CTO, Radware
David Aviv’s interview with the Cyber Risk Alliance at RSAC 2025
"How Adversaries are Rewriting Cybersecurity Rules: Adapting to AI-driven Attacks" - Adversaries are rewriting the cybersecurity rules. Shifts in the threat landscape are being fueled by attackers with political and ideological agendas, more sophisticated attack tools, new coalitions of hacktivists, and the democratization of AI. Radware CTO David Aviv will discuss how companies must adapt their cyber defenses and lead in an evolving era of asymmetric warfare and AI-driven attacks.
This segment is sponsored by Radware. Visit https://securityweekly.com/radwarersac to learn more!
Show Notes: ht...
About David Aviv
At RSAC 2025, Aviv stated that the democratization of AI is "changing dramatically not only the security posture but our lifestyle" and argued that human abilities will not keep up with AI abilities. He described seeing "agentic AI bots" that forward JavaScript anti-bot challenges to large language models, which quickly solve them and allow attackers to bypass protections. Aviv said Radware uses AI to build deeper security controls and to protect AI infrastructure, noting that running AI applications creates a new AI supply chain and new threat surfaces. In earlier appearances, Aviv discussed the shift to edge-centric architectures driven by 5G, stating that security must be pushed to the edge and that the 5G core, built on HTTP/web-based protocols, opens networks to web-based attacks. He noted that account takeover has gained momentum via spear-phishing, allowing attackers to assume identities and install stealth agents in cloud environments. Aviv also described Radware's approach to self-learning security algorithms, stating that in one week their devices produced about 130,000 attack events and roughly 99.9% of mitigations were performed automatically.
Transcript (35 segments)
M
Mandy Logan0:14
Welcome back to RSAC Conference 2025. We're here recording live from the Cyber Risk TV studio and broadcast alley at Moscone West. I am your host Mandy Logan and joining me for this interview is Dr. David Aviv, Chief Technology Officer at Radware. Welcome, David.
D
David Aviv0:31
Thank you, Mandy, for having me.
M
Mandy Logan0:34
I am so glad that you are here. And in the few minutes before we started our show here, fascinating conversation. I've got to tell you, it's always encouraging to me when I get to speak to a technology officer at this conference. I'm curious. Now, you have all that you have been working on with Radware products. You have all that you have been focusing on and looking to create. What have you noticed at the conference that either aligns with what you are doing or is completely opposite of what you are doing?
D
David Aviv1:06
Well, it's a tough question but I know anyway from what we have seen obviously there are many, many sections that we in Radware don't, we are not engaged with because there are many security domains that different vendors are offering. We at Radware we are mainly protecting our data, our customers' data centers, service provider networks for when they expose their assets publicly outside. As far as we know obviously there's a big hype on the AI applications, protecting AI applications. We do see a lot of concerns about identity, to protect identity, preserve identity. It has a lot of governance issues as well and beyond that, you know, the traditional I would say the enterprise security solutions as well as, you know, traditional application security, network security, there's a mix of many, many things here. As far as I see, we are quite aligned on what's going on in the market space. And we plan our own next generation of or next wave of products to cope with the new challenges of protecting AI apps and AI environments.
M
Mandy Logan2:43
And is that in your view the main thing that is changing the threat landscape?
D
David Aviv2:50
I would say that definitely the generative AI, or I would say the democratization of AI, is changing dramatically not only the security posture, our lifestyle, our life posture. Quite soon, you know, we start talking a lot of vibe coding, right? So everyone talks on vibe coding. Is vibe coding going to make obsolete the junior engineers? So it has an impact on every part of the industry and far beyond the industry because we're seeing people who have nothing to do with cybersecurity and in family life, in home life, everything, the adoption of GenAI and the way that it is.
M
Mandy Logan3:40
I feel like we're actively seeing kind of a rewiring of culture and mankind's thought process with having this much information available and also realizing that our human abilities aren't going to keep up with AI abilities.
