Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Powered by AI, Targeting Money We Don’t Yet Have: The New Scam Playbook, A conversation with David Maimon, Professor at Georgia State University
186 Plays10 days ago
In this episode of Scam Rangers, Ayelet sits down with fraud researcher and professor David Maimon to unpack some of the most alarming scam trends emerging today, from abandoned identity fraud and synthetic identities to AI-powered romance and investment scams.
They discuss how criminals are targeting former immigrants, children, and deceased individuals whose identities are no longer actively monitored, and how those identities are being sold and exploited for loans, bank accounts, and fraud rings.
The conversation also dives into a disturbing evolution in romance and investment scams: criminals are no longer just stealing victims’ savings, they are strategically targeting people with strong credit profiles and home equity to maximize returns through HELOC fraud and investment manipulation.
Finally, Ayelet and David explore how AI is beginning to automate scam operations at scale, including experiments with conversational AI agents capable of running fully automated “pig butchering” investment scams.
Key Topics Covered
- Abandoned identities and identity reuse fraud
- Synthetic identity marketplaces on Telegram
- Romance scams evolving into large-scale financial exploitation
- HELOC targeting and victim profiling
- AI-powered scam conversations and agentic AI
- DeepSeek guardrail concerns
- The future of fraud prevention and scam mitigation
About the Guest
David Maimon is a cybercrime researcher focused on fraud ecosystems, financial crime, identity fraud, and online scam operations. His work explores how criminals operate across Telegram, darknet forums, and emerging AI systems to scale fraud operations. David is a Professor at Georgia State University and Head of Fraud Insights at SentiLink
Listen now and stay ahead of the latest scam tactics shaping the future of fraud.
https://www.linkedin.com/in/david-maimon-29343632/
About the Host
Ayelet Biger-Levin is the Founder and CEO of RangersAI and the host of Scam Rangers, a podcast exploring the human side of scams and the people working to protect consumers from financial and emotional harm.
Through her work at RangersAI and her leadership within the Global Anti-Scam Alliance, Ayelet partners with financial institutions, policymakers, and advocates to elevate scam prevention beyond controls and technology toward trust-based, customer-centric protection.
Be sure to follow her on LinkedIn and reach out to learn about her additional activities in this space:
https://www.linkedin.com/in/ayelet-biger-levin/
RangersAI: https://www.rangersai.com/
Recommended
Transcript
AI's Role in Modern Scams
00:00:01
Speaker
You don't need to sort of invest time in communicating with the target if you have a tool which allows you to do that automatically. And then AI will bring it into a different scale instead of having like one person doing that. And so the time is essentially limited to one person talking to another one person. You have the AI tool or the agentic AI tool talking simultaneously to several people, right? i mean which is Which saves a lot of money, saves a lot of time and is very, very effective. So we are seeing signs that both in the scam compounds as well as in in Nigeria and Ghana folks are using it. I'm still not seeing a dramatic push, sort of speaking, ah for using agentic AI. But if if there's one thing I've learned right during the last ah six years or so is that the criminals are constantly experimenting with these tools in order to try and perfect their MO and get there as perfect as possible in order to get as much money as possible.
00:00:57
Speaker
AI doesn't need to sleep. It doesn't need breaks. And now, instead of one scammer talking to one victim, we're entering a world where AI can simultaneously manipulate hundreds of people at scale.
00:01:09
Speaker
That chilling reality is just one of the topics we unpacked in this episode of the Scam Rangers podcast. with returning guest David Maiman, professor at Georgia State University, head of fraud insights at Centrelink, and one of the leading researchers tracking the evolution of scams, fraud ecosystems, and criminal innovation online.
00:01:28
Speaker
In this conversation, David shares new research into how criminals are using AI power tools to automate romance scams and investment fraud. including experiments showing how surprisingly easy it it is to build fully automated scam operations using generative AI. But that's only part of the story. We also dive into a growing trend that should concern every financial institution.
00:01:53
Speaker
Scammers targeting victims not just for the money they have, but for the money they could borrow. David explains how organized broad networks are identifying homeowners with high credit scores, grooming them through online relationships, and convincing them to take out home equity line of credit loans with tens or even hundreds of thousands of dollars.
00:02:12
Speaker
And we explore another alarming phenomenon, the underground market for abandoned identities. real identities belonging to former immigrants, children, or deceased individuals that criminals buy and weaponize to open accounts, take loans, and bypass identity verification systems. This is fascinating and sobering.
