounties are okay if audits are the king of security in web 3 probably back bounties are like the queen and here I like to take number numbers when I when I can and this pretty interesting these are the total rewards paid okay paid not the maximum Bounty available but Total Rewards paid by top 10 projects in a platform like immuni so it it is only a platform it's not all the ecosystem but it's pretty interesting and these are the the figures and okay is a difficult to read but then there is like the Max is polygon with almost 6 million and the the medium is 85 keys and that means that in principle or better that means from a stat statistical point of you that's half of the projects which are visible in the platform here and we are talking about 67 records which decided to disclose the data because not all the projects are disclosing all the data about bounties being paid and uh is 85 Cas and that means that half of the projects are paying more than 85 so that means if you a startup and you want to deploy your protocol and you want to launch your Bounty you have to pay your Auditors then we are talking about you know money it's pretty expensive so there is always a tension between the budget that you have to develop and Market your product and also the budget that you have to use to secure your product and of course security is King yes but also you need money and I'm not saying that it must be overlooked but these are the figures so you can take your insights then another Trend which is let's say going on recently is the onchain monitoring that basically uh you can audit before you uh deploy the contract then you can also launch a bounty before but once the contract is launched and deployed and people is already using it then you can monitor and watching in real time what is happening there are a number of platform here I'm just citing fora but there are also other platforms and this is pretty interesting and now monitoring and posing is becoming like a new best practice this is a question mark because I you know I like to put questions rather than answers and this is a case very recent probably I don't remember 10 days ago or more or less and where the Pendle saved the 100 Millions because of this practice so they were watching uh transactions in real time they decided to pose the contract they took some actions and then they in some in this case they mitigated the uh risk um but here there is like you know there is a downside of course that means that there is a governance there is someone in charge to push the B Buton and posst the protocol and this is completely let's say contrary to the spirit of the centralization because we are talking imagine if Bitcoin I like to site Bitcoin in this case if there if there was like someone able to push a button to stop the protocol okay so um I know that we need to find like a tradeoff between decentralization and probably security and so so in this case we go back to the trilemma then I I want to complete to conclude talking about uh Layer Two security uh I like to say to people that uh if you don't pay the gas then you are the token this this is just a rephrasing of an old motto if you don't pay the product you are the product um by the way I have nothing against Layer Two I'm very bullish but it is very important to understand which are the limitation of the layers to that we are using I trust that all people in the room know this uh framework which is named stages where basically they work worked to Define uh a number of attributes to define the maturity level of a rollup so we are talking about rollups second layers um and uh this is very important because there are attributes like is the rollup provided with a fraud proof system yes no uh is the user allowed to opt out if there is an upgrade in the software yes no so there are a number of checks and uh the level of uh maturity is some somehow you can calculate yourself but there is a a site which is al2 bit where this is done and they I think they are also the somehow the authors of this U framework and by the way if you are curious uh the arbitrum rollup and optimism they are at the moment stage one so not completely decentralized conclusions so acts steadily happening 10% sorry there is a typo 10% of of funds affected auditing is not enough back bang is and realtime monitoring can be like a good uh extra extra goodie to to have more security but uh you have to make a trade-off between you know decentralization and uh and control with uh uh real time monitoring then uh regarding uh second layers security just follow stages and be aware of where you're putting your money and uh and then I I want to I like to conclude with this security is never a product security is a process so if you think that your protocol is secure because you buy an audit then you are completely wrong it's just like a stamp but security is a process security is about the culture that you have in your company or in your team and there is a continuous you know process of evaluating the risk taking actions and repeating and taking actions you have also to accept the you know the residual risk because it is impossible to have risk equal zero okay in it is impossible and by the way in web 3 is even more complicated this stuff because evaluating taking actions and repeating is something that you can apply easily in traditional software but what if you were talking about a smart contract that in principle you should be immutable then is not IM mutable then you can somehow have like a privileged access access to make some actions as administrators you know so and this is just to go back to my initial conector can we have a system which is completely secure completely decentralized and also usable question mark I don't have the answer thank you thank you so much any questions oh aming final someone vanina go ahead so one piece of Statistics that I recently noticed is that we you know hacks are on the rise that's there's no question about it but we're also getting better at recovering funds from hacks do you think that this is G giving projects the sense of false security that nah even if we get hacked we'll you know try and recover the funds and and how do you think we can keep recovering funds without giving that full sense of security thank you um well well uh yes I agree that is a false sense of security because you can like uh trigger a process where hackers or scammers they can start just creating this false cycle false virtual cycle of uh hacks and uh partial recoveries and so I don't know uh I think this very difficult to to um foresee any any kind of evolution of this and by the way uh just to mention another great person predictions are very hard especially about the future and then I don't have an answer either thank you unfortunately we don't have time uh for more questions we are already a bit ahead of the schedule but thank you so much David