AI, Quantum, and Cybersecurity: Protecting Critical Infrastructure in a Digital World

Aaron Crow (0:0.750): Hey everybody. Thank you for joining me for another episode of the protected all podcast. I'm super excited to have my friend cat on the, on the line with me today. we're gonna, we're gonna have some fun. It's going to be a little spicy, which is exciting. and so cat, why don't you introduce yourself? Tell us who you are and a little bit about your background.

Kat Wang (0:16.588): Thank you very much. A-A-Ron. That's how you pronounce it, By the way, before I introduce myself, I have to say this is very exciting to be here. So thank you for having me. And I really had to work when I read this Protect It All podcast title, because you've got inner capitalization. This is like very meta. You've got OT and IT and all that stuff all in one word, brother. Like, mad props on making me work hard for the title.

Aaron Crow (0:19.554): That's right.

Aaron Crow (0:48.096): It's all the little things that are in there. It's everything about it, right? And in today's world, it's easy to focus on one or the other, but this is really just around all different things and all the things that we need to do to make sure that we protect the world around.

Kat Wang (1:2.044): I do especially love the call to OT. It is a huge, huge problem in my mind. It's something that nobody really saw coming. And maybe some did, but not enough of us to know to safeguard it for the world of today. So my name is Kat Wang. I am a principal of the public sector division within SandboxAQ. We sit at the strategic intersection between AI and quantum technology. So not quantum computing, but quantum technology. We do cool things like quantum simulations, right, for material science or...

Aaron Crow (1:13.144): Yeah.

Kat Wang (1:33.073): pharmaceuticals to find cures for undruggable diseases. We do quantum navigation in support of DoD and then avionics, aerospace, and then also from a cryptography side, post-quantum cryptographic resilience in the cyberspace, which is what I focus a lot on within the public sector. So thanks for having me.

Aaron Crow (1:52.928): Now tell all the dummies out there, including myself, what the hell all that means? Like what is quantum? What are we doing? Like, how are we protecting? Like, what is this going to mean to me as an individual, as an asset owner, protecting ITOT? Like what give us the download, you know, 90 second dummy version of that.

Kat Wang (2:11.796): So wow, all right. So the way that we understand functions today of a class of computing is very binary, right? Zeros and ones. Think about it as an on-off switch, either it's on or it's off, or it's thumbs up, thumbs down. Quantum basically assumes that these two states are not the only states that it can be in. You can actually do a 360 rotation.

Kat Wang (2:38.292): down, et cetera. The idea is like these quantum physics, like things can exist in multiple states at one time. And the reason why this is important from a quantum computing standpoint is instead of being a very linear model, like if I want to get from point A to point B and moving like up and down, going in a linear path, we can start thinking about it from a more lattice oriented path, right? Like a three dimensional cube perspective.

Kat Wang (3:6.232): And you can, what that allows you to do is start to do computations in rapid, rapid contexts. Now, what that means from a cyber perspective is from cryptography, a lot of the ways that we protect our information, we encrypt it today, that's gonna be at risk. If you are not using post-quantum, like once quantum computing computers exist, really,

Kat Wang (3:32.617): If you were not using resilient algorithms that help to prevent quantum computers from decrypting your data, then literally everything that a cyber adversary has stolen for the last 20 years can be decrypted much faster, order of magnitude faster, as opposed to like hundred years, we're talking in months, right? So when that happens, then the entire security layer that we have, the security infrastructure that we have breaks down. And that's very bad.

Kat Wang (4:2.273): It means that really the only way that you can be secure in a post digital transformation and a post quantum world is with this.

Aaron Crow (4:13.346): Yup.

Kat Wang (4:14.228): Yeah, welcome, welcome back to the 1890s. Am I right? So that's what it is in not so many words. So that's what I help to do in the public sector is preparing our federal civilian entities, our defense entities, with the ability to inventory every piece of high value asset that they have. And of course, there's going to be, I call cryptographic debt, very similar to tech debt, like as you grow,

Kat Wang (4:41.054): there's gonna be stuff that's plugged into your platforms and accessing your network that you don't know about. And being able to get visibility into that, see what type of cryptography you're using and if it's vulnerable today, allows you to fix that and then also wrap it with NIST standards, for example, for post-coin world.

Aaron Crow (5:1.068): Well, and with that, just like in OT and in everything, there's going to be a, you know, prioritization of things because, you know, as we know in this, in today's world, especially in the United States, everything is classified. and there's different levels of classification and there's compartmentalization and there's different, you know, all of that has changed. Right. So we're going to have, we're not going to be able to fix everything instantly.

Aaron Crow (5:23.438): And we're going to have to understand what are the most important things. Let's make sure that those are protected. You know, the nuclear codes and you know, the things on submarines and all that type of stuff is different than, my recipe that's in my Google drive that is encrypted. You know, my mom's, know, chocolate chip recipe is chocolate chip cookie recipe is not nearly as critical as, as the, you know, the nuclear code. So.

Kat Wang (5:48.837): The crow family mother dough recipe needs to be protected. Okay, when the apocalypse comes and people need to have their like, crusty bread, you're going to be thankful that you use post quantum safe algorithms lock that shit up. Okay? Yes. I agree with you. What we're ended up talking about though, is like a prioritization within a prioritization. Like, if you're going to build a house, you're going to have a list of priorities at a macro level, but then each one of those priorities is going to listen to another

Aaron Crow (5:53.102): That's right.

Aaron Crow (6:1.304): That's right. That's right. That's right.

Kat Wang (6:17.076): list of priorities. And they're in line with the rub. This is the reason why our cyber warriors, our cyber defenders are constantly trying to keep up and it sucks. Like it's so hard, right? And, and on top of that, you're fighting for budget, you're fighting for priority and reality. Like if, if this was any other scenario and if leadership and the rest of the organization was building in mechanisms to account for security.

Aaron Crow (6:27.052): Yeah. Yeah.

Kat Wang (6:46.222): and speed, then we would already be, every time you spend a dollar on something, you put a dollar over here for something else, for security, to protect that. But there's just too much. So that's part of the reason why I went to the grassroots cryptography, not only for today's vulnerabilities, but tomorrow's. It's like, at some point, let's say we were doing all the right things.