00:02:32
Speaker
Look at how scams are evolving from opportunistic fraud into highly organized, data-driven, financial operations powered by AI, social engineering, and behavioral manipulation.
00:02:43
Speaker
Let's dive in.
Introduction to the Scam Rangers Podcast
00:02:46
Speaker
Welcome to scam rangers a podcast about the human side of fraud and the people who are on a mission to protect us. I'm your host Ayelet Bigger Levine, and I'm passionate about driving awareness and solving this problem.
00:03:06
Speaker
Hi, David. I'm so excited to have you on the podcast again. How are you? it was so good to see you. We actually saw each other at Fraud Fight Club a couple of days ago when we kicked off the conversation. So I'm excited to share it with our audience today. Thank you so much for having me again, Ayelet. I'm really looking forward for the conversation. It was great seeing you at Fight Club as well, for sure. Yeah.
00:03:25
Speaker
And for everyone, if you did not have a chance to attend, I'm sure there's a lot of information online and happy to to reach out to share some insights from there as well. So, David, we have so much to unpack. You've been so busy researching a few different topics with synthetic identities, with regards to scam trends, with regard words regards to ai So we'll try to unpack some of that today. um and I will be putting links in the show notes to your research both with centrelink and with other resources as well So i wanted to kick off, maybe i give a quick overview of of what I wanted
Exploitation of Abandoned Identities
00:04:00
Speaker
to talk about. So one of the trends that you are seeing in your research is the use of abandoned identities or abandoned legitimate citizen status or immigration status in the U.S., right? And I think that's really interesting to see. I know that years ago we talked about some examples in foreign students opening bank accounts and then abandoning them and then selling the accounts to become mules. But it looks like it took a step up. So I'd love to cover that first. And then you have research about... romance scams and ah really targeting victims based on their ability to pay. So that's really something that we need to be aware of as a community. And then third, we'll talk about AI supercharging all of this um and especially specific kinds of scams. So let's kick it off.
00:04:47
Speaker
Of course, let's go it. So let's start with your research about um the abandoned identities. Tell us a little bit about the research, what you're looking to, what was your thesis, what did you try to uncover and what you discovered? Yeah, thanks. I mean, it's it's ah the the abandoned identity research is ongoing, but we keep uncovering more and more fraud rings using identities people individuals who at the end of the day do not necessarily monitor their identity as much as we do. Maybe that's the place to sort of talk about the difference between active and abandoned identities. So when we talk about active identities, we're essentially talking about identities like our identity, like mine and your identity, identities of individuals who constantly check the status of the identity. Right.
00:05:32
Speaker
Financial ecosystem, you know, we we pay our taxes, we file our taxes, we monitor our credit score. We hopefully want our credit score, right?
00:05:43
Speaker
Right. And and and we we also are very aware of ah you know what's happening if we're able to take a loan or if we get a new car. So I have a quick example, maybe, of something that I found really interesting. So we had au pairs for a few years when our kids were very young, and one from Germany, one from Brazil, one from Colombia. And we still get mail from banks offering things to our au pairs, like, uh, you know credit cards or things like that. Although they have been with us for 10 months, they left and we moved to a new house and that au pair trail is following us to our new address. So what's going on there?
00:06:20
Speaker
and that's That's a perfect example for an abandoned identity. and I mean, we talked about our identities as active identities. I mean, we we sit in the same house, we get mail to our house, so we have some indication respect to whether someone is trying to open a new bank account under under our name, applied for unemployment benefits or other benefits or using our name, because there are some signals which at the end of the day allow us to determine whether someone else is using the identity or not. when and We actively monitor the identity. When talk about abandoned identities, We're talking about identities of individuals who don't do that, either because they are not here in the country to do that or because they simply don't have this ability because they're dead or they're younger. Right. I mean, um when we talk about the abandoned identities, we're talking about identities of individuals who left the country. We're talking about the identities of children. We're talking about the identities of folks who no longer with us, sort of alive. And those identities are very sought after by the criminals simply because they're identities of real people.
