DSS 2026 — Why Crowdsourced Security Is Dying (Code4rena, Hats Finance, and What Comes Next) Dmytro Matviiv Source video: https://www.youtube.com/watch?v=p462_m4FU9I Automatic captions — not human-reviewed. Complete spoken-audio coverage is unverified. Source language tag: en (English). Source export SHA256: 1347ebc36c32a47193fe2e0d139306b7ed7887d8aef49c3ec9826761c2d23b84 [0:00] Yeah, let's let's do it. I'm Dmytro. I'm [0:02] from Ukraine, and actually I hold my [0:04] life in cybersecurity, as everyone is [0:06] here. And today I would like to talk [0:08] about especially crowd source security [0:10] since actually you do crowd source for [0:12] the last nine years. [0:14] So, [snorts] [0:16] yeah. [0:17] Today I would like to focus only on a [0:19] stuff that related to contest and crowd [0:23] source security, yeah? So, if you [0:25] actually remember, we had a platforms [0:28] before. It was Hacker One, it was [0:29] Bugcrowd, and etc. Then actually new [0:31] platform is appear on the market as [0:33] Immunify, HackenProof, and etc. And best [0:37] years for everyone probably was 2021st [0:40] where we actually had new models on the [0:43] market which is contest, yes? And then [0:46] we have a new reality in 2024 and 2026. [0:50] So, [0:51] here [0:52] we should find the common name, yeah? [0:55] Crowd source audits, contest, hackathon, [0:58] wherever everyone had trying to create [1:01] own name for this space, but the idea is [1:05] is one. Many researchers, one codebase, [1:08] and one goal, find vulnerabilities, and [1:10] of course earn money, but not on the [1:12] slide. And so I call this chapter is [1:15] beginning, beginning of everything. [1:18] So, what's the problem we trying to [1:20] solve this time? So, first of all, there [1:23] was not enough [1:25] security researchers on the market, yes? [1:27] New smart contract languages and etc. [1:29] appear, and everyone wants to know that [1:32] everything is perfect. [1:34] Cyber security firms were were fighting [1:36] for security specialist, and everyone [1:38] was saying, hell, no, we actually know [1:40] how to do move audits, rust audits, and [1:43] etc. But at the end everyone was [1:46] find out that this is was the same [1:48] specialist cross-sharing through each [1:51] auditors. Also, at the same time, there [1:54] was lots of security good engineers who [1:56] had a full-time job as a developer who [1:59] actually built a code base and they [2:02] like, [2:03] hmm, why I cannot just doing some [2:05] part-time job legally? They asked it [2:08] actually the HR to make some [2:12] extra legal agreement to participate in [2:15] etc. So market was growing more and more [2:17] and everyone wants to have transparency [2:19] because like if some languages appear [2:21] have to learn bugs, have to actually be [2:24] faster than bad actors. [2:26] So what's happening then? [2:28] Best times when actually Code for Arena [2:31] was announcing about first contest for [2:33] more than $100,000, [2:35] everyone was talking about this that [2:37] hey, we are going to earn money, not [2:39] much specialist and at the beginning it [2:41] was a few guys who participated and they [2:43] earned good money. [2:45] So if you actually remember it's public, [2:47] all data is public, yes, so nothing is [2:50] to hide here so you can make a photo and [2:51] etc. So with those year actually was [2:55] documented more than 160 crowdsource [2:58] audits on a market [3:00] and more than nine [3:02] million plus actually rewarded bounties. [3:05] And there was more than 200 unique with [3:08] critical vulnerabilities. Everyone [3:10] actually could learn. So many good [3:12] actually initiatives was around. I [3:14] remember even Pashov groups [3:17] I met guy at the conference and then [3:20] they collected together good talent is [3:22] and created a new firm security. So lots [3:26] of talented actually created on those [3:28] time. And the business model was pretty [3:30] nice, yes. So [3:32] companies so we talk about Code for [3:34] Arena for example, they earn around [3:36] 20-30% from each actually contest. So by [3:39] having contest for 100,000 you could [3:42] earn actually as a business [3:44] to 20-30,000 dollars actually [3:47] and work. So if you remember there was [3:50] an article, unfortunately my designer [3:52] didn't put it, but actually there was [3:54] article