EP 282: AI’s Role in Scam Detection and Prevention

This is the Everyday AI show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business and everyday life.

Scams are not easy to detect anymore. They used to be extremely easy to spot. You could spot them miles away. But with AI means more sophisticated scams. So now more than ever, it's so important for your business to really take scam scam detection and prevention extremely seriously when you're talking about cyber security. So that's what we're going to be talking about today on Everyday AI. So thanks for joining us. I'm excited for today's conversation. If you're new here, my name is Jordan, I'm the host and we do this every day and we bring this to you live, unedited, pretty much unscripted as well, helping you understand AI to grow your company and to grow your career. So we are going to be Talking today on AI's role in scam detection and prevention with a guest. But before we do, we're going to start as we do every single day, going over the AI news. So if you haven't already, make sure you go to your everyday AI.com and sign up for the free daily newsletter where we're going to be going over both today's episode and a lot of AI news because there's more than we can get to in our little recap here. But let's, let's start at the top. So AI safety researcher has left OpenAI for rival anthropic so the former lead safety researcher at OpenAI, Jan Leakey, has joined rival AI startup Anthropic after resigning from OpenAI earlier this month. Yeah, just in two weeks span we saw a very, very public resignation from Leakey on Twitter and he's already joined rival Anthropic. So Anthropic is obviously backed by Amazon with $4 billion in funding from them and they focus on Super Alignment. So in his new role, he'll be focusing on super alignment mission, scalable oversight and weak to strong generalization and automated alignment research. So OpenAI did dissolve their Super Alignment group, which was strange and emphasized the importance of AI safety in the tech sector at the same time. All right, our next piece of AI news is, well, Nvidia is closing in on Apple and could quickly become the second largest company in the United States. So Nvidia's shares surged by about 6%, a huge jump, reaching a record high and bringing its market capitalization to 2.8 trillion, just shy of Apple's 2.9 trillion valuation. So who knows, that could happen today or in about 30 minutes when the markets open here in the US. So following a strong second quarter revenue forecast, Nvidia stock did did hit an intraday peak of $1,149, up 9. Sorry, up 8% during the session. Nvidia's remarkable growth this year has seen its stock or its share value more than double and is obviously attributed to its success in the AI industry and a significant five fold increase in the data center segment. Huge. If, if only someone out there, you know, I don't know, a year ago told you that Nvidia was the most important company in the US that no one was talking about. Oh wait, we did that all right. And last but not least, a former OpenAI board member has dished on why Sam Altman was fired from the company. So in a podcast interview, former OpenAI board member Helen Toner gave some more details on the drama that went down in November 2023 when we saw Sam Altman getting fired and rehired days later. So if you did miss that, the OpenAI OpenAI CEO and co founder Sam Altman was unexpectedly fired n November 17th, but then was rehired November 20th three days later. So at the time the board said Altman had been, quote, consistently not candid. And so Toner was one that voted to remove him, him being Sam Altman, and then she later resigned from the board. So in this new interview, Toner revealed some new details and said the board was not informed about the launch of Chat GPT and that they found out about it on Twitter, which is wild to think about. Toner also accused Altman of lying to the board on multiple occasions and said the board at that time could no longer trust him. She also detailed other employees accounts of Altman's reported toxic atmosphere and psychological abuse. So the drama is always brewing in AI, but that's, hey, that's why we cover it every single day. So we're not here to chat about AI news all day, we're here to talk, which I'm extremely excited about. AI's role in. In scam. Sorry, in scam detection and prevention. So if you're joining us live, get your questions in now. So thanks for everyone, Brian from Minnesota and Woozy from Kansas City, Fred and everyone else, thanks for joining us. But please help me in welcoming in our guests for today. I'm extremely excited to bring on to the show. Uh, there we go. We, we have him there. So uh, please help me welcome Yuri Vinos The Chief Innovation Officer at Aura. Yuri, thank you so much for joining the Everyday AI Show.

Thank you for having me.

All right, absolutely. So I mean, let's just start at the top. Tell me a little bit about what you do at Aura.