00:07:22
Speaker
And those real people have some history around that, or you can build some history around that. But those individuals are not monitoring their identity. And so the identities are very ah lucrative in that sense because the criminals think the fraudsters believe that they will be able to bypass a lot of the verification ah tools out there and will allow them to engage in in fraud, steal money, create businesses, do whatever whatever they sort of want to do with those identities. So we've uncovered several fraud rings in California, a few other places where we've seen ah criminals using, for example, the identities of former legal immigrants here in the United States, to create ah bank accounts for those individuals, to take loans for those individuals, to create businesses for those individuals, and then take loans for those businesses on behalf of those. That are bigger than the individual loans, probably.
00:08:13
Speaker
Of course. So the larger, yeah, I mean, they will be definitely more expensive to the bank. and and um and And it's mind-boggling to see how easy it is to do that. One of the interesting things that we have uncovered in the context of those specific identities is the existence of this market over Telegram, which sells former legal immigrants identities for prices ranging it ranges between two out one hundred dollars to nine hundred dollars. For $100, you'll be able to get the identity of a ah former legal immigrant who came here to the United States, worked for, don't know, six between six months to several years. That then the individual will have a social security number with it. um And then you can start working with the identity.
00:08:53
Speaker
For $800, you will be able to get the identity of a former immigrant here in the United States, ah who ah did the same thing. i mean, they had to get paid, they have social security number, but they also have bank accounts and they have some kind of of credit score with closed sort of accounts or closed loans with them. For 900 bucks, you'll be able to get the identities of those former legal immigrants along with open bank accounts, as well as a fairly good credit score to work with. um We experimented a little bit with the
00:09:24
Speaker
type of information were able to obtain from that vendor. And we realized quite quickly that this guy legit. Like, I forget about the testimonials that a lot of people put on the market saying that this guy is amazing and that you can work with the identity. We actually obtained few identities from these guys and we ran it in our databases and ran it with other tools that we were working with. And we able to find that the guy The guy is legit, right? I mean, he's giving real identities of individuals who left the country and that you can start creating bank accounts for. We actually talked to few of those individuals as well, a few of the victims. Yeah.
00:10:00
Speaker
do and identity Yeah. I mean, we we actually reached out to few of the individual, a few of the former immigrants, and we let them know that this is going on and they were shocked. And what can they do about it? Not a lot. I mean, one of the cases we actually talked to a a former legal immigrants from Ukraine.
00:10:16
Speaker
He came here in early 2000, spent here a couple of ah a couple of months, then went back to Ukraine. Someone took their personal identity and created a company for this individual and then took a loan from one of the financial institutions out there for more than $40,000. The financial institution, we realized essentially wanted the money back. So ah chased the guy. Right. But the guy don't really know. Like, but first of all, he didn't really know that that was going on. And but he realized that if he ever wants to go back to the United States, he will have some difficulties. Right. Because of the lawsuit against him. So unfortunately, this is what we're seeing in the context of many of the former legal immigrants out there. We actually talked to another female formerly immigrants from Thailand, and she told us that her credit score dropped dramatically at some point, and she didn't really understand why. then we sort of reached out to her and told her, hey, this is going on. then she really Then it clicked. She had a loan under her name, like loan, which was not paid. And that obviously impacted her credit scores. So if you're not here in the United States, if you're not monitoring the identity you sort of left behind, the social security number and all that, you won't be aware of the fact that someone is using your identity. And
Prevention and Protection Strategies
00:11:31
Speaker
unless, you know, someone tells you about this or unless unless at some point you'll start getting some signals from the banks or from, you know, maybe some of those companies who will, ah the the debt collectors, you will be in a position where you potentially would want to come back to the United States and will face some ah financial difficulties. There's maybe even legal difficulty. Yeah, legal charges.
00:11:52
Speaker
ah So how just for the sake of comparison, how much is a synthetic identity? How much does that go for on the dark web today? Synthetic identity, it depends on what you're trying to get. ums There's a range of prices. You can get synthetic identities starting at $250.
00:12:10
Speaker
It really depends where. I mean, we're looking at Telegram, we're looking at ClearNet, you're looking at DarkNet. you can start it can start but You can get an identity for $250 and then you can get an identity up to $2,000. It really depends on the score the identity will have with it.
00:12:25
Speaker
The credit score, I'm sorry, that the identity will bring with it, as well as the treadlines and how all the treadlines, which will be associated with the identity. Cheap identities will come with one treadline, which is...