D
David Aviv4:02
Yes. Well, there is a lot of debate on, you know, the AGI, the let's say general AI that will obsolete people as human beings. But, you know, it's very hard to predict, very hard to predict. I'm a very big, I can be considered as optimistic vis-à-vis the human nature and the way humans will handle. But one thing we can say for sure, my lifetime I've seen already the web revolution and this is another big revolution, maybe bigger than the web. Don't know yet, but it might be even bigger than that one. It will probably impact every part of our daily life and we see the early signs of that. And obviously as a security, as a technology officer, providing and having to address and protect those kind of new challenges. We are now at the race of understanding the new trend, the new threat landscape. What does it mean? What does it mean for us? How can we craft new wave of products to be aligned? And for that obviously we listen very carefully to our big customers.
M
Mandy Logan5:38
That's good. I mean you as an engineer and product overseer, you're looking, of course you have to have data, yes, and you need it from your customers. And is it with your background in telecommunications and electrical engineering, is there any aspect that you look at this very start of generative AI being adopted broadly and go, oh, this is just another means of human communication?
D
David Aviv6:09
Yes. Well, that is a, it takes us back to our previous session before, but this is evolution and revolution. Evolution in the sense the way we communicate and we can create new things, but a revolution in the technology side, the way we can adapt new things and the way it can lead the usage, the new way of using neural networks in order to solve problems, to understand problems. This is going to bring us to completely different areas. I'd like to give a very simple example.
M
Mandy Logan6:52
Yeah, please do.
D
David Aviv6:55
In security we model, we tend to model threats and by modeling the threats we get to understand what kind of models we need to build in order to outpace them. So there were a couple of problems, very tough problems like business logic attacks for someone that is building protection for APIs. Traditionally if you use traditional machine learning methods, well, you can do as much as you can but you cannot go deeper than that because of the limitation of the CPU and memory. Well, guess what? With the new generative AI technology that we already have, it's much easier to model a much complicated model that can now uncover threats that we couldn't even see before, we couldn't understand before. It's like a chess game. You can play ahead two moves all the time and now with the new chess I can play, I can see ahead 10 moves at a time.
M
Mandy Logan8:19
So that makes me think if we go back to a simplistic view of an adversary mindset versus the defense mindset, is what you're saying that generative AI is allowing defense to more quickly adapt to how adversaries are creating the evolution of the modern threat landscape?
D
David Aviv8:39
Absolutely. If you know how to use it. It's a traditional, you need AI to combat AI.
M
Mandy Logan8:50
Well then how is Radware using it?
D
David Aviv8:53
So we use it in, we use three layers of that. The first one we call it, you know, it's secure by AI, which means we use the AI in order to build deeper security controls. So our algorithms become better by simply using AI. What does it mean? It means that we can now solve problems or uncover threats that we couldn't uncover before, couldn't understand before. Simply by using new technologies, new generative AI technologies in which we augment our current solutions. So this is the first layer. The second one is protect the AI infrastructure. Our customers start to use AI applications. Well, using AI applications brings up new threat surfaces.
M
Mandy Logan9:58
Well, and doesn't it also increase old threats? I mean, it is increased what has been around forever.
D
David Aviv10:05
Absolutely. Amplifies many things, right? Adds many things. I give an example. If you run an AI and a regular application, you know, we were used to the supply chain, the ecosystem, and there were many attacks through that supply chain. Now, when you start running AI apps, you have a completely new AI supply chain. It's a different thing. It's a different beast. So, well, how do we address it? How do you discover AI assets? We were used to talk on shadow IT. Now we are talking on shadow AI. There are many new AI assets in the enterprise. Well, there's a new task. Discovery of them, label them, map the risk of them, safeguard them. So, it's an endless list that we are now just addressing and starting to view and understand and based on that start to build solutions for them.
M
Mandy Logan11:14
And with your triple layer AI assisted technologies within Radware, what is something that a security team or a SOC analyst would see while using Radware? What is a benefit that they would see? What would directly help them?
D
David Aviv11:28
There will be two, I would say, I will speak in two dimensions. The first dimension is you will have better tools to remediate the problem, to bring the RCA, to shorten the meantime to repair, to remediate and find the root cause analysis via new tools. I'll give you an example. Today in our traditional SOC, some events, some critical events will bring up the experts to deal and spend an hour, two hours, three hours in order to resolve the issue. However, with our new agents, you bring up the new agent for remediation. It cuts down the MTR, the meantime to repair, to two minutes.