Kat Wang (7:12.696): every zero day that existed came from something that we didn't see coming. Right? Many of them we do, but some of them, a lot of them we don't. And it's because there are the economics are horribly flawed in favor of the adversaries. They have everything to gain and we have everything to lose. They have to win once we have to defend successfully every time that is losing economics. So when intrusion happens,

Aaron Crow (7:31.180): Yep. Yep.

Kat Wang (7:42.000): and all other cyber defenses fail, your cryptography cannot. Because if they take that treasure out of your castle, the only thing that's keeping them from using it is the lock around it.

Aaron Crow (7:55.279): Sure. Yeah. And you see that now looking at AI and hopefully I'll say hopefully, hopefully not as widespread in the public sector, in the government space, in the three loaded organizations. But you see, I think there was recently a dating app or something that was vibe coded using AI and all these women's profiles that were using this app that was released. There was no security whatsoever. Whoever developed it, they were not a

Aaron Crow (8:24.904): cyber or they didn't have a cyber lens behind it. The application was cool. It a fun UI, but all of the people's information, I think mainly women that were in this app, all of their public, all their personal information that they put into this app. Again, I think it was like a dating thing was released because there was no security around it. And when you're looking at something, you're just assuming, well, this is safe. Like it's on the app store. It's on, you know, there's no reason why I shouldn't be able to put those things in.

Aaron Crow (8:54.190): So those are also things that we have to really be considerate around because as AI comes in, it can be a tool to help us to code to make things function, but something working and something being ready for prime time, ready for release, or ready to be available to be used in the scenarios that we're talking about here when it comes to national security, when it comes to my electricity working, when it comes to getting on a train and making sure that the safety measures are...

Aaron Crow (9:20.118): are not impacted. Like we have to be really careful around those types of things. And AI can help us, but it can also hurt us if we're not thinking about cyber as part of the operation and functionality of those systems themselves.

Kat Wang (9:35.165): Okay, so I don't even know where to begin with all the gold that you just dropped. So okay, let me take it high level for a second. The I've discovered in doing these types of conversations a lot that there's really three key things that I want anybody who, you know, here's this podcast to take away from them. The first is

Kat Wang (9:56.657): know what your data is worth. And we can touch on that as it relates to that dating profile and any other app that we use in our daily lives. The second is secure the basics. And the third is educate your network. We inherently put all of the responsibility, the onus on the consumer, the user. When we are the most uneducated segment of them all, that seems

Kat Wang (10:24.233): horribly misbalanced. you're telling me you have to tell, you know, if your grandma decides that she's gonna join a dating app, you're gonna force her to be responsible for the data that she puts online. And those terms and conditions are debilitatingly long for a reason. Mostly for the legal jargon and the DPA, the data protection rules that exist within them. like,

Kat Wang (10:51.209): Something's gotta give. So that's why my third pillar is educate your network. AI, you wanna poke that bear? Okay, I have had so much coffee today. I am so ready to poke this bear. I'm gonna walk right up to him and punch it in the nose. here's my hot take on AI.

Aaron Crow (10:54.499): Mm-hmm.

Aaron Crow (10:58.220): Yeah.

Aaron Crow (11:5.870): There you go.

Kat Wang (11:13.871): It is, I really go between days where like, my life is so much better with AI in it. And then days where like, my God, it's going to happen. Skydance coming to take over and we're all going to die. So it's, whoo. I'll start with some very real examples. And I'll start with the psychology of it how America is approaching it versus the rest of the world.

Aaron Crow (11:21.410): Mm-hmm.

Aaron Crow (11:26.774): You

Kat Wang (11:43.509): So while Fortune 500 was focusing on AI to replace the workforce, our cyber adversaries said, cool, we're going to use the same AI to supercharge our workforce. So cyber adversaries didn't get leaner with AI, they got meaner. And we did the complete opposite, right? It's like showing up to a knife fight.

Aaron Crow (11:52.952): Mm-hmm.

Kat Wang (12:13.511): with the Roomba and good intentions. It's like, okay, well, we have this here. It's not really AI. Well, it kind of is. It's like, no, it's not. But what we're going to do, no, it didn't. Whatever you thought it was going to do would replace the human being that you just had. It's not doing that. And by the way, you haven't applied any zero trust principles to it. So now it is more of a liability than the value that you got out of it. I am being very crude and very generalistic right now. So

Aaron Crow (12:34.946): Mm-hmm.

Kat Wang (12:43.902): I will say take that with a grain of salt, but that, is fundamentally, I think the problem here. We did not approach AI with the same creativity, and productivity for a force multiplication of our workforce, which is what it really needs to be. We took it and use it as a cost cutting mechanism. Right. And my friend Ali King likes to call it corporate ozambique.

Kat Wang (13:12.999): When you think you're getting leaner, you're actually losing the majority of your muscle mass.

Aaron Crow (13:17.452): Well, it'd be like back in the day when we were going to the moon and the NASA engineers, when they started using calculators instead of slide rules. It's not like we fired engineers because they were faster. We gave them a calculator so they could be better at their job. We didn't look at it as a way to, well, I only need two engineers now. I don't need 10 because that guy can do work way faster. So why are we looking at AI any different? I remember being in...

Aaron Crow (13:43.669): in calculus class in my freshman year of college. And I have this big TI-92 calculator that's the size of a tablet nowadays. I'm just dating myself because this is back in the early 90s. But I was able to use that in my math class. And you could actually put in a calculus equation in that program, in that calculator, and it would give you the answer. But it didn't matter that I knew what the question was and I knew what the answer was. But I still had to show my work to get from A to B. So it helped me to be able to

Aaron Crow (14:13.358): to check my work, it helped me to be able to be more confident in my work, but I still had to understand the 59 steps between the question and the answer to be able to show that I can get from A to B. And that's the way I see, that's the way I use AI, it's enhancement, it should be a 10x multiplier of your workforce, not trying to, well, I don't need those people anymore, or I can have fewer of them because I can just AI everything. Like that is so short-sighted in all.

Aaron Crow (14:43.638): avenues.