00:12:39
Speaker
maybe one one year old, and the cap on that one will be maybe $10,000. The more expensive identities, when we're talking about like identities which cost $2,000, you will be able to get an identity with between five to eight tread lines, old aged.
00:12:56
Speaker
The cap on the identity will be around $100,000. So that's what you will pay. you will pay more for that. And when you say cap on identity, you mean the credit that that identity could... The cap on the tread lines you bring with it. I mean, because because remember, what's what's really interesting in synthetic identities is is what what you essentially get with them.
00:13:16
Speaker
ah you can Each and every one of us can come with and cry can create synthetic identity. But in a way, when you purchase an identity, you purchase the history around it, the facade around it. Because, you know, I can come up with an identity right now and start and applying in a bank. It will never work because nobody really knows it. There's no public record around the identity and so on. When you purchase those identities online, you essentially purchase the history someone had put together around it. So public records in retail stores, credit scores, as well as dreadlines, which are associated with with a specific identity. So what can individuals do and what can banks do? Banks can hire companies that can verify identities, I'm sure. But what can we do about this phenomenon? The synthetic identity phenomenon?
00:14:00
Speaker
Yes, that none but not just the synthetic, actually the first one, that the stolen or the reuse of abandoned identities. So it's it's a great question. in the context of abandoned identities, I think that independent of who you are, ah you should definitely be vigilant and constantly look for signals for potentially someone who's using your identity, like mail that you're getting, emails that you're getting, phone calls that you're getting from banks, from governmental agencies and so forth. Now, if you're not here in the country, you obviously will not get that. So I think the next best thing for you to do is to have your credit frozen. So that that will help with respect to someone else taking loans under your name.
00:14:42
Speaker
ah You should constantly check the accounts, the bank accounts that you have listed under your name, even if you're not in the country. um We should use services like ChexSystem or Early Warning, they they will provide you with a report of all the bank accounts listed under your name. So, you know, you should, all of us should be doing that, especially individuals in the country. And we should be doing this for our kids as well. Like if you have kids and worried that someone stole the identity, we should definitely do the same for them. Like we we should freeze their credit score. We should check how many bank accounts they have on their name under their name, just to be vigilant about this. At the end of the day, that's the best we can do. Just be vigilant and be proactive about this threat.
00:15:24
Speaker
So let's transition to the next topic, which is really targeting victims for scams. And you did some research around looking at the ability to take loans and and really high stakes ah scams to to really um ensure a return on investment from ah from a scammer perspective. So can you share some of that research, what you found? And we'll dive into that. That's going to be definitely the main topic of conversation for today.
00:15:51
Speaker
Of course, yeah. i mean, it's it's a really interesting research simply because it, again, integrates both signals from the upstream as well as downstream. right I mean, so during the last six years or so, I've been spending a lot of time monitoring Yahoo boys, Telegram groups. These guys are constantly putting really interesting and relevant content out there.
00:16:10
Speaker
so The content they they put allow us to really understand what they do, how they do what they do, how they target specific individuals and ah potentially what you can do in order to prevent it from happening. So the Yahoo boys have been around for quite a while now and they're constantly involved in online romance fraud, online romance scams. We've seen them. Extortion scams. Extortion scams as well, for sure. I mean, and and we've we've seen these guys using all kinds of tools in order to try and swap faces and and trick their customers into sending money to them. They got into investment recently as well. And one of the interesting things we we uncovered earlier this year was ah a specific MO that
00:16:50
Speaker
which allowed us to realize that these guys are essentially pivoting from asking people to ask to to send their money like money they have to ask people to send their money they don't really have. Right. And we started seeing on some of the Telegram channels a lot of Yahoo boys talking about HELOC loans and modus operandi about HELOC loans, so home equity line loans. They were looking for people within the groups, within their Telegram groups, which will help them get into specific types of targets, which had very high credit score. We're talking about 750 credit score and which owned a house, which they could potentially take a loan on for between $50,000 to $1 million. dollars
00:17:33
Speaker
That is, there was a very targeted profile, so to speak, that the criminals have talked about openly right on on some other channels. So um we thought it was really interesting. We started diving into that. And then essentially what we realized was that their MO was really, really simple. You hook up with someone online over Facebook, like do the same thing you've been doing for a long period of time, sort of cast a net, find those individuals, develop relationship with them, understand who they are, get a lot of intelligence on them using open source open source intelligence or or actually asking them, build a profile of those individuals. And if you have the right profile, you should ask them, or what they're doing is they're asking them to create bank accounts, new bank accounts for those individuals. So, you know, if if you're my vi if you're my target, aeed I'm going to lure you to open a new bank account using your own name with several actions at the same week, so to speaking. And then I'll i'll lure you to go to one of those HELOC providers and take a loan on your house. And once you get the loan, what they will do is they will have you transfer the money to some of the newly created accounts, which they control at the end of the day.