Right, so I drive innovation functions at Aura, which is a Boston based cyber security company for individuals and families. We help everyone stay safe online and it's just becoming such a massively difficult job to stay safe online. One of the things that I've actually done lately is scam protection technologies and scams are on the rise. And as you can imagine, I'm going to chat about using a lot of AI to help everyone be safer online. But as a matter of fact, a lot of bad guys are equally using AI, are equally empowered with all those technology and tools and they make our job much harder. So it's a very interesting dynamic and it's been, it's been up and down and we see a lot of changes within this several last years.

Yeah. And maybe walk us through those changes. I can only imagine what the role of a chief Innovation officer at a company providing cybersecurity looks like and how much it has changed over the past couple of years, especially with this, you know, kind of surgeons of, of large language models and generative AI. But, but Yuri, maybe just walk us through how has just your, not just your role, but how has the industry changed with now generative AI and large language models?

Yeah, absolutely, Jordan, and that is the right question. I think the nature of scams, you know, scams is not in new concept. We've been, we've all known scams for many years. A lot of time, a lot of times when I hear scam or something is a scam, we think about something like a spamming message or something ridiculously dumb, but that somehow we tend to click. We rarely think about extremely sophisticated attacks that you can barely recognize. Well now we got to the point where they got smarter and smarter every single day. So what we are seeing is that, and it all started with. And by the way, scams are on the rise, right. So we know that the impact that monetary impact and the amount of victims is growing year over year. So this problem is getting only bigger. We have first seen a very interesting tendency with mimic and writing styles and just a quick background information. So many decades ago everyone was trying to hack a system, a computer or a system. Well that now that it's, it becomes very difficult to hack a system. So now everyone tries to hack a human being or us typically by impersonating so their first goal is to impersonate someone you trust, whether it's bank representative, your friend, or anyone else, customer support, doing a refund, whatever this is, right? Financial institution, they just want to establish this trust and credibility. So they literally are trying to hack our cognitive ability or trust by default, sort of mental disposition. And there is a very interesting thing that is very hardwired in our. In our brains. We tend to authenticate a person over texting through their writing style. So, for example, if your mom writes you hey, Jordy. And that is how she starts every single text message. And there is someone that says mom, and the message starts with hey, Jordy. And then they might use some specific language in addition to that, you almost immediately start to ignore everything else, like you have complete trust over that message. And that is exactly what's being exploited right now by the scammers, as they are using large language models to mimic someone's writing style. All it takes to do, by the way, is to have several references for sample communication letters or examples of how someone is writing. So basically, you need to communicate with someone, and when once you know how they write and of course, better write to you, then they can mimic that style and then can try to impersonate someone that, you know.

You know, and I have so many questions, and I, I love that example that, that you gave, right? Like, yeah, someone could just very easily impersonate you just by looking at your. Your writing style. Right. Like, I have so much of my own writing online. I was a former journalist. It's out there, so it's pretty easy. But, you know, I'm curious because, you know, I think that previously scams maybe targeted larger companies, enterprise companies, because there was more to gain, right? And maybe that's why scams had to be more sophisticated. But, you know, I've even had friends who are small business owners, you know, not, you know, thousands of employees, but just small ones now being targeted by scams. You know, with AI, do companies of all size now, Yuri, need to be on the lookout. Whereas I think maybe before it was just larger companies. Like, are you seeing that in your role?

We definitely are. And this is anecdotally, like, I don't have a hard date on that, but I have a feeling that if you fall. Have you followed any of those rumors about losing jobs to AI Jordan?

Yeah, I might have talked about that once or twice here on the show.

I think this is what has happened to scammers. I think their job has been taken over by AI. And I'm joking. About this, of course, but I think what's happening is that in the earlier days we were seeing someone physically writing all those messages. So there were some sort of call center, chat center people that were trying to create all that communication. Right now that's not what we see happening right now. Most of those communications being automated. Think about someone Instagram account being hacked. What they do, they take everyone within the contact list, they analyze the profile of the contact that you have, and they create a tailored message using AI so you can send message from a hacked, hacked account, can send message to all your contact base within five minutes. Obviously this is automated. This is a large language model dealing.

