The One Question bounty closes Thursday 10 December.

Join in
We Them Media logo We Them Media Support us

Beyond Audits: Securing Web3 in the Age of Lazarus - Gracious Igwe

The picture loads from YouTube. The player loads only when you press Play. Turn on captions in the player if you need them.

Gracious Igwe argues that smart contract audits alone do not keep Web3 projects safe. The talk cites 2025 loss figures, says most hacked contracts had been audited, and describes how the Lazarus Group uses fake job applicants, social engineering and supply chain attacks. It then covers operational security for individuals, organisations and protocols, and makes the case for continuous monitoring.

Not by We Them Media Channel: Web3Bridge. Published Sunday 26 October 2025.

Titles are the owner’s. It plays from their channel.

Transcript

This transcript may contain errors. This video belongs to Web3Bridge. Letters mark each change of speaker.

These are the speakers’ own words. We have not checked their claims.

0:02 A: All right. Hello, everyone. GM, GM, everybody. Yeah, my name is Gracious, and I'm a security analyst and also a security researcher. I'm very pleased and excited to be here, and it's an honor to be standing here speaking to you guys about web security. So, like I said, I'm a security analyst. I work at the Security Alliance, also known as SEAL, and we're a nonprofit crypto-native security collective. Okay, crypto-native security collective dedicated to protecting Web3 protocols through rapid response, collaboration, and shared intelligence. So today I'm going to be speaking about securing Web3 beyond audits, and, um, the title of my topic is Beyond Audits: Securing Web3 in the Age of Lazarus. So my reason for giving this talk is because I do not believe that smart contract audits are sufficient, because despite all the efforts, all the audits, and all the money we've poured into security, um, we're still losing billions to to hacks and attacks. So I'm going to tell you why, in my opinion, I do not think that smart contract audits are enough for this space. So let's get down to business. Uh, yeah, the numbers don't lie. The numbers we are seeing in 2025 are unprecedented. Um, in the first half of 2025, Web3 recorded approximately $3.1 billion in losses to hacks. And, um, what is interesting is that This is, this is already more than all the losses that, that was recorded in the year 2024. And interestingly, 60% of those losses were from the Bybit, um, February 2025 hack, the $1.5 billion Bybit breach. Now let's take a look at the attack vectors. We have control hijacking, which, um, accounted for $1.7 billion in losses, and then phishing attacks, which accounted for $410 million in losses. So together, these figures imply that Web2 attack vectors- Web2 attack vectors are dominating smart contract exploits, right, in the Web3 security space. So the question is not if another attack is going to happen, it's when and how it's going to happen, and if we are prepared to tackle it. Because despite the widespread awareness of this problem, the ecosystem is still fundamentally vulnerable. All right, so let's move on. So that brings us, um, to a critical misconception in the industry, which is that audits equate to comprehensive security. But then the data says, or rather statistics says otherwise, because 92% of hacked smart contracts in 2025 had undergone some form of audits, undermining the belief that an audit badge means a protocol is completely secure. So if audits are supposed to be our safety nets, then why are we still falling through the cracks? Well, the reason is because we've created, um, a sort of- we've created an ecosystem where the appearance of security takes precedence over actual security measures. When you look at current auditing practices, um, they are limited by their narrow focus on code one-time assessments, inconsistent quality, and inability to detect human-driven threats. And that leads us to the question of who audits the auditors, and are there even quality standards, or are there even standards in the space that ensures that we get quality audits and assessments, right, in this space? So moving on, Now, this guy, um, uh, now this guy, Bruce Schneier, he observed that amateurs hack systems but professionals hack people. And I think that this distinction is important in understanding the current Web3 attack landscape. So if you're, if you're active in the Web3 security space, and you're someone that regularly reads audit reports and postmortems, you would notice that recently, um, reports are dominated by, um, human-based or human-layer-based vulnerabilities. But as at 2022 and 2023, reports were dominated by code-based vulnerabilities. So in recent years, attacks are driven by operational failures, and social engineering have become significantly more common within the space. By 2024, high-value attacks succeeded primarily through compromising human and organizational layers rather than technical exploits. So basically, the tech is solid, is, is, is good, but the human factor is what is the actual problem. It's very vulnerable and is susceptible to attacks, and attackers know this. They recognize this. And, um, They recognize this and they're taking advantage of that. So, so Lazarus- how many of us know about the Lazarus Group? Please raise up your hand. Okay, so the Lazarus Group has executed more than 40 attacks resulting in theft exceeding $3 billion in the DeFi space. Their strength extends beyond technical vulnerabilities to understanding human behavior, and they don't just break code. they break trust in the entire system. There's something I refer to as the decentralization illusion. This is- this means that we all know that the tech is decentralized and it's solid, but then the humans operating this tech are centralized and they are vulnerable. And this creates a kind of, um, disconnect and creates an opportunity for attacks and hacks to happen. So how does the Lazarus Group operate? We have insider positioning, uh, they apply for legitimate jobs. Sorry, okay, so they apply for legitimate jobs with fabricated credentials. They work as contractors through consulting firms. They become