Kevin Surace: How AI-driven relay attacks bypass MFA and how biometric hardware prevents them

Welcome to another CEO wisdom with Kevin Service. We had him last time. It was pretty fun. We discussed about his background. Didn't focus much on uh his great business activities. So he's CEO and CTO at Advance and Cher token apart from all the other stuff that he does. We're going to focus uh on that today because uh >> you want to talk about Broadway and film today?

>> Yeah, we could uh as well. But I I thought I'd have you on for a second round because Yeah. you have a huge background and I I think I did do you full honors last time we spoke. So tell us Kev what you're up to nowadays. What's your your keen priority and focus? >> Yeah, look I I'm busy with a couple of things because they're both very important.

one is artificial intelligence and and uh specifically how that's being used in the software development life cycle and we can talk about what that means and the other one is cyber security that because of AI is now a bigger threat than we've ever had right and it's it's changed the game on uh how we're dealing with uh ransomware hacks breakins it's just incredible and so I I

get to participate in both and invent both So, um, happy to talk about either one, >> right? Why cyber security? We've seen Vibe coding pick up as of late and apps like T app getting hacked, you know, like all this sudden. So, why uh is it a growing concern? >> Yeah, here's what's interesting. So here's why it's a growing concern is the number one 90% this is the best way to say 90% of all ransomware oriented hacks happen uh right in the front door they happen because people have hacked your credentials.

Now anyone listening to this is going to say that's not a problem because I've got MFA I've got two I've got an off app. Okay when I said credentials I was including your MFA and your O apps. Now I'm gonna teach You want me to teach everyone how to hack an off app or uh or an MFA? >> Yep. It's this easy. >> It's so easy. Teens are doing it. >> And then they're holding companies ran, you know, for ransom for $20 million.

Okay. Here's all you do. >> First, I use AI to generate uh a pixel perfect uh internal employee website. Okay? Something that has access to data, whatever that is. So, I generate that. I can use VO. I can use GPD5. I can use a lot of things, right? VO is great at it. Generate a website that looks exactly like this one. It takes it about a minute and I have a pixel perfect copy.

Okay. >> Then I add a few things on the back end to relay me information. Let's say the hacker. And now what I do is I send 10,000 emails or thousand emails to every one one email per employee at some company, some bank, right? Whatever it is, or an airline or insurance company. Those emails are also generated by AI as to be perfect. So perfect that they can't be recognized that they are a fishing email.

So the email comes doesn't look like fishing looks like it's from your IT department, your HR department, whatever. And it says uh important please log in. We think there's some false data in your account or we think something happened or we can't process this payment to you or whatever it is, right? So you click on the link and you go there and it is your website.

You see it that's the website. Now you failed to look that there is some tiny little thing that's slightly different in the URL. So instead of like employee.mamazon.com it's employee.amazon.comlo for example but you don't see that. It's so close visually that you don't catch it. So you go it's your website. So you go ahead and you log in. You put in your ID and your password.

Okay. But behind this fake site, it's relaying this information to the hacker. Follow me so far. So, the hacker is logging in on my actual employee site. Okay. Now, the site comes up and it says, "Oh, um, go to your off app and it'll come up and it'll say, is this you logging in?" And confirm it's you logging in. What do you do? You confirm it's you, but it's not you.

It's the hacker. the hackers logging at the same time. That's called a man-in-the-middle uh real time relay. So, u it happens in real time. Your information is relayed to them. And by the way, the other thing it could have done is sent you a code on your phone. You get the code on your phone, you enter it into the fake site, it relays over to them, they enter into the real site they're in.

It's a real time relay. Man-in-the-middle realtime relay. So, this is being done all the time. So, you know, companies just rolled out MFA and O apps over the last year or two, and now they're literally hacked every hour with this very simple formula of a relay attack. And um it works every time, every single time. In fact, uh something up as no matter what kind of training you've given your employees up to 10% will go see it's the real site feel safe because of MFA or O apps and log in and hand it all to the hacker.