there where Code Arena was [3:56] promoting a guy who actually earned more [3:58] than 1 million just for 40 months at the [4:02] Code Arena for [4:04] participating in the contest. [4:06] And then actually became a boom. So when [4:09] something is appear, everyone trying to [4:11] step in as a business, work, find [4:13] models, and etc. And this is what was [4:16] happening. So [4:17] Spearbit, which is Cantina, they [4:19] actually raised money. There was a [4:21] public announced that they will going to [4:23] crowdsource security market and improve [4:25] this market. There also actually [4:28] was opportunity for for this because [4:30] everyone was knowing that Code Arena was [4:32] charging 20 30% for [4:36] listing companies and they could [4:37] actually build strategy to get [4:40] investments into it. And then actually, [4:43] if you remember, there was a boom of [4:46] millions contest. So it was actually [4:49] make a DAO on the Sherlock. There was [4:50] actually Blast, Uniswap, [4:53] Arl, etc. So there was like a boom of 1 [4:56] millions [4:57] platforms and even companies who wants [5:00] to have a audit, they were fighting to [5:01] be listed at the platforms and actually [5:05] show that they have money to make a [5:07] contest. And everyone was thinking that [5:10] contest is cheaper than doing internal [5:12] audits, but the reality always was and [5:16] and is that contest is always expensive [5:19] than actually doing private audits. [5:22] And actually, so everyone wanted to be [5:24] in, yeah? New talent, learning boom. [5:27] Everyone was just making some courses, [5:30] training actually pipelines, and etc. [5:33] And then actually, if you remember, [5:35] there was a Cantina elite security club. [5:38] They announced it that if someone is [5:40] joining, they cannot participate in [5:42] other publicly contest and etc. Everyone [5:46] was trying to step in into this contest. [5:49] And then was a turning point which is [5:52] actually happening. So [5:54] crowd become too crowded and more [5:57] researchers, same price, submissions [5:59] overloaded, early AI entering to the [6:02] games, more duplicates. So, everyone [6:05] trying to rethink um [6:08] technology. So, if you actually take a [6:11] look, so previously there was around 500 [6:14] that price pool for submissions. So, in [6:17] 2026 there was less than five times [6:21] public contest and price for per [6:23] submission was just $80. [6:26] So, there was a perfect storm, yes? So, [6:29] crypto winter, platform competitions, [6:31] hungry researchers, um [6:34] AI wasn't the main boom yet. So, [6:36] actually [6:38] what's happening the next? I will take [6:40] one minute extra, okay? So, zero [6:43] yes. So, California actually release [6:46] zero fee for the companies. So, come and [6:49] not paying. But the reality, if you [6:51] actually company and you have register [6:54] your entity, let's say in US or [6:56] somewhere, you also have to pay money [6:58] salary. I'm talking as a CEO for you, [7:00] yeah? So, to keep company at least for I [7:03] don't know, 10 people in US is [7:06] more than $100 per month. So, actually [7:10] what's happening? So, companies, [7:12] platforms start [7:13] competing with each other. Everyone [7:15] trying to show zero [7:17] bug fee model, [7:19] conditionals pools. If you remember, [7:21] let's say someone was saying, hey, you [7:23] have to find 10 critical [7:27] So, that was the reality. Then severity [7:29] based when like previously the main idea [7:31] was there is a pool and you take pool [7:33] even you have low, medium bug. Then [7:35] actually critical, high, low and [7:37] everyone has to participate. So, what's [7:39] happening then? Two pioneers die, yeah? [7:44] Because of business model first of all. [7:46] Second, it's my last part, cows. Cows [7:49] actually was related to what? Everyone [7:52] trying to do with AI, everyone trying to [7:54] participate in new models appear on the [7:56] market, pay submission and etc. And [7:58] where we move, last slide from my side. [8:01] So, we move to [8:03] from, hey, we need more talents to we [8:06] have AI. So, next what we will see is [8:09] actually we move to the way when AI [8:13] researchers and actually researchers [8:15] will burn their AI tokens to save tokens [8:18] of companies to actually make AI audits. [8:22] Yeah, thank you.