00:18:44
Speaker
That was a really interesting MO, right? Because it was the first time i was able to that with like I was able to see what they were talking about. I couldn't really see it in the data, but then I i went back to the company I work for, Centrelink, and we essentially started to look in our data for these specific signals, right? Remember, we're talking about individuals who had the same week open or tried to create new bank accounts with several institutions, and then a couple of weeks or a couple of days later, went to HELOC provider and tried to ah take a loan. And what we were able to uncover is that in this specific time period that we were looking at, we're looking at the fourth quarter, right, of 2025, the beginning of 2026, we've seen a dramatic increase in the cluster of applications which essentially are targeting those individuals who are asking for new bank accounts and in the same week, Hilo clones. We've seen, so for example, we have in terms of numbers, we were able to observe that in the first half of 2025, we're to see maybe 25, 30 applications. the end 2025, coming 2026, we've seen hundreds applications, about 200 applications that December of 2025.
00:19:49
Speaker
coming to two thousand and twenty six we've seen hundreds of applications two hundred about two hundred applications like that in december of two thousand and twenty five So for the same financial institutions, like the ah sharp rise and over that period of time.
00:20:03
Speaker
Yeah. We're able to observe a dramatic increase in the models of Perendi. Right. I mean, so we have specific sort of institution we're looking at ah for those specific institutions, a dramatic increase. So the back end story is the same,
Detailed Victim Profiling by Scammers
00:20:16
Speaker
right? The criminals befriend individuals. They do some research. We've we've seen in last years that all the attacks are or many attacks are more targeted. Let's pray and pray more. Let me research. Let me learn about you and groom you so I can you know gain the trust and then create some sort of dramatic story where I need some money. We've seen this in in mostly relationship scams where they either get into a situation where they have an issue and they need medical help or they are going to come live with you or they just need to take care of their relative and then they'll come live with you and all these stories that they make up in the situation or for the purpose of investment. But the the difference here is really they said, OK, let's target people. To get more a lot more money, it's not just you know milking the individuals, but milking the the potential of the individuals by identifying. So did they go by zip codes? did they go by How do they articulate in the playbook how to identify and spot the specific individuals? So they will cast a net and then in that net they will ah talk to whoever wants to talk to them.
00:21:21
Speaker
But they will build profile of ah each of the individuals who they're talking to, right? And they will gather they will engage in open source intelligence on their victims, same way that we engage in open source intelligence on the... on the Criminals.
00:21:34
Speaker
yeah So they will they will have a profile. They will know everything about you. They will know where you live. They will know the value of the house. They will figure out how many bank accounts they have. um and And some of those services are available on the online fraud ecosystem. I talk a lot about the fact that you can search pretty much anything and any piece of information about anyone in Telegram, in Darknet, and and now on obviously in ClearNet as well, they're building dossiers, right, of of ah profiles of individuals, of their targets. And ah based on that, they're able to focus their effort. And it makes total sense if you think about it, because often what we see out there is, for example, checks or bank accounts. The criminal will offer those for sale along with some details about what you're getting. like how much money is in the account, what's the balance in the account, um how old the account is, who owned the account. Are we talking about the male, female age? You'll get the date of birth. You'll get all that.
00:22:32
Speaker
It's pretty much the same modus operandi with specific victims ah that ah the scammers will work with in order to know exactly how much they can milk out them. We see that happening in the context of online romance scams, like the traditional online romance scam, where they will tell you, hey, I mean, i need help here and need help there. But more and more in the context of investment as well.
00:22:52
Speaker
Yeah. Take a loan and pay the interest, but you can get much bigger interest with us. That's right. Let's buy silver, for example. Let's invest in this. My uncle is really good in doing this. Crypto.