Hacker gets in, they only need a few minutes. By the time they're discovered, they're out. They've sucked out data and they hold you ransom. Okay, it's that fast. So, I've just taught everyone how to do this and the tools to do it and it's ridiculously simple. Now that you can hack anyone who has 2FA legacy, what's called legacy MFA or O apps, we invented something at token called both the token ring and the token bio stick.

And um and and the thing with these is that they do something unique. One, you have to have biometrics. two, they will not light up from 3,000 or 10,000 miles away. You have to be within three feet of the computer logging in because it's the local browser that kicks you off and says use your token product, right? And so you've got biometrics, you got proximity, you got the fact that it can't happen remotely, and you got something else.

When you registered, say at employee.amazon.com, amazon.com, it tied your biometric fingerprint cryptographically connected to only that domain. So if someone were somehow even to get as far as trying to get you to log in to the fake domain and they were three feet from you somehow, it still wouldn't do anything because it will fail the process of uh uh sort of entangling your fingerprint with that information from the domain, right?

Those have to match. And so the whole thing falls apart. And so if everybody was using a token product, none of these hacks could happen. You can't hack token. It's it's basically fishing proof. You can't um uh you can't steal any MFA code because there's no code to steal. There's nothing to share. If you lose the ring or you lose the bio stick, it only works with your fingerprint.

So, who cares? It's it's it's the future. It's it's the way we're going to protect our identity, right? So, it's biometric. It's proximity. And it's uh you know, sort of co-mingled, if you will, with the with the actual domain of registration. So there's no hacking, there's no spoofing, there's nothing. So that's that's the whole thing with token. Uh it's token ring.com if anyone wants to look at it.

And it's a gamecher in cyber security. And so companies are rapidly moving to this technology and we're the leader in it. And uh uh companies based in Rochester, New York, spun kind of spun out of Rochester Institute of Technology. >> Very cool. Is it still very profitable to be a hacker? like uh do these folks >> yes >> uh get jailed more and more and uh do they still sleep at night?

Is it like a good life choice to be a hacker? >> There are a few people who have gone to jail but it's very hard to get caught. Most so most of the hackers are um uh Russia, North Korea, China and often they're government sponsored and and uh and Iran those four countries and they're mostly government sponsored so you're not going to arrest them. They're not arrestable.

the ones who got arrested recently, part of the scattered spider group. They were teens, they were in the UK, a couple of them got caught. It does happen. Um, but mostly companies are paying. You know, we're talking about tens of billions of dollars or maybe hundreds of billions of dollars of of uh of uh losses due to this. And so, it's a lot cheaper to protect your employees with um you know, with a product that I've described from token than it is to get hacked.

and you will get hacked with your MFA. I mean, and and the pro it was one thing up till a few months ago where you didn't know, but the Affleck hack, the Hawaiian Airlines hack, the Quantis hack, um the Clorox hack, all of these were social engineering hacks like this. There were other tricks in social engineering, too. People call it and say, "Oh, I'm Charles and um uh I forgot my password.

I don't know my password. I don't you know, whatever." and they give them a password for a one-time login. Well, again, if you had the token product, what are you going to do? Lose your password and lose your token? They're going to say, "Come back into it. You have to get reregistered, right?" Doesn't work that way. So, this protects against all the social engineering hacks.

It um it's, you know, it's it's the future. And I think it's going to be the future on um even on Zoom calls because you don't actually know if this is me. This could be an AI version of me today. I haven't I have my AI um avatar that you can interact with. It's at my website, my kevincer.com website and um I can put that on Zoom anytime or put it on, you know, a podcast.

>> Well, it's still easy to know right now and I think the human mind will evolve to spot these AIs. I think the human mind will stay like at the forefront and I think there will also be apps. Uh, for example, there could be a screening on Streamyard. They would have an API. That would be a good company idea, by the way, to automatically detect if Kevin or Charles is a deep fake.

And they would pay per API call. I think that's a billion-dollar uh company for sure. Now, you mentioned you're going to get hacked. Like, I mean, if you're an enterprise, sure, why? Because there's a bunch of low-level thinkers in companies. But if you stay small and agile and you don't hire employees, you have agents, I don't think that's a possibility.