Kat Wang (14:44.709): And I have, it is a full time job. And this is coming from someone of sound mind and judgment who works in this space, right? Not some, another typical user that I would say who kind of dabbles a little bit in AI and has a free version of Gemini and Claude. I spend a considerable amount of time understanding how to write LLM prompts. And what I have found from it is like these

Kat Wang (15:13.973): prompts salt the responses. So if you are not careful with how you write in your prompt, they're going to give you the answer they think you want to hear. It's gotten so bad where I actually turn AI off in my search results. I don't want to see what they provide because in many cases, I'll ask something very basic like, does mint pair well with stone fruit?

Aaron Crow (15:17.166): Mm-hmm.

Aaron Crow (15:24.334): correct.

Kat Wang (15:43.145): this horrible example, but like something like that. And then AI will say, yes, it does. Okay, sure. And then I'll go, I'll actually go through all the listing results. And it gives me a much more like robust, like overview of the reason why it does or why it doesn't. That's a, that's a very benign situation. There are other times where you ask a question, if you are not careful with how you're leading the question in the prompt, they will

Aaron Crow (15:48.216): Sure. Yeah.

Kat Wang (16:11.690): LLMs will put that back into your response. And then when you ask them to validate that with sources, the sources are subpar. They would not hold water in any like paper if you were going to write on that or just it's blatantly wrong.

Aaron Crow (16:21.752): Mm-hmm.

Aaron Crow (16:29.868): Well, and it's so, again, you and I, I'm very similar. I'm using AI all day long, every day, in almost everything that I do, whether it's looking for, I'm doing research on whatever, right? But I have all the different AIs. Like I'm using them against each other, right? Intentionally, like I'm gonna try it on this one and I'm gonna put the same prompt in another one and see what comes out. Then I'm gonna adjust my prompt and do it again.

Kat Wang (16:51.111): yeah.

Aaron Crow (16:58.794): in a different window that doesn't have the history of the one that I just did to see if I get a different outcome. And then I'm taking the outcome from one and then pasting that into the other one and telling it, hey, this came from this other AI poke holes in it. See what you can find. Like, because this is what's going to happen. This is where from from a cyber perspective and from a from an overall risk perspective, we need to understand there's all these examples of here lately, there are too many to list, but we've seen, you know, grok go off.

Kat Wang (17:7.509): You're so...

Aaron Crow (17:28.172): where it went off and it was like creating images of Nazis that are multiracial. And then you've got the other side where it's SharePoint, or not SharePoint, but Co-Pilot within Microsoft that's perusing or previewing all of your emails. And a bad actor realized that they could put in a prompt in a hidden context or hidden text that the AI would then exfiltrate data from the AI.

Aaron Crow (17:57.747): You logged in as you so it has access to your files. It has access to your email like these are in the same thing just came out about Google, the Google one that's built into Gmail. Same thing is happening there, right? So we're just seeing these different attack vectors that we never thought about and everybody's like, well, that's really helpful. I've got Siri on my phone and I wanted to summarize my text messages like, do you? Are you sure about that?

Kat Wang (18:23.549): Or maybe they do, maybe their text messages are super benign, who knows. I love that you brought up the Google Gemini example, because I actually, so much, it's August 29th. So much happens that is reported on. Now granted, like my phone, your phone, that's already hyper-optimized to be able to show us these things on our news feeds, right? However, during Black Hat, Def Con, a lot of these announcements came out, right? August 6th, hackers.

Aaron Crow (18:26.488): That's right.

Aaron Crow (18:34.755): Mm-hmm.

Kat Wang (18:51.743): found a way to utilize Gemini to exploit a hidden malware prompt in the description line of a calendar invite. So you tell Gemini, summarize my calendar, it'll look through all your calendar invites and somebody just hid a executable command within the description and ends up, we talk about OT, ends up infiltrating into the network and into the smart home.

Kat Wang (19:20.689): opening the windows, turning off the lights and turning on the broiler. Like talk about it. OT, IOT, IOT. Like this is a absolute shit show. It shouldn't happen. Like if we think we can expect or predict any, every permutation in which AI is going to operate, we'd be wrong, right? And then not one day later, no, hackers.

Kat Wang (19:48.713): found a way to upload a poisoned file into ChatGPT and basically established a lot of connectors, like you said before, got access and using your login credentials and was able to initiate action without any user interaction and X won't write data. Three days after that, August 9th.

Kat Wang (20:6.481): Same thing, jailbreaks and agent flair do a zero day exploit, enable data theft from the AI agents. Like this is agentic AI vulnerability. And that was already integrated with cloud and IOT systems. So you can imagine the amount of fuckery that could happen from a situation like that. that's August 5th through August 9th, that's four days. And half of the, or arguably more of the world's hackers are in one place.

Aaron Crow (20:15.215): Mm-hmm.

Aaron Crow (20:35.019): Mm-hmm.

Kat Wang (20:35.965): When that all went down. So it has been what, 24 days since then? 20 days since the 9th? I shudder to think of all the different things that we are not, that we're missing, that we're not accounting for, that's not getting reported on, that's happening in the background continuously that we don't know about and won't figure out until later.

Aaron Crow (20:59.522): Yep. Well, and the frustrating part about most of those things is it's turned on by default. Like, it's not like I went into Google and I configured and I said, hey, I'd like it to summarize my emails or I'm prompting and asking it to it just started doing it automatically. Right. And it's doing it in Microsoft. It's doing it in Google workspace. It's doing it in your Gmail account. And it's by default doing those things. And in theory, they're like, well, it's not

Kat Wang (21:7.220): Hmm.

Aaron Crow (21:28.748): chat GPT so it's not public, it's just in your container, your company's container, but that doesn't matter. Like there's still risks to those things and we have to be asking these questions differently because you know, what got you here won't get you there. We're dealing with different problems and different attack vectors. We have to look at these problems differently than we have in the past. All these things, and I'm not saying that these things can't be solved and we can't.