You know, and I'm curious, maybe could you walk us through like an actual use case or example? You know, at Aura, I'm sure that you have many clients, some you can't talk about, some you can't. But maybe just walk us through typically what scam detection looks like for businesses. Because I can only imagine there's, you know, so many now different ways that, you know, companies can, can, you know, fall victim to these scams. Whereas maybe before it was a little easier to spot. Now I feel it can hit you from all sides with generative AI. But maybe could you just walk us through, you know, like a use case of, you know, essentially here's what companies like, our types of clients are seeing and here is how we specifically act against it.

Absolutely. So we have a consumer offering, which is a consumer app, that is a suite of tools that has everything you need in order to stay safe online, which is an absolutely state of the art set of technologies that will help you, including scam protection. The way we help businesses is by making every single employee of those businesses more protected. And we know that we all have some portion of our work on our personal devices. So we believe that by protecting every single employees and families of your employees, the overall security posture of the organization is going to be way stronger, way better. Now what we are offering within Aura is we have a message protection technology that filters all unknown incoming messages or messages from unknown context. We do the same thing for calls, which is where it gets very interesting as we can understand the intention of the call, we can understand the early signals of the call signals of the scam. Sorry. We can also understand that something funky is happening on the call in the middle of the conversation with an unknown number and then we can identify that as well. And lastly, we're just about to release our last technology that scans every single email in your email inbox. And we have so much our inbox are over cluttered. So we believe that you have to protect yourself within every single communication channel. And for every single communication channel, we have created a little AI assistant that will meticulously check every single thing you can possibly think of and see if that is similar to other scams, if that is trying to explode your psychological bias. Like is there any sense of urgency? Is there any psychological pressure happening? And all of the other things. Is this a first time you're receiving this email from that specific email? Does. Is that, is that email look similar to other email you're constantly exchanging messages with, but you have never received emails from this before? So all of those. I can keep going on and on and on. There are hundreds or dozens of those triggers that we're looking at. And of course it is so difficult for us to look at every single one of them for every incoming message. But it is so easy for AI to augment your ability to recognize riskier communications.

Yeah, exactly. I'm sure it's a challenging task to take on just because of how easy scams are now. And speaking of that, you know, maybe, maybe with some, a little bit of humor here, but I love this question here from, from woozy asking. And I'm gonna ask the, the, the flip side after we hear your answer. Yuri, but so woozy here asking, what's the worst scam attempt that you have seen or heard about? Because I'm sure that they're getting lazier and lazier and not very good because of AI, but maybe what's, what's the worst one that you've seen or heard about?

Can I, can I give you something that is somewhat related to our. But more like anecdotal.

Yeah, yeah, absolutely.

I was. Yeah. And it's probably more. I'm answering the question the worst scam, the most ridiculous scam I've seen happened in Indonesia very recently. I think a man started to. Someone has been using online dating and they found a potential partner through an online dating platform. And then they got married and the person realized that, I think it was like a week or two weeks after marriage that this was another man that he married and all of that was a scam. So someone was trying to impersonate someone else and this is how far it went, which was very, absolutely ridiculous. So this stroke me as just how creative some people are when they're dealing with scams in terms of the AI and the most ridiculous scam that I've seen in a more serious manner. I was pretty surprised by the Deep fakes. I think the deepfakes are being real and we can chat about this, Jordan, if you want.

Yeah, yeah, absolutely. Because you know, now, so there's, there's different things, right? So there's, there's AI clones, right, which a lot of companies are, are using, I think in smart, responsible ways, right, like to be able to provide training and personalized learning and development. So there's authorized ways that you can essentially, you know, clone yourself. But then there's these unauthorized way, right, which is deep fakes when someone without your permission makes, makes a version of you. So yeah, I mean, let's maybe talk about the technology side first and then we can get into maybe some things to look out for. Yuri. But you know, how has just the availability of this generative AI technology changed deepfakes because yeah, it seems like it's easier than ever before to create something like this.