I think uh like you said, it's social engineering, but if the if there's an AI agent working for me, I think I would actually double check the email and make sure it's not fishing. And I I'm not really hackable on my side because I'm smart. I just I can outsmart these hackers. I just know if it's fishing or social engineering. What do you think about that?

Yeah, I I what I think is that uh the statistics are you're going to snare 3 to 10% of the people you write to. And uh nobody is writing, you know, nobody's trying to do ransomware on some older person. Nobody's trying to do ransomware on a single person. They go after a company because you can send hundreds of emails and you can get one on the hook and that's all you need is one, right?

All I need is one person in a hospital and I've got access to all the records. This is easy, right? And I know statistically I will get one in the next hour. I I just know that it's not inarguable. So if I'm going after you as a podcast one, there's a high likelihood you're going to look at it more. There's a you know I don't know that I can hold you for ransomware.

What data would you have that I'd want? See, that's the problem. Like it doesn't bo It doesn't matter. But if you go after an airline and you steal their whole customer list, they don't want that released. Then they'll pay you something. I don't know if it's a million or five million, but that's that's cheap actually to just shut that problem down. So they pay it, >> right?

Yeah, that makes sense. But do you believe companies are going to scale in size? Like I tend to think that the future is probably one CEO, a couple of exacts, and the rest is fully AI agents. And I don't think they would give access to the other companies or freelancers that they work with which would make their companies way more secure like and way more fortified.

Do you also believe that to be the future of companies in terms of size and >> I'll answer that. All right. The companies that are getting hacked are large insurance companies, large retail companies, um uh um uh airlines have been a real target lately. How you going to run an airline on one person and a bunch of agents? Well, you do have agents, but they're human agents, right?

I mean, you can't, right? So, I agree with you in the tech world, as you know, you know, the the the there is a founder out there that alone is going to build a $10 billion company with the founder and some AI and some agents. We we know that, right? That person's been born. Um, but there's still going to be airlines and airlines will take, you know, a 100,000 people to run if they're of any size.

And so, uh, if you pick on an airline, pick on a hospital, right? You're not going to run a hospital with one person in an office up top. There's there's humans. And, you know, 50 years from now, will it be all robots in hospitals? It might be, right? Robotic operations and but but where we are right now, yes, when you think about this, we think about tech companies and that's because our heads in, oh, you can go build an incredible tech company that can generate leads and do go to market do and you can do it all with agents.

Fair. But that isn't a physical company. And in the end, you got to have physical companies. Second of all, even in a tech world, right? Microsoft has, I don't know, 250,000 employees or whatever, a couple hundred thousand employees. Where are they going? They're going nowhere. They're growing. I mean, they're laying off low-end people, etc. But they're not going to go away.

They're not all of a sudden going to have one person, right? Not going to happen. Not in our lifetime. I do think they're going to go through massive rounds of lay layoffs once um Agentic fully catches fire. Just that Agentic hasn't fully caught fire yet. Um and yeah, Microsoft in my opinion, I mean, it's still the stock market that keeps it alive and probably its partnership with OpenAI.

I don't see Microsoft as still valuable. I mean where I I get wrong in this that most uh older generations I still think they they use Microsoft and Outlook but these folks are are retiring right but um yeah that that's >> Microsoft well question Microsoft does how much business what's their revenue I don't know >> 100 billion a yearish something like that 100 billion 100red billion doesn't go to zero in any in our

lifetime impossible might might go from 100 to 90 but everything they're doing is actually growing that revenue base their revenue's growing quarter over quarter, not shrinking. So, it isn't a stock market thing. They do real revenue. So does Google. So does Facebook, right? So does Facebook. Could you eventually run a Facebook with very few people?

Probably. >> Yeah. Because Microsoft is in infrastructure, right? That's not about to go away. But my point is like a very strong core of their revenues related to other enterprises and legacy companies which in my opinion are either going to yeah acquire these younger startups more agile that are adding tons of value. I do expect a couple of other open AIs to see life in the next couple of years and I think that's going to drastically change the landscape.

Now tell me about uh GTM and go go to market when it comes to cyber security because I did do GTM a lot for cyber security companies and it's not the easiest thing to sell. One because you're selling to enterprise but two because you're selling a version of reality that hasn't happened yet. It's like hey you could get hacked and they won't feel the pain until they actually get hacked and when they do they probably already have a solution.

