DSS 2026 — The Audit Gap: A Four-Year Empirical Study of 23,818 Findings and 218 Exploits Christian Vari Source video: https://www.youtube.com/watch?v=96YwISrzBn4 Automatic captions — not human-reviewed. Complete spoken-audio coverage is unverified. Source language tag: en (English). Source export SHA256: 79703665f080ff2146d6846f3311edf45303998f5e406bfd9bf6eb1452894050 [0:01] Do you guys like data? Because we do. My [0:04] name is Christian Vari. I'm the head of [0:05] audit operation at Ox Security. And we [0:08] have prepared a report about the public [0:12] data that we have found about the [0:14] findings and exploit that have happened [0:16] in the last four years. [0:18] Uh very briefly, Ox Security is a [0:22] security firm that is working on [0:24] blockchain since 2017. We have done more [0:27] than 600 engagements. We have more than [0:30] 50 senior security researcher. [0:32] And [0:33] we actually [0:35] published this report that you can find [0:38] on our website and that I will introduce [0:40] here briefly. So actually we explored [0:43] all 27,312 [0:46] public audit findings from 22 security [0:49] firms and we [0:51] also checked another population in this [0:53] study. So the actual exploits that [0:55] happened. We checked all the live [0:58] ones from, you know, we we got the data [1:00] from rec.news and the total amount of [1:04] the losses is about 10 billion. But the [1:07] important things here are how actually [1:10] audits work and how [1:13] and what is the root cause of the [1:15] exploit losses. So as we can see here, [1:18] we have two different columns and on the [1:20] left we have the percentage of the with [1:23] the category of the audit findings and [1:25] we can see that the first one is logic [1:27] errors. Then we have code quality, then [1:30] we have access control and input [1:31] validation. But if we check actually [1:34] where the money is lost is very [1:36] different because we have private key [1:38] compromise, phishing and social [1:40] engineering, bridge exploits and also [1:43] access control. So the only thing that [1:44] we found in both of them is access [1:46] control. [1:48] The interesting thing is that the [1:50] remaining three categories that we have [1:51] for audit findings just it's just about [1:55] under 4% of the losses. [1:59] And another important is that [2:02] more than, you know, the 50 percent 51 [2:05] percent in our study of the losses is [2:07] about what we call human vector attack. [2:10] So private key compromises I said, but [2:12] in general all related to not to the [2:14] code quality, your code base and these [2:16] kind of things, but to operational [2:17] security. And you know, [2:19] just last week we have we have [2:21] experienced another another thing like [2:23] that that will that will be accounted in [2:25] the next version of the repo. We update [2:27] it every quarter. [2:29] Another important thing that we that we [2:31] found is that obviously there is a big [2:33] domination of [2:36] big incidents. So we we see here that [2:38] the largest incident is 15 percent of [2:41] the world losses and the largest 20 [2:43] incidents account for 66 percent of the [2:46] world losses. So actually there is a big [2:48] domination of few incidents. [2:52] So here [2:54] this is what we learn about that and [2:56] what we would like you to think about as [2:58] you know protocol owners and people that [3:01] manage protocols and their security. So [3:03] you probably go through an audit and you [3:06] got your design reviewed, the [3:08] architecture, maybe you have done [3:10] fuzzing, all this kind of stuff that is [3:12] very very important for your code base. [3:14] But also [3:16] we need to work a lot more on the [3:18] operational side of this and also on how [3:21] to respond to that. So having really [3:23] clear who have the keys, who are the [3:25] operators that can do this kind of [3:27] things, having clear workflow and [3:29] processes for that and also what to do [3:31] when something happened because thing [3:33] will happen. [3:34] And I have, you know, I I will leave you [3:37] with a question to all the, you know, [3:39] protocol owners and people that work on [3:41] that. So is which part to a catastrophic [3:44] event is outside the scope of your [3:47] audit. So you have been audited, that's [3:49] great, but is there a part outside of [3:51] that that still can lead a catastrophic [3:54] loss? If yes, probably you should think [3:56] about that and do a threat modeling for [3:59] that and work more on the operational [4:01] security in your company. [4:04] I leave you with this QR code. You can [4:06] scan this actually and you can go or you [4:09] can reach our website at [4:10] research.iotsecurity.io [4:12] where you can check the [4:15] the full report. We also have [4:16] dashboards, interactive, all these kind [4:18] of things and also it's better explained [4:19] the methodology that we used and this [4:22] will be updated every quarter. So this [4:24] is for you and this is the QR code. [4:26] Thank you.