Aaron Crow (21:52.973): get the value out of this like we talked about earlier, the calculator, the value add, we can absolutely do that. We just have to be thinking about it. Like you said a minute ago, how will this impact? Is there, what is the risk? Not if there's a risk. What is the risk from a cyber and availability and thinking about all those things? And what does this bring in? What risks does this bring in unintentionally that I'm not thinking about? That's how zero days happen. That's how all of these types of things happen is because we weren't thinking about it. We were thinking about the outcome, meaning we want the best

Aaron Crow (22:22.754): possible case scenario, not what are the off-ramps that happen that when it doesn't go as planned, what are the bad things that can happen on that same path?

Kat Wang (22:33.023): So you're absolutely right. This podcast is not to poo poo on AI, where they're required, like we need to have AI in our lives to function. I could talk a little bit more about the criticality AI provides from a survivability, like a humanity survivability standpoint leader. So that is key, but like,

Aaron Crow (22:39.598): Correct. Correct.

Aaron Crow (22:45.067): Absolutely.

Kat Wang (23:1.329): We also need to consider what truly needs to be connected to AI. Does it truly need to be connected? And once we draw the line in the sand of what the risks are, whatever risk is too much risk, let's analog, baby, let's not worry about it. What really scares me though, like of all of the things that AI can and cannot do, because the cannot do part is part of the same risk.

Aaron Crow (23:6.744): Mm-hmm.

Aaron Crow (23:18.434): Yep. Yep.

Kat Wang (23:31.125): profile as what it can do is this idea of data poisoning. We talk about the China problem a lot in my line of work. You probably too. And I will say this like many things about people who understand the methods, the TTPs of the PRC, but one thing

Aaron Crow (23:46.478): Mm-hmm.

Kat Wang (23:57.449): that is so apparent with all of the typhoons that we're seeing is their ability to play the long game, right? Like we have an incredible military force assembled and know it is unmatched with any other nation state in the world. But the question becomes, can we carry on? Can we support three different wars at one time? our resources and our attention severely divided, but also like what are the risks?

Kat Wang (24:25.257): where we no longer can respond and we talk about operational technology and critical infrastructure. But then if a lot of our platforms, and I don't just mean DoD, like DoD has a very strong approach, the sector for army that I just kind of better understood that their plan around AI and AI adoption within US Army, I think they're approaching it the right way. But.

Aaron Crow (24:50.499): Mm-hmm.

Kat Wang (24:52.105): the critical infrastructure that underpins their ability to operate is highly vulnerable. So why, if you pick a very strategic location, let's say an air force base, it depends on energy or power in order to open certain gates or to launch their planes. They'll just be sitting there, unable to mobilize. And the the data piece of it, this is why, you the first thing that I really want people to take away from this is know what your data is worth.

Aaron Crow (25:9.922): Yep. Yep.

Kat Wang (25:22.101): Some people say prioritize the data, certain things are worth protecting and other things have a shelf life, they don't matter. That isn't necessarily true in my opinion when it comes to the massive quantities of data that we could start aggregating and then turning that data into information. So a good example of this is, I talked about by a lot, behavioral warfare and cognitive warfare. Behavioral warfare is a fascinating one.

Aaron Crow (25:35.970): Mm-hmm.

Aaron Crow (25:45.422): Mm-hmm.

Kat Wang (25:50.941): What is the value of knowing what the temperature is in an office building? Most people would say probably nothing, but behaviorally, if you were to change that thermostat and do it in a way that people could no longer change it back, a lot of things start to happen. like psychological studies have shown that you disrupt sleep patterns and normal homeostasis.

Aaron Crow (26:9.698): Mm-hmm.

Aaron Crow (26:16.110): Mm-hmm.

Kat Wang (26:18.121): Physically like biologically the other thing is is you then disrupt interactions with each other like people start to get a little bit more irritated and angry possibly aggressive Right, right. I know your kid comes in and change a little thermostat. Are you gonna be happy camper? Nope. So all of those things happen and and here's the thing like that takes an adversary coming in and changing it all purposefully and then disabling

Aaron Crow (26:26.784): Mm. I do.

Aaron Crow (26:33.782): No.

Kat Wang (26:45.650): the changes. And we all know, and sometimes these military facilities, are inaccessible, or it is very difficult to get any technician out there to fix that. That is in a scenario where the adversary actually does something. We already have situations where these dibs, these military bases have problems with thermostats because they weren't installed properly, and they already are sitting in a hot box and being very uncomfortable with these types of working situations, right? So that's...

Kat Wang (27:13.383): That's one example of disabling sensors or executing behavioral warfare. But what if cyber-affirms decided to actually change the data itself? Like slowly salt or manipulate the sensor data within, I don't know, something like critical infrastructure and start to disrupt that. What happens? And let's look at imaging data. Turns out...

Aaron Crow (27:21.560): Mm-hmm.

Kat Wang (27:42.269): that there is only one pixel difference in some images that will turn AI from recognizing a missile launcher to a civilian truck.

Aaron Crow (27:53.548): Mm-hmm.

Aaron Crow (27:56.270): You

Kat Wang (27:56.487): It really calls to question our entire concept of truth, right? So there's so many examples where data contamination and poisoning of AI data, the stuff that's already trained on, is a real risk. And that would be a classic PRC move, in my opinion, for the long game.

Aaron Crow (28:19.074): Well, and we know firsthand now, you know, how do we, it's all security in depth, right? So we look at all these critical places you're talking about, whether it's a power plant or it's a base or whatever, you know, it's so easy now using AI to create frauds, fakes, deep fakes, voice, video, all these things are getting so good. We know we're gonna have to, and that's why I'm talking the quantum thing. Like we haven't even stepped into that realm yet. We're just talking about things where, you know, you can literally,

Aaron Crow (28:48.714): Now, especially think about this. So I've had 72 episodes, I believe, of this podcast on the internet. That's video and audio. I talk a lot. I'm a professional mic, good quality mic, right? So they have high quality versions of my voice. It would not take very long to create a clone or a fake, a deep fake of me, right? I know firsthand celebrities that are getting deep fakes and it's not hard to do.

Aaron Crow (29:18.732): Right? So imagine that at scale or, know, you think about a penetration test or a physical, somebody, a bad actor trying to get access to something they can have, they can represent your boss and say, Hey, cat, Aaron's going to be at this facility tomorrow. Make sure that you've put him on the visitor list. So they don't even have to hack the visitor list. They just have to social engineer it so that you put them on the list and then they're let in the front door. Right? There's, there's an infinite number of these scenarios.