So how do you actually sell that at scale successfully? Yeah. Well, I look what has helped us is all the major hacks this year use the uh trick that I uh took you through earlier, right? And so three years ago, two years ago, one year ago, it was presumed that someone would figure this out and would use this hack. But today, it is the hack that everyone uses to get in.

And so, um, enterprises are now aware of that. And so to some extent the hackers did our bidding for us, right? They basically said this is how you do it and everyone's doing it. So if you're using MFA, it's obsolete. And so remember companies rolled out MFA because passwords are obsolete at some level. They're easy to hack. Right now MFAs are, you know, traditional legacy MFA is easy to hack.

So they're going to have to beat the criminals and go to the next thing. So um I think it's a pretty easy cell. Anyone I talk to in the enterprise, this is me, right? and I'm not really in sales, but I talk to them in the enterprise. They go, "How many can I get and how fast can I get them because I need to close this door?" Now, part of it's also just a legal risk, right?

So, before these hacks were happening, they could say, "It's just theoretical." But now that the hacks have happened and they're widely reported on this method, well, they have to do something. They can't just sit there and go, "Well, I know about it and I know there's a fix for it. I know there's a solution, but I don't want to do it." Really? That's not First of all, if you're a CISO and you said that when you do get hacked, your CEO fires you.

Why? Because you knew there was a solution and you didn't do it, right? You ignored it. Why would you do that? Well, we just rolled out MFA and I know it's hacked every day, but that's not my problem. No, it is your So, you're right. You know, look, selling to the enterprise is never easy. But the good news is once you're in, they buy 25, then they buy 2500, then they buy 25,000, then they buy 250,000 units, right?

By big numbers. And so once you earn the trust, it grows very quickly. And I think uh um I think that's what we're seeing. We're seeing because of these hacks and the way they're all being done today, enterprises are trying to close that gap, and we're just about the only solution that closes it. And say if we take it from a cold email perspective, would it help to actually send a case study that is very close to their company?

Would it help to actually go on the dark web and scrape links of people pretending to be them and send them to send it to them in a in an email and be like, "Hey, people are actively trying to hack you right now." What would be an innovative approach to sell them? I I I talked to a a bank recently and he said, "Look, you know, we have uh you know, 50,000 attempted breakins a day here, right?

So, they're very good at thwarting most of it." I mean, the numbers are like that. So, me showing up with one more that says, "Oh, look, people in the dark web people, they're going to go, "Yeah, we already know that, right? I I'm not going to do their job better than they do it. They already do that. They scan for it. They know about it. What I can do is say, I know you're very worried about this number one hack, which until I walked in your door, you didn't have a solution for.

I have the solution. You either have budget or not. I don't have any budget. Okay, then we're done talking. So, I'm not finding people who go, "Oh, I don't believe that can happen." They go, "Oh, I've been looking for a solution. This this is a problem. I know it's happening. Maybe it already happened to them once or twice. They're very worried about it.

They know that all their training against fishing emails and using MFA and using off apps is for not right that doesn't work. So they're already there. They're looking for a solution. My biggest problem in go to market uh is getting in front of people enough so that they become aware of the solution. And you know in GT probably better than I do. You probably have better stats.

But you can't just pop in and say hey I came out of nowhere. No, it's got to be oh I read an article and then I read this and then I saw something else. Then I got an email then I got a cold call. and I then I saw an ad on LinkedIn, you know, by the fifth or 10th time or 13th time or 21st time, you finally respond and go, "Oh, I should actually take a look at this.

We actually have this problem and I need to deal with it." So, I think the problem in GTM today is uh is that in order to get above the noise, you have to be seen over and over and over again. You have to have the patience >> to put enough content out there to be seen over and over again, >> right? and probably know who they're actively using and just suggest either like a complimentary approach.

Not like we'll replace the their full stack, but just here's the the hidden angles that >> That's right. We don't replace any stack. >> We close we close the big open door. That's 90% of the problem. And they're >> they're looking for that solution. So they run in a token and they don't get it the first time. You know how it is. you see stuff go by, but trust me, by the fifth or tenth time they see it, they go, "Oh, I got to talk to these guys." That's what that's about, right?