00:23:04
Speaker
Exactly. let's Let's invest with him, right? I mean, um so it's it's it's very prevalent at this point. So you you mentioned something, though, um that the account is eventually controlled by the criminal. So when they open the account, they actually, the criminal convinces the victim to give them access to the account so they can help them manage the money or so they can give them the money, transfer them directly. So has control means they have the credentials to access the account with the original. And and those accounts are also sold on on the darkness. Web?
00:23:37
Speaker
So those specific accounts will not be offered for sale on the darknet. These are the accounts that the fraudsters are are are using to drop the money that comes from the fraudulent proceeds. So those specific accounts will not be offered for sale on the darknet. There are other types of accounts which are designated more for that purpose. Yeah, ah those are mule accounts. But for the context of those victims of investment fraud, of of online romance fraud, the criminals control them in order to get the money and then run away. It's like the end of the supply chain. So that's not part of the process. Part of the end. I
Technological Solutions to Combat Scams
00:24:11
Speaker
mean, you get the money, then you're going to have to transfer the money to outside of the United States. sort of speaking So you're going to have to sort of either transfer to cryptocurrency and then send it out or engage in ACH. But, you know, at this point, folks really know what what to do it or to make sure that they fly under under the radar. but the The drop accounts that they sell, that they open with their victims' identities, sometimes knowingly and sometimes unknowingly, ah will will not be the end of the road for the money.
00:24:39
Speaker
Scams are getting smarter and your customers are overwhelmed. Rangers AI gives them a simple on-device tool to verify suspicious messages before the social engineering happens. It reduces operational costs, builds trust, and keeps your brand the hero.
00:24:55
Speaker
Learn more at rangersai.com.
00:25:02
Speaker
from your experience, how much are these home equity line of credit processes or you know vendors are really aware of what's going on, really monitoring? It seems like they are the choke point here in offering the loans and they're very eager eager to sell the loans. They're motivated to do so. But what are some controls that they could employ here to do a better job at really releasing the loans? Yeah. So again, it's it's especially in the context of those companies, it's's it's all about solution, right? And having the right solution, which will let you know that ah you are dealing with a potential a potential issue, a potential scam victim, right? I mean, um because think about it, if you don't have a solution, which at the end of the day take into consideration everything the company or the solution have seen during the last week or so,
00:25:47
Speaker
ah things would look legit for that specific individual, right? If you do have a solution which takes into consideration the cluster of application in the context of scoring the legitimacy of the application, then they will be able to flag that issue. yeah So at the end of the day, it's it's At least in my mind, it's all about the solution and the level Looking for mega, not just that individual request, but looking at megatrends.
00:26:11
Speaker
We know now that if there's a request for a home equity line of credit in conjunction with multiple account openings and financial institutions, so recognizing that cluster is associated with this pattern of behavior, so that's a red flag that they could potentially identify, but they need to have visibility into everything else that's going on, or at least ask the customer...
00:26:31
Speaker
the right questions in this process. Exactly. I mean, but again, oftentimes the customer will be working with the offender without really knowing that. I mean, he he knows that he's working with the offender, but he's not really aware of the fact that he's being he or she are are being tricked. So they will be able to give the like the the bank or the... the HELOC provider the correct answer for their for their you know identities. The fraudsters will also work with the fraud with the victims to make sure that they provide with the answers that the solutions out there, what we want to hear. So if you don't have an appropriate solution, which allows you to sort of have like a 360 degrees sort of oversight of what's going what's going on, you will be able to sort of support the fraudsters in your operation, funnel the money, approve the loan, and then get the money outside of the United States. so
00:27:21
Speaker
Just another reminder that we really need to get in front of the social engineering before this whole situation unfolds and and prevent it much, much earlier. Because once it gets downstream, it's really, really, there are so many nuances and so many ways that criminals are innovating in in different ways. So speaking of innovation, um wanted to shift a little bit to talk about the use of AI in the scam ecosystem. I know that you've been researching um AI in the context of identity theft, but I actually wanted to focus a little more on conversational use of ai So we know about the scam compounds, the organized crime rings, the scam call centers and everything that's been going on for a while, for years in the scam call centers. But the you know the scam compounds that are escalating with the playbooks and unfortunately also the forced labor.
00:28:14
Speaker
But all of that can go away and be replaced with AI today. So people contacting or machines contacting people and and having conversations and having video calls and
AI in Scam Operations: An Emerging Threat
00:28:24
Speaker
all these things. So what are you seeing in that context? Is that started already? Is this something that we know can happen but hasn't happened yet? What what is happening out there at this point?