Aaron Crow (29:45.945): But this is the type of thing as a defender, as a warfighter, as a cyber protector, any of these types of things, we have to start thinking about these things. And unfortunately, what I see is so many of these organizations, these people, these groups, these companies, they don't have enough money, time or effort to really even focus on it. You walk in and they've got a firewall within any any role. So it doesn't matter if quantum's coming. They already have the front door open. You don't need quantum to get in there. You just walk in.

Kat Wang (30:16.351): quantum. a quantum has just as much like the thing is, we got to stop we got to stop thinking about things as positive or negative. They are nice, right? They are neither of those things. And the quantum on top like the the the fuel of quantum on top of the fire that already is AI is what I call the AI quantum collision. It's like feeding gremlins after midnight. It's like it's so so bad. And what you're saying from a deep fake perspective,

Aaron Crow (30:23.694): Correct. 100%.

Aaron Crow (30:35.576): Yep.

Kat Wang (30:43.879): This is happening today. Like, I don't know if anybody heard about the crypto, the deep fake crypto heist in Hong Kong in 2025 earlier this year. Hong Kong finance worker joined a seemingly ordinary call, right? Zoom call. Everyone looks familiar. Voices all sounded right. Turns out that every single participant on the video call was a deep fake. It was generated using stolen audio clips and AI.

Aaron Crow (30:57.486): Mm-hmm.

Kat Wang (31:9.949): So it ended up resulting in an 18 and a half, think, million dollar crypto high siphoned out of the company's accounts, legitimately approved by someone who thought that they were getting the authorization to do so by their executive leaders. Right, now that was an instance where it was all audio. I think the cameras were off in that case. So every time I get on a video call with someone, I can't see their face, very skeptical. Who are you?

Aaron Crow (31:31.202): Mm-hmm.

Kat Wang (31:38.293): But you know what, it turns out it doesn't even matter because in 2024, I was attending, in October, so almost a year ago now, I was attending the Singaporean International Cyber Week Conference. And I walked up to a booth and they said, do you want to be part of a little demonstration, ma'am? I'm like, I guess, sure. Actually, Erin, do you have that photo? Are we able to bring it up?

Aaron Crow (32:6.562): yes. Yeah, I'll bring it up. Keep talking, I'll bring it up.

Kat Wang (32:9.841): Okay, so when that happened, this was a continuous like monitoring conmon platform that allowed you to med the IT at the sock level, make sure that every connection and interaction that you were having was truly authentic, given the amount of deep fakes that are occurring for social engineering and other types of cyber crime. So they had a toggle they said, who do you want to be? Well, there's like

Kat Wang (32:36.559): four men and then one woman. The one woman use case was Taylor Swift. And for today's purposes, since she just didn't engage, I thought this was a pop culture homage to her now that she's engaged. They in real time took a typical camera that was on top of a, wasn't even a laptop, it was a camera that was embedded into a smart TV, pointed it at my face and had me talk for five minutes while

Kat Wang (33:5.833): they superimposed Taylor Swift's entire face onto mine and it looked exactly the same. And the mouth matched up perfectly. This is what I looked at like that day and this is what they did when they superimposed her face onto my body. So if you go to the next one, I think there's another one that shows you that the whole, the setup that they had going there.

Kat Wang (33:34.751): but you could be Keanu Reeves, you could have been a former president. Like they have a lot of different options. It's insane that you can just have these done. And your tablet can do this today, right? You can have those deep fake apps on your phone, I think on your iPads, and children play with this all the time.

Aaron Crow (33:55.331): Yep. And it's not super complex, but they're able to do it. And like you said, in near real time. So this is the other one showing the actual system that they had there.

Kat Wang (34:5.949): Yep. So that's the entire cohort standing behind me. That's me standing right there. And then this is the face that they cut out just to be able to show. this technology, they're saying that it could do is essentially detect if it was a deep fake that was happening on a call. this is a year ago. This is in Singapore. This is another reason why having global perspective and understanding how other governments are approaching quantum, AI, supply chain, all of these

Kat Wang (34:35.667): key threat vectors within cyber right now is so important because understanding how other people are approaching it will give us inspiration and maybe, you know, incentive to follow suit and prioritize some of these things in kind.

Aaron Crow (34:53.378): Yeah, I mean, it's terrifying to think about that, right? As you can literally, as you're standing there, put your face, your body, you're standing there, remove everything, put your face over it, it moves to your mouth, all that type of stuff in near real time. And again, this was a year ago and we know how fast things are changing now. So if they could do that a year ago, what could they do today? Make it even more realistic. So these are all the types of things and this is not supposed to be a, I don't think either of us are trying to say, A, you should be terrified of life.

Aaron Crow (35:22.124): You should have a healthy fear and questioning attitude. This is something I've been teaching and as I build up teams, I've always done this all the way back to, you know, beginning of my cyber career of have a questioning attitude. Don't take it at face value just because the expert I'm doing air quotes. If you're just listening, the expert says that it's a certain way. Ask for proof. Trust, but verify like ask for.

Aaron Crow (35:47.545): How did you get to that thing? I wanna see it myself. I wanna understand how this thing works. I wanna understand, have you considered all the risks? Or are they just saying, well, it worked 99 times out of 100, it was fine. Okay, what about that other time? What did it do then? Those are the things that we need to think about, because nothing is perfect. And there's always a risk. Risk is never zero, and it never will be zero. So we need to think about all of these big picture problems.

Aaron Crow (36:15.046): and think about that as we're solving. So, you you look at OT, you look at critical infrastructure, you look at DoD and a lot of these places, sometimes we just need to keep it simple. Like, do we need AI to control the thermostat? Do we need to have a smart thermostat? Like, there's a benefit to it for sure, but there's also a risk. So we just need to ask the question. It doesn't mean the answer's no, it just means, are we understanding what the risks are and what the reward is, and is that risk worth the reward of automating, of putting the thing online?