trusted community members in governance roles. They maintain cover for months and even years. Before attacking. Then we have social engineering, multi-platform campaigns. They impersonate executives, they use deepfake technology for attacks, they create artificial urgency. And then we have supply chain infiltration. They compromise development tools, libraries, and frameworks. They infiltrate communication platforms and community spaces. They target service providers and contribute malicious code to open source projects. And then lastly, we have long-term relationship building. So they create fake LinkedIn profiles, they engage in technical discussions, they offer job collaborations, and they extract organizational intelligence through conversations. Then we have notable incidents, the Running Bridge hack and the Harmony Bridge hack. So, The Running Bridge hack, um, Lazarus created fake developer profiles and they spearphished the employees of the Running Bridge, and then they gained access to validator nodes and drained about $624 million. While for the Harmony Bridge hack, they used deepfake-enabled video calls to gain trust and eventually walked away with $100 million. Now it's interesting to note that neither of these hacks were caused by code bugs. But they were as a result of human and organizational failures. And both of these projects, or rather protocols, had- Both of these protocols had passed multiple security audits, but then they suffered massive breaches. And this reveals a critical gap in traditional security approaches. There's something referred to as the audit blind spot. So you build your project, you get an audit, But then an audit is not going to tell you that your key storage practices are below par. Your audit is not going to tell you that- it's not going to prepare you for a social engineering attack. Your audit is not going to tell you that your employee is a North Korean state actor. So that is referred to as the audit blind spot, which is comprehensive code review but paired with complete operational blind spots. Then the decentralization paradox is also a factor. Um, both protocols promoted decentralization as a security strength, but in practice, it did not eliminate human risk. Rather, it spread it across potential attack vectors. And then lastly, we have the human factor. Lazarus succeeded by exploiting human vulnerabilities rather than breaking code, demonstrating that true operational security must address both technical systems and human behavior. All right, so I've been talking about operational security, operational security. What exactly is operational security? Well, OPSEC involves implementing proactive procedures and practices to protect security assets from attacks. So basically, OPSEC is just like treating the operational and human layer just as important as you treat the technical layer, which is your code and smart contracts. I think it's important to distinguish OPSEC from related security disciplines such as SecOps and AppSec. SecOps, which is security operations, is primarily focused on reactive security measures, including real-time monitoring, threat detection, and incident response protocols. And then AppSec involves designing, building, and maintaining secure code through bug bounties, audits, and- bug bounces again. So SecOps is like reactive, is the thing that you do after you've deployed your projects. So like you do real-time monitoring, threat detection, and incident response. And then AppSec is like proactive, like the ones we all know, which is code audits, writing secure code, and so on and so forth. So most Web3 security resources are focused on the AppSec factor, but then operational security failures cause the majority of losses, highlighting a critical misalignment that needs correction. Then we have the OPSEC framework. The OPSEC framework is basically, um, operational security measures that you can implement. This is not comprehensive, like this is not everything. These are just like the key ones that I selected. So we have key management, which is the secure handling of private keys and signing procedures, endpoint security, securing individual devices and machines, personnel security, education and human factor management, application security, securing applications and interfaces, code management, secure development and deployment of code, and then access control, managing your infrastructure access and permissions to ensure that only authorized personnel have access to your system. So here you can see a real-time monitoring dashboard for web security incidents, and it was built by Chainlight. Name is Lumos. So it, it just sums up everything I'm talking about since 95% of emergency incidents in April 2025 were caused by basic avoidable security lapses. So Web2 attack vectors such as social engineering, key leaks, supply chain compromises accounted for over 3 times the value stolen by smart contract bugs compared to smart contract bugs. And then from January 2023 to date, control hijacking led to $3.31 billion in losses. But then in contrast, smart contract vulnerabilities accounted for just $900 million in losses. And then we have major cases and studies. There's the Bybit hack I talked about earlier. So North Korean hackers evolved from targeting individual protocols to critical infrastructure, and the Bybit incident compromised the safe wallets. SafeWallet platform used by thousands of DeFi users. So what happened was in the Bybit incidents, um, North Korean hackers didn't just- they didn't just target, um, individual protocols, rather they targeted the core Web3 infrastructure, which is the SafeWallet platform. And Bybit, being the biggest users of the SafeWallet platform, were actually the victims of the hack. Then we have Radiance Capital. Um, attackers build credibility over extended periods before striking. The Radiance hack involved a fake security researcher who spent a year establishing trust before deploying malware. So, this was what I talked about earlier where I said your code is not going to tell you whether your employee is a North Korean state actor. In the Radiant Capital hack, a North Korean state actor had sent a file containing malware to