M
Mandy Logan12:25
Oh wow.
D
David Aviv12:27
Two, three hours to a couple of minutes. So that's one dimension, being able, what we call it, anti-fragile network, moving to an anti-fragile network, not a resilient one.
M
Mandy Logan12:41
Yes. Okay. So anti-fragile, that's a really good turn of phrase.
D
David Aviv12:47
That is Nassim Taleb, this is not mine. This was phrased by Nassim Taleb. So what we mean, and we adapted that term, is that resiliency can absorb the shock but stays the same. Anti-fragile networks absorb the shock and changes and adjusts, it changes to cope with that and plan for the next one. It means that the new SOC will have new intelligence capabilities to understand the attack and then predict, move to a proactive and a predictive mode in which we'll be able to anticipate for the next wave, make it better, remediate it better.
M
Mandy Logan13:44
Now that is fascinating. It's making me think of a lot of different things, but I am curious. Do you feel that with the advances adversarial AI use is bringing, that the industry as a whole or perhaps subsections are on the verge of a black swan event?
D
David Aviv14:06
Well, it depends of the viewpoint, you know, which kind of industries will evolve on that. But it's going to take an extreme impact on all of us. You know, I must be frankly with you. Let's say a year and a half ago when we started the journey on generative AI, I wasn't that sure that the impact will be that big. I wasn't.
M
Mandy Logan14:38
You mean in the end results for your customers or?
D
David Aviv14:42
No, generally what can be really achieved with those kind of things. But as the months go by with the new capabilities and I would say the new ecosystem that's been developed around it and the speed of that, I personally was quite amazed on what can be achieved utilizing those kind of new tools.
M
Mandy Logan15:12
Can you provide an example?
D
David Aviv15:14
Yeah, it's like, give an example. One of our customers, big customer, e-commerce, came to us, told us, you guys, you provide us an anti-bot solution. For detecting scrapers, our business model is based on those, you know, detecting those bad scrapers and avoid them. Otherwise, we expose our catalog and everything, right? But you as well as the entire industry provide a solution which is based on providing suspicious guys, suspicious actors a challenge. And you expect the actor to solve the challenge in order to be sure if it's a friend or a foe. I mean it's a good guy or a bad guy, human or machine. Let's say different ways to deal with it. But we start seeing a new wave of agentic bots, agents, bot agents, AI agents that when we send them the challenge, which is a JavaScript challenge, the agent sends it to an LLM engine, large language model engine. And the engine is able very fast to understand the challenge and solve it. So they can escape our entire catalog. We're in big, big trouble. So this is a way where you can see the adversaries are using those kind of new technology, new agents to defeat the defenses. So our job now is to bring new tools that are defeating those agents that try to defeat the defenses.
M
Mandy Logan17:20
But in that moment when you realize that that is going on. I mean that's a complete almost I feel like a watershed moment like it just makes you stop and go, oh, we have just entered whole new territory.
D
David Aviv17:31
Yes. A new paradigm. Absolutely. It's a new paradigm. Yes. It's a new paradigm of everything. And that's why being able to understand, to run, to test, to understand the new technology and then be able to understand the new risks that can come through that.
M
Mandy Logan18:00
Going back to what you said, that's like 4D chess. Having to stay ahead of all of that which is incredibly satisfying for those of us who want to do that and understand the mission toward it. And David, I appreciate your conversation here with us today and I definitely want to encourage our audience and all of the CISOs to look into Radware's information because these are key factors for the changing threat landscape and for the next, I mean we're here, the future is here, it is now. And thank you for joining us. To find out more about Radware, please visit securityweekly.com/radware-rsac. For all RSAC 2025 coverage by CyberRisk Alliance, please visit securityweekly.com/rsac. Thank you again, David.
D
David Aviv18:44
Thank you for having me.
M
Mandy Logan18:46
You are so very welcome. Stick around. We'll be back after the break with more interviews live from the RSAC Conference 2025.