Aaron Crow (36:44.014): You know, you can have your refrigerator be connected to the internet. Okay. Do you really need it to be done? Like mine are not. Not because I'm like a weirdo. It's just because I have a refrigerator that works and I just don't see a use case where I need to have my freaking fridge connected to the internet telling me how much milk I have or don't have. Like I have three kids. They're going to tell me themselves. I don't need to connect to the internet for that to happen, right?

Kat Wang (37:3.829): You

Kat Wang (37:12.233): securing the basics too. Like at some point we're going to have to buckle and say, all right, fine. God damn it. Fine. If you really need to have this connected to the internet or have AI injected into this forcefully injected into it, then secure the basics. Like if you cannot control whether or not this happens, I get it. Just make sure that you monitor everything that's coming through from a cryptography, like a data asset standpoint for data resilience. Cause like

Aaron Crow (37:25.484): Right. Right.

Kat Wang (37:39.993): Everything comes back to data. also want to say economic security is national security. Data is not just like a digital clone of our identities. It's not just our credit card numbers, our bank account statements, or, you know, our daily habits. It is intellectual property. Like

Kat Wang (38:9.169): The amount, the speed at which certain innovations can happen now just by hitting the market. We saw this between DeepSeek and NVIDIA. Like the speed at which a lot of these things could catch up, especially from other states, nations in relation to the US is already shortening substantially. Why should I as a nation state or cyber adversary, invest in my own?

Aaron Crow (38:29.192): huh.

Kat Wang (38:37.357): when I can go out and steal it from someone else. So secure the basics. It's a huge priority for any nation that is developing their own innovation.

Aaron Crow (38:49.542): It's one thing. a lot of my career I've been spent in power utility and I've supported every kind of power generation, substation, critical manufacturing, anything critical infrastructure I've been part of. And one thing I really loved and at the time I didn't understand it when I first got into the industry. You go to a nuclear power plant. I supported, I owned the OT cybersecurity and the networking and really the...

Aaron Crow (39:15.374): the functionality of the plant from that perspective, not how the plant ran, but how do we get data in and out? How does all that kind of stuff? And in it, you walk into a power plant, especially a nuclear power plant, it looks like it's from the fifties. Like the panels look like what you saw when we launched demand to the moon, apparently, in the sixties. So it's. But, know, exactly, conspiracies. Right.

Kat Wang (39:38.277): We're a lot of ground today, Erin. A lot of ground. Conspiracy!

Aaron Crow (39:45.543): So in the 60s like so a lot of they have you know, primary secondary and tertiary systems They have their primary system is analog and it's always going to be analog. That's what their license going right? So they also have digital stuff in there, but it's only in a secondary or tertiary Setup, right? So if all the other things fail so like what you talked about How can I display this information and and and fake the numbers, right? That's that's what stuck snap was right. It was stuck snap was

Kat Wang (40:12.693): Mm-hmm.

Aaron Crow (40:14.198): The control room, the operators, everything looked fine. But in the actual outside, it was spinning out of control because it wanted to cause damage, right? But everything on the screens that the operators were seeing was green. It's thumbs up. Everything's running as expected. So it was saying nominal and it was actually experiencing not nominal. So it was adjusting those numbers. But when you have when you design a system like what we have in nuclear and obviously you can't design tertiary systems and everything because it's expensive, it's complex, et cetera. But

Aaron Crow (40:43.082): In our critical environments, we need to think about that because if the digital system, it's very easy to manipulate the numbers that are coming in. But when it is hardwired and there's a physical connection to the position on a valve, it's really hard to fake that, Almost not impossible, but almost impossible, right? You have to have physical access and you have to physically go change set points and, you know, re you know.

Aaron Crow (41:8.546): calibrate things so that the calibration is off. Like there's a lot of steps that you could do that in a power plant or any place that has a lot of those types of sensors. It would take a huge effort and you would have to incrementally do it. So nobody noticed it. Like it would, it would be a big effort. But when I digitize all of this and I could do it with AI or a nation state attacker can just go do it with a, with a malware. it becomes a lot easier. So all this to say, we need to think about these problems, these big problems a little differently and, and, and prioritize.

Aaron Crow (41:38.607): the functionality of what we're protecting. And is it again, going back to my chocolate chip cookie recipe, or is it the nuclear football? Like those are two different scenarios. We should approach, I shouldn't spend the same amount of effort or money or resources to protect one as the other. It's the same as with, I can have two PLCs. They have the same risks to my environment. One is controlling a turbine. One is controlling the ice machine in the break room. They have the same risk, but they're not the same.

Kat Wang (42:3.476): Yeah.

Aaron Crow (42:6.478): priority or impact to my business. although people would get pissed off if the ice machine stops working, it's not going to shut my business down.

Kat Wang (42:13.853): Right, right. Do you ever watch, did you ever watch the movie Live Free or Die Hard? It is shocking how much pop culture these days are echoing. mean, there's certain things that are just wildly off for entertainment purposes, like Mission Impossible. But the most recent one, Dead Reckoning, I almost said Red Dead Reckoning, that's one, Redemption, that's the one.

Aaron Crow (42:20.557): Yes.

Aaron Crow (42:39.436): Red Dead Redemption. Yeah.

Kat Wang (42:43.497): But there are certain things about that that at least in theory are very real and very big concerns, right? Although I feel like it is very derivative of Skynet and Terminator. Either way, I do think that some level of awareness, this is where the third pillar, educate your network. There's some very practical things that you can do. And it's really hard to know where to start because on the one hand,

Aaron Crow (42:56.856): Yep. Yep.

Kat Wang (43:12.905): the children, the children are on these devices, they do not understand, they're vulnerable demographic, they do not understand any of the risks that are coming in. And we do not introduce these concepts to them until, in my opinion, way too late. Like even the concept of media literacy, we talked about cognitive warfare too, and informational warfare is another form, behavioral informational cognitive.

Aaron Crow (43:27.662): All right. Yep.

Kat Wang (43:39.059): Like these are all things that kids need to be aware of. And I hate that we're putting it on them, but they do need to know. But that's one thing that I think we need to invest more in. And anybody watching this, you are now part of this fight. If you care, please make sure that you tell your children or you tell your immediate network.

Aaron Crow (43:59.043): Mm-hmm.