Absolutely. So if you think about how human beings are communicating with each other, it's really, at least online it's really three things. It's texting, it's phone conversation or audio only conversation, and then it's video and audio conversation. So of course with texting it gets a little bit easier. It's getting more complicated with mimicking someone else's writing styles. But it is still manageable to detect the scam. Now with voice, things become murky, right? Because one can generate voice synthesizer using like less than a minute of someone's talking as a reference. And then people will think that, oh, but that's, that still, that still doesn't sound right. Right. You can still understand those little bits and pieces. So it's a voice synthesizer. Yes, that is right. But this technology is getting so much better, so aggressively that I'm not sure if that's going to be right in one year. And, and just to add to that, people add background noise to mask those imperfections of the voice synthesizer. So you might receive a call from your CEO and the phone number was also spoof. You're not using the call protection. Okay. You can get a spoof phone number, incoming phone number that is from a random number that says the name of your CEO and then someone who talks with a voice very similar to your CEO with a lot of background noise, like at New York street or something else. Like lot of background noise and says, hey, I just sent you something I need you to do. Blah, how likely you're gonna obey.

You know, and, and it, it might seem, you know, if you're not following, you know, if you're not following the space, something like this might seem far fetched but it's, it's really not. Right. Like so, so like I said, I've had friends, you know, even you know, get hit with these types of scams. You know, whether it's just email and you know, but multiple, on, on multiple platforms as, as well. So may. Can we talk about that? You know and maybe some, some safeguards that people need to take because I guess if you get a scam it might be easier to detect if it's one platform. But what happens then when it starts hitting us from multiple places and it does seem very sophisticated and maybe coordinated from multiple platforms. So maybe number one, how can business leaders look out for that? And then number two, what should they be doing about that? Or common, you know, common sense steps that they should be taking to prot themselves and their businesses.

Right. So the good news is that multi channel scams are still rare or much less frequent that a single channel scam. So my first advice to just everyone across the board is that as, as dark as it may sound, don't trust anything that you might feel might think be risky or suspicious. Like if there is a clicking on the link, sending the money or doing something involved like just you received a WhatsApp message, pick up a phone, call that that person and say hey, I received that. What, what do you mean? If you received an imessage, text them over, insta, whatever. But hacking two channels at the same time is much less likely in terms of the impersonating. So that, that is a very simple advice that all of us can follow and that will help hopefully will eliminate most of the scams. But of course for businesses I think the more targeted, sophisticated, multi channel attacks are becoming more frequent and then the question becomes like what do you do? So obviously you have to train people you work with. And I gotta say that cybersecurity trainings might be the most important trainings in the world. Like we all that that is a necessary evil. But I think there is a massive opportunity to make them much more entertaining, memorable, joyful and I think we should absolutely do that. So just, just be a little bit more empathetic with your workforce. No one wants to read a huge manual of how do. The question is how might we make this truly enjoyable, truly memorable? I think that's one thing. The second thing, obviously you have to have a person inside your organization that everyone knows they need to report. They can, they can write if they see something suspicious. So someone has to be physically present there to protect you. And I think with those, between those three things, two, fa, responsible person and of course some training in terms of critical thinking. Not all of us know about inability of trust. Like we used to trust everything we see. Now it's just not the case and we're not wired that way. So I think there has to be some training. Of course you should be, you should be safe. Safer.

You know, Yuri, you bring up a good point because I think that, you know, over the last, even with social media, you know, over the last 20 or 30 years, when it comes to consuming information from the Internet, I think we give it just a, a one off, you know, does this pass, you know, my own human detection, like yes or no? It seems like, you know, maybe that's a good habit that might be come a bad habit, right, where we just look at something quickly and we're like, yep, this seems legit. I trust it. It almost seems like we might have to rewire our brain as scams get more sophisticated to really look into like, hey, yeah, this, this, this video looks real. This, this voice looks real. This, this text looks real. But we might have to unlearn some things. So what are maybe some, some of those key things to look out for, for? So you already mentioned, you know, single channel scams. Hit them up on a different channel. But maybe what are some, maybe telltale signs or, you know, something that businesses should be looking out for that normally you're like, ah, I normally wouldn't look at this or look into this. So maybe what are some of those, not red flags, but some of those things that might be under the surface that we should be paying closer attention to. Are you still running in circles trying to figure out how to actually grow your business with AI? Maybe your company has been tinkering with large language models for a year or more, but can't really get traction to find ROI on Genai. Hey, this is Jordan Wilson, host of this very podcast. Companies like Adobe, Microsoft and Nvidia have partnered with us because they trust our expertise in educating the masses around generative AI to get ahead. And some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to use Genai. So whether you're looking for ChatGPT training for thousands or just need help building your front end AI strategy, you can partner with us too. Just like some of the biggest companies in the world do. Go to your everydayai.com partner to get in contact with our team or you can just click on the partner section of our website will help you stop running in those AI circles and help get your team ahead and build a straight path to ROI. On gen AI.