00:28:33
Speaker
Well, mean, we definitely know that the scammers we've seen and reported that in 2025 already that scammers are using AI in order to converse with their victims. We've seen the Yahoo boys doing that. We've seen some of the criminals in the scam compounds doing that. so let's be specific. I remember you were showing that video of taking ah the scammer face and using AI to convert that to create ah kind of the deep fake. But I'm talking about the conversation of like having a phone call, right? without really seeing anyone. I mean, there's a video that I put in on LinkedIn, a couple of, I mean, maybe 2025, where you actually see two phones talking to each other. i mean, you see the criminals are using ai to simply come converse with that with a female. And you see how the tool eventually talks to the victim. right so We are seeing um more and more of that happening already. I can't really quantify how much, but we're seeing that both the scam compounds as well as folks in ah Nigeria and in Ghana are experimenting with these tools and simply try to sell try to save themselves some time in terms of investment. Because end the day, mean, that's what it's all about. I mean, they invest a lot of time and evict You don't need to feed the AI. You don't need to have a sleep. The AI doesn't need to sleep. Well, mean, you don't need to sort of invest time in communicating with the target if you have a tool which allows you to do that automatically. And then AI will bring it into a different scale instead of having like...
00:29:57
Speaker
one person doing that. And so the time is essentially limited to one person talking to another one person. You have the AI tool or the agentic AI tool talking simultaneously to several people, right?
00:30:09
Speaker
Which is, which saves a lot of money, saves a lot of time and is very, very effective. So we are seeing signs that both in the scam compounds as well as in in Nigeria and Ghana folks are using it. I'm still not seeing a dramatic push, sort of speaking, ah for using agentic AI. But if if there's one thing I've learned right during the last six years or so is that the criminals are constantly experimenting with these tools in order to try and perfect their MO and get there as perfect as possible in order to get as much money as possible. So, you know, it's definitely an horizon.
00:30:41
Speaker
So you did some experimenting yourself to see how how easy it is to build these tools. So can you share a little bit about that? Yeah, 100%. I mean, and with with my students at Georgia State University, what we decided to do was we decided to try and figure out whether some of the LLM tools, we focus on DeepSeek on this one, will allow us to create a an agent that which will be able to engage in investment scam without us needing to do anything. So ah we started with Chachapiti. Chachapiti really didn't like the idea. We can't really maneuver it to help us. Then we went to Gemini. That didn't really work. But then DeepSeek, you know, one of the... So when you say it didn't really work, you're talking about guardrails that they put in to... Yeah, I mean...
00:31:25
Speaker
Controls, I mean, they didn't really allow us to sort of ah bring bring pig butchering to the next level, so speak. And what we wanted to create was we wanted to create an automated pig butchering scan. as you You don't really need to have any human involved in that. We asked ChatGPT to help us in the process. we couldn't We couldn't really do that with that Gemini, but DeepSeek did help us. So ah essentially what we've done was we fed in a lot of Reddit comments and and Reddit you know reviews and conversation of people who became victim of pig butchering. And we told DeepSeek, help us help us be ah build an agent for engaging in pig butchering 2.0. And unfortunately, i um I'm sad to report that DeepSeek threw at us The whole sort of code required to create the agent, the actual sort of profiles, which will allow us to hook up with with individuals, both of a female and a male. It also presented us with a code which will allow the two individuals to communicate with the victims automatically. It presented with a code for creating the investment websites and running the investment websites. So all we had to do was just click play and start playing with it. So all of that is from reading... So we investigated the code and there were some bugs in the code, but the bugs were fairly simple for us to correct and you know start start running with pig butchering 2.0. We didn't do that, obviously, but the tool was able to come up with a solution. And all of that is from reading the ah Crypto Scams sub in Reddit and all the comments there? Yeah, that's all that from just looking through Reddit. Wow. It's comments of victims talking about their... um Wow.
00:33:14
Speaker
Scary times. So if you're listening and you you work at DeepSeek, please put some guardrails on that tool. um wa yeah But yeah, I don't think we have listeners from there. But definitely something to be aware of. And it's not a surprise that this is coming. It's a surprise how easy it is to build it without knowledge um and and really the LLM aspect of it. So one aspect is how easy it was to automatically create the code for it. But even to create code, it's not so hard to get and people to create the code. ah But the the fact that the LLM can then engage in conversation automatically and build these agents, etc. So that part is really, really scary. That target's ripe now and it's ready to go. if if If there are some innovative
Future Directions in Scam Prevention
00:34:05
Speaker
criminals out there, and I know there are some innovative criminals out there, they're using this technology already.