Kat Wang (44:5.491): the very basic foundations of having two factor authentication, having strong pass phrases, not reusing passwords, really making sure when you download something that you read the reviews or determine whether not you truly need that does whatever you're thinking about modernizing from a digital standpoint, you truly need to be modernized. Don't have a TV in your bedroom. Don't put Alexa in your bedroom. Like they're always listening.

Aaron Crow (44:30.434): Yeah.

Kat Wang (44:33.234): The education piece of it is really critical. I have people over the age of 65 constantly contacting me like, Cat, is this fishing? I'm like, I don't know, but unless you absolutely need to click on it, don't.

Aaron Crow (44:45.986): Don't. And they've gotten, you know, diving into that really quick. I've talked about this a few times, but you know, I've been doing this for 30 years. You know, I've, I'm, I'm, I see it firsthand. I'm building, fishing, you know, scenarios, tabletop exercises, and I almost got hit this last year. I got a phone call and a voicemail and it was from the, Travis County Sheriff's Department or Williamson County Sheriff's Department where I live.

Aaron Crow (45:15.278): and it was a, sounded like a good old Texas boy. called him back. I had a conversation with him. The phone number that came up was from the police department. Cause I checked all of that and I did my due diligence before I even talked to him. I called that number. they had somehow gotten that number and, and, and spoofed it. and you know, they, they had this elaborate. Scheme to, you know, try to get money. and it had me for, cause again, my spotty sensors are going off from the beginning.

Aaron Crow (45:44.547): But they didn't push too hard in the beginning. They were like, hey, you have to show up at the county courthouse. You have to take care of this or there'll be an arrest warrant issued and we're to have to come after you. Like, okay, well, what do I do? Well, you have to go to the courthouse or go to the sheriff's department. Like, okay, well, that's reasonable. A hacker's not going to send me to or a bad actor is not going to send me to the sheriff's department. It's very unlikely the sheriff's office is going to be the one doing this, right? So I go through the process like, yeah, but you have to show up with...

Aaron Crow (46:11.532): with money because you have to pay it while you're there. And if you don't, we'd have to take you into custody. But if you do, then we'll just take it all care of it. Right. And it was all because they say that I had a summons for jury duty, which I did, but I filled out the form and got excused from jury duty. I think I was traveling that week or something. So they knew enough that I had a jury duty summons and that I was, and I did not get, well, I guess they probably just assumed I didn't go.

Aaron Crow (46:39.918): or they were trying, hoping that I didn't go. Anyways, long story short, they got me to the point where I took money out of my bank account, went to an ATM and got cash out. And it wasn't until the next step where, cause again, my spotty senses are going off the entire time, but they just, they didn't push too hard and then nothing was out of order. Cause I was going to take this money and go to the sheriff's office. So there was, there was no risk. And then when I took the money out, they're like, okay, now go to the CVS and buy a gift card. And I'm like, you're.

Kat Wang (47:6.206): Give card.

Aaron Crow (47:9.058): fucking kidding me. No, I'm not doing that idiot.

Kat Wang (47:13.346): Wow. Okay, so I have a little...

Aaron Crow (47:15.202): But I see this all the time, right? And I still almost got hit by it because they hadn't, they were smart enough to not push it in the beginning. anyways, it's just going to say, nobody's against this. It can happen to anyone. We have to be on guard at all times.

Kat Wang (47:32.341): So a little confession, that was me, sorry. No, it wasn't. I know. So this goes back to my exact point about what data is protecting. You can protect all these things that you think are high value and then you don't protect other things that you don't. Like the fact that they knew you had a jury summons makes me believe that they infiltrated some kind of local sheriff department or a judiciary system like summons a database.

Aaron Crow (47:34.574): Yeah

Aaron Crow (47:37.752): Damn you, cat!

Kat Wang (47:59.667): because they know that they have your name, they have your address, they have your county that you're in, they know when you're supposed to show up and they have like, everything else is publicly accessible information. They could see what is the nearest sheriff's office, which is near CVS, right? There's probably only a few different banking institutions, including local credit unions that you're probably part of like.

Kat Wang (48:24.597): the amount of social engineering that kind of comes together once you have that piece of information. People will be like, well, who really cares about jury summons? This is why. They will be able find any opportunity to use that data to their advantage. And this is why we have to know what our data is worth. It starts with us.

Aaron Crow (48:40.398): Absolutely. It just goes to show everything. When I worked in the power industry, I had security clearance because I had to be able to talk with all the government agencies and give us intel on things. I was also badged at a nuclear facility which has its own background process and both of those databases were hacked by China. As we know, this was years ago.

Kat Wang (48:56.959): Thank

Aaron Crow (49:5.066): all everybody that had security clearance or was going after security clearance, they have all of that information. Like the funny thing is, is with the nuclear side, not only did they do a government level topsy or security clearance background check, but they also had a you had to go through a psyche vows. You had to have it. You answered a five hundred question questionnaire and then had to have a one hour interview with a with a psychiatrist to make sure that you weren't crazy.

Aaron Crow (49:34.467): I guess somehow I passed. don't know how, you know, you I don't know. I must have tricked the system. I said on attack or something. I'm not sure. But you know, I was able, you so they have not only my personal information, but all of my psyche valve, like all of that stuff had to have been in there, which means now what do they not know about me? Like they probably know more about me than my wife knows about me after 17 years. So it's just, it's, it's terrifying. And that to be said, assume they know.

Kat Wang (49:36.489): How does that happen?

Aaron Crow (50:3.426): So how can I protect against that knowing that they know the answers, knowing that they know my social security number, where I live, where I drive, what cars I have, that I went to jury duty or didn't, assume all of those things, and then look at every situation or when you're designing a system or you're trying to defend the system, assuming the worst case scenario, not the best case scenario.

Kat Wang (50:24.297): Yeah, agreed.

Aaron Crow (50:26.542): Awesome, so 10 the next 5 to 10 years. I asked this question everybody next 5 to 10 years. What's one thing come up over and we've already talked about a lot of this. What's one thing coming up over the rise and you see that's that's terrifying and maybe one thing that's exciting and you know it could be less than 5 or 10 years like Sky Nats going to take over next year. You know what's your predictions?