the Radiant Capital team. And then when they clicked on the file, it affected their machines, and then they were able to- North Korea was able to steal or drain their funds. And then we have the WazirX hack. Attackers bypassed hardware security, wallet security, By manipulating what users see before signing, and then victims unknowingly authorize malicious transactions while believing they are legitimate. So we have personal OPSEC for developers and users. Effective operational security starts at the personal level. It includes preventing information leakage, sensitive information like travel schedules and role updates. Can provide openings for social engineering attacks. Then secure physical environments, extend security to development spaces and communications channels, and then rigorous key management protects against, protect against social engineering attacks and physical access vulnerabilities. And lastly, we have social media awareness. Attackers build detailed profiles by aggregating information. for multiple online resources. Moving on, we have OPSEC at the organizational level, which includes vetting your hires, like thorough screening process for your team members or your new employees. Then cultural balance, navigating security needs, insider threats, managing risk from compromised team members, infrastructure hardening, strengthening systems against potential vulnerabilities, team protocols, and lastly, secure communication. And then OPSEC at the protocol level. So at the protocol level, OPSEC involves more than just deploying multisig wallets. It requires clear process for governance and treasury protection. So for those of us in governance, these are the measures that you can take to ensure that your operational security is solid. We have multi-signature security. Effective multisig requires geographic distribution, operational independence, and so on and so forth. Governance controls prevent attacks through snapshot-based voting, quorum requirements, and monitoring. We have treasury protection, time locks implementation, incident response, and admin key management. All right, so security as a life cycle. This is like the crux of everything I've been talking about since. Web3 security must be approached as a comprehensive life cycle. This means that you must work towards approaching secure Web3 security as a continuum. That is, as a continuous process. It's not something that you just do once and then you forget about it. You have to continue doing it. at every stage of a project. So we have pre-deployment security, which includes threat modeling, secure development practices, comprehensive testing, upset planning, and incident response. Then we have deployment security, which includes real-time monitoring, anomaly detection, community alerting mechanisms, response team, and continuous threat intelligence. And lastly, we have post-deployment security, security, which includes operational security, systemic improvements, and regular assessments. Then the continuous approach versus point-in-time audits. So the continuous approach is just like continuous approach to security versus just having an audit once and forgetting about it. So the point-in-time audits are like the traditional audit models. And they force teams to react only after something has happened or after an attack has happened. And then it often takes days to, to respond. So like, there's usually a delay after like something has happened. And then in that process, the attacker might escape with the money and the money, not be able to recover it and so on and so forth. But with a continuous security approach in place, It involves constant monitoring, regular expert reviews, and threat detection to catch problems early and respond faster. Then we have for point-in-time audits, they quickly become outdated because security is always evolving. It's not something that is, it's a constant, right? It's always changing. So like protocols update and threats evolve. Leaving systems vulnerable to new risk. Then shared responsibility. The continuous approach brings users into the security loop through bug bounties, making security a collaborative effort. So the mindset shifts. So as well as we try to adopt security, the continuous security approach, We need to change the way we think about security. So instead of thinking, oh, we got audited, we are secure, we don't need to do anything else, you start thinking that, oh, we are continuously securing our system. Instead of just saying, oh, security is just a requirement, then you start saying security is part of how we operate, security is part of our infrastructure, you know. Then audits prevent attacks to continuous vigilance prevents attacks. All right. So key takeaways. Security is not a product, it's a process. Yes. Constant monitoring, continuous vigilance performs points-in-time security checks beyond audits. Audits provide a necessary foundation, but they cannot address all the vulnerabilities. And then human vulnerability. OPSEC is now the primary attack vector. And social engineering compromises attack even the most sophisticated smart contracts. All right. So how you can contribute. For developers and teams, move beyond smart contract audits to include OPSEC assessments. Establish multisig management procedures. Create incident response plans. Implement continuous monitoring. Contribute to open source security tools. For investors, look beyond audit badges to assess OPSEC practices. Evaluate multisig configurations and and key management procedures, and then supports the development of security tools. And then you, as the user, you can report potential security threats to the community, contribute to open source security resources, and verify protocol security claims rather than relying on audit badges. So at the end of the day, security is everyone's responsibility, and we must come together to ensure that we are able to shift security from where it is right now to a more continuous and human-centric security culture in the Web3 space. Okay, questions?

More from this hub

Something wrong, or a video you want off this page? Write to wethem.xyz@gmail.com and it is fixed or taken down within 48 hours.