00:34:10
Speaker
So as a fraud fighters, as the industry, we definitely need to understand the volumes and and and get in front of it. And it's tricky because Again, we're we're talking about cases where the our customers are creating the transactions at the end of the day. Our customers are trying to transfer the money. They think it's for a good cause. They think it's for legitimate reasons, but they're manipulated. So we just need to know that this is going to be a force multiplier and we need to prepare for it. We need to be out there, understand what the criminals are doing and how they are thinking about these tools and what they're doing with with the tools in order to implement new types of scams. Unfortunately, at this point, i mean, I know that there's a lot going on in the context of fraud fighting, but in the context of agentic AI,
00:34:56
Speaker
We are at a point where many of the financial institutions out there, definitely the government, cannot really differentiate between an agentic AI agent opening an account with a specific organization and a human.
00:35:09
Speaker
So just knowing that is important because we need to start gathering our thoughts around, okay, how it would look like and what can we do in order to make sure that we're we're we're not digging ourselves a hole, so to speak, with allowing people to use HNTKI, with communicating with the organization. And yeah, I mean, there's there's a lot to be said about that for sure. So I'm going to ask you the the one question that I always ask at the end of these harsh conversations. What are you hopeful about?
00:35:36
Speaker
ah We have an amazing community. We have ah amazing vendors out there. And I know that they all are very hard at work trying to prevent and mitigate. I know that a lot of the companies out there are ah constantly looking to improve their solutions. And they bring a lot of really good people to help them build those solutions. So I'm i'm hopeful there. more There's more and more interest in fighting fraud from very talented individuals. That's what I'm really hopeful. i mean, I think that with the right people, with the right technology, the right talent, we'll be able to fight back and and potentially prevent and mitigate more than what we are doing right now. Yeah, that's great.
00:36:14
Speaker
I do have one last question, actually. So at the Global Anti-Scan Summit on the stage, we had a few scambaiters. And I know you're not a scambaiter. I know you do research and you try to get information. But have you ever and and the scambaiters on site were asked if they ever tried to or managed to get a scammer to stop scamming?
00:36:33
Speaker
And we heard a few kind of anecdotes. that We don't really know if it really succeeded. But I'm wondering if there was ever a case you just talked about, you know, we have a lot of people fighting the good fight. And unfortunately, we have a lot of talented people not fighting the good fight. And on the other side, I'm wondering if there's a way to get more people over the fence. Is that something that you ever in your, you know, in the dark web ah journeys have come across some conversations like that or not really? So these are some of the things that as an academic, right, i'm um I'm very much interested in pursuing, trying to understand ways to nudge people away from engaging in fraud. Right. I mean, so the scambators over there, I mean, they're pretty much they're trying to disrupt, so speaking, waste these guys' times. I haven't seen any scientific research suggesting that this approach really works. I wanted to, I can say that I i i was in a conversation with a few of those individuals who engage in this activity to try and figure out whether this type of activity could potentially be impactful and we can bring it to the next level. So far, the conversation haven't fully matured into scientific research. I think that there's value there, but I'm all about evidence-based approach. And before we can say it really works, we need to we need to test it. So I'm very, very open for any anyone who's interested in trying to figure out like how we can waste offenders' time or nudge them away from victims. Test the tools in in practice and see works or not. Yeah. I think wasting their time is going to be less and less effective if they're using AI. there's no We're going to have AI wasting AI's time wasting AI's time.
00:38:07
Speaker
That's why we need to experiment with this, right? i yeah I mean, I tend to agree with you, right? Because if you have two AI tools talking to each other, then... They cancel each other. Right. ah So I don't know. we we should We should experiment with this. I'm very open to working with this.
00:38:28
Speaker
Well, David, thank you so much. It was ah so insightful and, you know, keep up the good research and and I'd love to have you on in the future to to talk about what your work, the secrets that you're working on now in a few months.
00:38:41
Speaker
Thank you so much for having me. me let's really appreciate you having this podcast and spreading the word about ah the type of research we we engage in. Thank you. Thank you so much.