Kat Wang (50:50.323): I'll give you one answer that applies to both. What excites, what are our greatest challenge or threat is also possibly our greatest opportunity and all it entirely depends on how we as a.

Kat Wang (51:5.877): I don't know, global, as far as humanity goes, approaches the situation, there is a global population decline, a massive one. Anybody who hasn't read Peter Zion's book, The End of the World is Just the Beginning, it's a real pick me up, but it's very important, it's very important. And they go through the systematic dismantling of every major industry that underpins humanity today because of this global population decline. In short,

Aaron Crow (51:14.786): Yeah. Yep.

Kat Wang (51:36.009): The way the birth rates have gone historically until now have been increasing over time. And anthropologically, you do see this inside of societies. Like at certain level, the birth rates and the population will peter out. They will reach a kind of like a tipping point. And now we're seeing that come down. And what happens then when it comes down is a lot of the people that are around to do certain jobs are just simply not going to be there anymore. Economies are going to buckle.

Kat Wang (52:6.075): and certain countries will be in severe trouble. There are some people out there that are not too quote unquote worried about the China problem because of this particular point is the Chinese and is a along with Russia and it's unfortunately some cases, Japan, other societies where the population decline is so severe and so significant that their country simply won't be able to sustain with the amount of people that they have anymore.

Kat Wang (52:36.447): So that, how does that turn into both our opportunity and our biggest challenge? That's where I wanna see AI. I wanna see AI making smarter predictions about how to harvest food so that we are not so dependent on a workforce to do a lot of that harvesting. Some of that is already in place, but we need to see that increase 10-fold. We need to see more in the way of

Aaron Crow (52:56.738): Mm-hmm. Yeah.

Kat Wang (53:6.005): efficiencies for transportation and I'm talking about a massive transportation on the shipping containers that get to and from and just general production of goods in general. A lot of automation happening in the manufacturing space, in the agriculture space, that needs to happen. We need to then as a species learn how to level up in every sense of the word.

Aaron Crow (53:15.704): Mm-hmm.

Kat Wang (53:34.387): think about all of the most interesting, fun, utopian aspects of the sci-fi movies, like Next Generation, Star Trek Next Generation is great example, right? Where the concept of money doesn't really exist anymore. We're back on academic achievement, a contribution to the exploration and the growth of the human population.

Aaron Crow (53:44.747): Yeah.

Aaron Crow (53:52.013): Mm-mm.

Kat Wang (54:3.057): outside of this planet, that type of stuff, we see this in some cases with Occupy Mars. I want to see people skilling up, utilizing AI to think more strategically about big world problems that have scale. We do this at our company today. So the ability to do quantum simulations to find cures for undruggable diseases. This will save lives in ways that we cannot do right now because the pharmaceutical industry is

Aaron Crow (54:26.701): Mm-hmm.

Kat Wang (54:32.773): very expensive with a very high failure rate, right? You're not paying for your meds being so expensive because of the drugs that worked. You're paying for the meds that didn't work. So if we can reduce that and we keep the small dwindling population that we have globally healthier and living longer with a better quality of life, then that helps to reduce the impact that we see at a larger scale. So

Aaron Crow (54:41.368): Right. Right.

Kat Wang (55:0.947): we need to stop thinking about AI is like, it's gonna help us create efficiency in the workforce. No, we need to turn every single human being that can read and write into a scientist, into a strategist, into someone who could solve big problems in order for us to survive as a species.

Aaron Crow (55:23.214): I love it. And I see one of the big values in that space too, right? You think about, I was just listening to, I listen to a lot of podcasts, just listening to podcasts about health and longevity and I'm not so short in the tooth anymore. I got the gray hairs, all the things, right? But if you think about the way medicine has been done, and there's a reason why they call it a practice, we've really had to build medicine around the masses.

Aaron Crow (55:49.611): even though it doesn't specifically work for me because you and I may have the same problem, but something may work for you and not work for me or vice versa. Imagine the fact if we could feed all that data personalized into an AI and build a custom solution for me that would would benefit and help me to get to the same place to fix my ailments or whatever that may look like. And, you know, fix my hormone levels. And I mean, you're not a six foot

Aaron Crow (56:19.234): you know, to man with a red beard like I am like. So obviously there there are differences between the two of us. So obviously some things are going to work for me enough for you. So if we can imagine that at a larger scale, like those are things I'm excited about what tomorrow looks like. The ability to use that for good. But obviously we have to be careful because then it could also, you be bad and that it does something it says is good. We we inject it and then, you know, I fall over dead. Right.

Aaron Crow (56:46.294): Some people may like that, but that's for another time.

Kat Wang (56:49.813): Not on my watch, Aaron, not on my watch. No, and you're 100 % right. bringing it back to protect it all, cyber underpins that. This is how we protect it from being bad. And we're never going to be perfect, but just lock that shit up. Get those crown jewels in a vault and make sure that that vault is secure. I've seen people hacking into some of the most secure vaults in the world with a wet paper towel and a clothesline.

Aaron Crow (57:12.588): Exactly.

Aaron Crow (57:19.810): Yep. Yep.

Kat Wang (57:20.405): clothes hanger, right? Like this is, this is the world that we live in. You don't have to be high tech to be a, uh, you know, a hacker. So, uh, you're, you're absolutely right. There's, there's great, great opportunities ahead. Let's just make sure the security keeps up with the speed.

Aaron Crow (57:29.197): Absolutely.

Aaron Crow (57:36.460): Absolutely. Awesome. Hey, thank you for your time. What is a call to action? How can people get more information about what you guys do? Maybe dive into more about quantum and all the cool stuff that you guys are doing and working on.

Kat Wang (57:47.559): Yeah, so I am very active on LinkedIn. tried to shockingly I find myself on panels, on stage or on podcasts twice a week these days. So it's a great way to kind of keep up on the events that I'll be attending. If you want to link up, send me notes and messages. I'm happy to connect. And yeah, just thank you everybody for being part of this and caring enough to want to invest in the security of your future.

Aaron Crow (58:15.436): Yeah, absolutely. Thank you so much, Kat. I'm sure we will see each other again and until then have a good weekend.

Kat Wang (58:21.599): Thanks, Erin.

Transcript lightly edited for readability.