There are lots of signs of a scam and just, just I think adjusting to the reality that you cannot trust out of the gate to most of the think things you see and you hear is a good rule of thumb. And then of course, how can you be on, even on a higher alert? Right. So another good sign of scam is sense of urgency. Like if someone is asking you to do it right away, this might be exactly where you're, you should wait and hold off. If someone is applying any psychological pressure, like hey, I just had a call from CEO, you have to do it now. Don't buy that, just don't react to that. You have to learn to ignore that because that is more likely to be a sign of a scam. And this of course happens when you believe you are talking to someone you can trust. So calling back, involving anyone else within this organization, double check and being extra careful is of course, course my, my large advice. And then scams are being extremely creative anywhere from jobs interviews, you know, fake job postings, to sophisticated corporate scams. But it's difficult for me to chat about that because they are, the variety is, is high but, but the principle stays, always stay unchanged. Yeah.

And you know, even when, when you talk there about just the sophistication, I mean it's, it's getting really good and I think people need to be aware of it that even if you are a, you know, a smart, intelligent business person, you can easily be duped. Right. There was something we talked about on the show a couple months ago, I believe it was back in February where a finance worker essentially got on a zoom call with what he thought were all his co workers. It looked like them, talked like them, it was video and they were interfacing. Right. And this company, a very sophisticated scam, ended up, you know, essentially stealing $25 million from this company just because of the level of sophistication. So you know, in instances like that, Yuri, you know, I know there's no, you know, catch all but even just when we talk about the future of scams, because I think, you know, we've, we've talked about, you know, as an example, GPT4O and Google Astra, you know, their Project Astra, where AI models can kind of see and react in real time. But you have to think that this type of technology will also exist for the bad guys as well. So when it comes to the future of scabs. How should we be looking at that? What should we be paying attention to? And how can we actually detect them as they become more and more sophisticated?

Absolutely. I fundamentally believe you have to use a technology. If someone is using AI against you and you are, you know, not, not using any reduction for, for a better say, I think you are, you, you become more vulnerable. That's just the reality of things. So I think using AI back to, which is almost the beauty of AI, it's, it helps you to augment your vision. It just helps you to recognize the risk. It's almost like coming up with this risk score, right? This communication seems more risky or less risky. And here's why I think that is, that is a very good thing. I also think that unfortunately we don't have a really good answer to deep fakes right now. That's just the reality of things. Someone is forging videos. It's absolutely doable to impersonate someone, Someone's voice, someone's video appearance. That's massive, right? That's the impact of misuse or potential misinformation is. Disinformation is massive. As we believe what we see. I, I think we just all have to be more, much more critically, become much more critical thinkers and just know about that. Know that you cannot trust things that you see online or through digital communication, recorded communication out of the gate, as we used to many years ago.

Yeah, yeah. And you know, I like, I like what Monica is, is saying here. You know, kind of this concept, I think, you know, people come up with like a family password, right, in case they're ever, you know, being targeted. You know, but maybe, I mean, should businesses be doing something similarly like, you know, before something is signed off? Like, should businesses have like a pat, like a. Almost like a family password that says like, yes, this is real, you know, this isn't, you know, a scammer. Is that taking it too far? Or might we see something like that being commonplace in the future? Yuri?