I better talk to these guys. Let me fill out a form. Let me have someone call me, right? I think cold emails are very hard today. Cold calls very hard. I mean, we all get too too many emails, >> especially with their security systems, right? >> Yeah. Oh, just all of that's hard. They're they're going to have to find you in content, in advertising, in on Reddit.

um lots of placed content. Placed articles, right, are very good. Earned articles are great. You know, when you're not advertising, you're just saying, "Hey, you are all aware of this problem. By the way, there are some solutions and you might you might take a look at this one as an example, right? You have to you have to be educated." >> Yeah. >> Let me ask you, with AGI coming, um would that thing go rogue and just hack all of us?

Um, and yeah, like what would that also mean for encryption? Let's say we pair that with the quantum computers. Are we doomed or Yeah. So, this is a great question. Putting my AI hat on because the other half of my life I work only in AI. Uh, well, look, um, let's let's let's take the term AGI out of it because I think we all get hung up on that. And it's artificial general intelligence with me.

that's more intelligent than any of us or than all of us or look, Excel is more intelligent at math than I'll ever be at math and we've lived with that. Now we have these things that are more intelligent than us at language at probably illustrating and in some cases video and analysis and sort of variety of things and medicine and and we're learning to live with that and we're learning to use them as tools.

Today the thing can't go rogue because it doesn't do anything. It just waits for input from us. Now as we start to build agents as we are um these agents if we weren't thoughtful about them, you know, we could we any of us could create an agent that would do dumb things today, right? We could create an agent that by itself would eventually hack into our own systems and grab all the information and stick it out on the public web.

like I could create an agent to do that but accidentally do that because I wasn't thinking about what are all the things that can happen in the agent's life and how the agent would respond right so I think we have to be thoughtful about these agents that we're building because that's the first time that we're letting things work autonomously without us otherwise chat GPT you know I

ask it's a query it answers it's done it's all completed right it's there's no more task agents have ongoing tasks and live forever unless we kill So, it's an interesting, you know, it's going to be an interesting time. And then they're still not that dangerous because it's just running at Azure. I mean, it could give away data, but it can't harm you.

But what happens when we've connected those agents internally in humanoid robots five, six, seven, 10 years out? Well, now a humanoid robot, even if it's programmed not to, you know, could go ary, right? something could happen and it could start destroying things and literally hurting anything, hurting the cat, hurting the dog, hurting you, right? Uh so I I think once we get into the physical world, I think this is a much more interesting conversation to have since it's software running at Azure and anyone could unplug the server.

I don't know what can it do. It's just software. It talks to other software mostly. It doesn't even know how to talk to anything else. well can steal financial data information and that would be enough to bring the world down if people don't have their digital cash anymore. Of course, there would be digital currencies, but I think that'd be pretty catastrophic and ruin reputations, right?

>> Well, what Okay, so what could it do? Unless someone wrote an agent to do something like this that gave us access gave it access to systems that it wouldn't normally have. uh a a an LLM by itself can't do that. Can't do anything, right? But we could write, you know, you and I could sit down and say, "Let's write an agent that just pounds away and eventually breaks into the World Bank and I don't know what it transfers all the money to itself, right?" Uh or transfers it out into space, so it's gone, you know, aes it away.

I don't know. You could probably come up with some scenario, but someone would have to write that. So, that's going to take uh a criminal mind to do it. It's not the LLM by itself. that doesn't know how to do it. >> Yeah. Yeah. I agree with that that part. Uh Kev, where can people find out more about you again? >> kevase.com and you can talk to me on kevinace.com.

Go to the bottom, go to my virtual Kevin, click video, and you'll you'll see me actually talking to you and you can What's really cool because I can see I can see you. My agent can see you. My avatar can see you. Um you can ask it what's behind you. Like what do you see behind me? Oh, I see a bookshelf and I see a radio. looks like um some kind of older electronics and then I see some awards maybe you you know it'll tell you and you go I'm freaked out by this because it has eyes.

It's pretty cool. >> There you go. That was Kevin.