I, I don't know. I think we might. I think, I think organizations are different by their nature and everyone is different. If you are a very small business, you're in one place. But then if you're growing rapidly, but you still don't have a CISO or someone who's responsible for information security in your company, I think you become higher on the radar for sophisticated targeted scams, but you don't necessarily buttoned up your security and that makes you even more vulnerable. So I think, you know, if you are an Average consumer just, you know, again following those basic rules will help you to eliminate most of the nonsense. If you are a larger company or a larger org, definitely use some tools or better invest in the security, invest in someone who can take care of this or those things get, gets more sophisticated and it's just difficult to manage this part time.

Yeah, yeah. You know, I'm curious what not keeps you up at night but what do you worry about? Right. So obviously you know, you and your company provide solutions but what is that thing that you still worry about specifically when it comes AI being used in business scams? What kind of keeps you up at night in that regard?

I think the video, like I said, I think the, the deep fakes is something that we just haven't cracked. This, this, you know, we, we didn't figure this out. We haven't figured this out yet. I, we, it's very difficult to recognize a scam. It's very difficult to understand how to deal with that. It's obviously someone is impersonating someone. They have the exact same appearance, the exact same voice. How do you differentiate sort of what's the ground source of truth? How do you not, you know, Internet has used to be the most free place in the world and now there is so much misinformation or garbage information there. It just not something that I'm worried about from a corporate standpoint or how do we absolutely will. A lot of extremely smart people are working to solve these problems. But I do tend to think that it got to the point where you can trust what you see. And I don't like that, of course. And I don't like where it's going.

Yeah. Yeah, great. Another great question here. So Monica asking, do you see insider cyber attacks at medium and large businesses? Something I didn't even think about. Right. So we're always thinking about, you know, these threats from the outside, but yeah. What about at large enterprise companies? Is that something that comes up and if so, you know, what steps should we be taking to maybe mitigate that or maybe help prevent that?

Well, I think that this, this is a separate topic from scams. Right. So in insider threats, I think they've been here forever. I don't have again, I don't have the hard data on that, but I think, but I think this is something that like the, the Trojan horses if you will, they have been happening for decades. I, I have seen mass targeted scam attacks. So imagine that Someone has identified 25 key employees of your company and someone has sent them almost simultaneously a Message from a, from the CEO and the CEO says to do something immediately. All it takes for one person to buy it out of 25 and you might be compromising someone else's company security. Security, right. So that is something that is happening quite often. But unfortunately I can comment on the insider, insider threat. I think it's a different beast.

Sure. No, that makes sense. So I mean Yuri, we talked about a lot in today's conversation. So you know, we went over, you know, as an examp, you know, not paying as much attention to, you know, single channel attacks and maybe verifying them on multi channels. Kind of the concept now that with AI people are hacking humans instead of hacking systems. And you know, how easy it can be to replicate someone's writing style and how people are using, you know, large language models to do that and to, you know, launch more and more kind of cyber attacks. But you know, as, as we wrap up, maybe one is the one important, the one most important piece of, of tactical advice that you have for business leaders when it comes to understanding AI's role in scam detection and prevention.

Just imagine that it is impossible to recognize if the communication is a scam. It is so good that it's, no matter how a critical thinker you are, it's just not feasible for you to know, to differentiate what's real from what's not real. So how can you protect yourself in such an environment? And I think again, going back to 2fa, not making hasty immediate decisions, calling back, using second communication channel is what actually will help you to verify that you are communicating with someone you can in fact trust. I think that trust is something that everyone online will have to work a little bit more to gain and we just have to stay cautious about that.

I love that. I love that. So, so, so much good advice for business leaders, especially with this new sweeping. I mean so, so many AI scams out there. Very sophisticated but some, some great advice here today on the show. So Yuri, thank you so much for joining the Everyday AI show. We really appreciate your time.

Thank you for having me.

All right, and hey, as a reminder everyone, there's a lot more. If this was helpful, please let us know. You know, repost this tag tags. You know someone in your organization who needs to hear this, but also go to your everydayai.com we're going to be sharing a lot more about what Yuri talked about in today's episode. A lot of links to some of the stories that we talked about and as, as aura so you can check out a little bit more about what they do. So thank you for joining us and we hope to see you back tomorrow and every day for more Everyday AI. Thanks, y'.

All.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going for a little more AI magic. Visit your everydayai.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.