FRAUDFORWARD
#103AI, Payments & the Future of Fraud Operations — Live from Safeguard
53 min
What’s up, fraud fighters, and welcome to Fraud Forward.
I recorded this live from Safeguard’s AI Deep Dive Retreat at The Broadmoor in Colorado Springs, and y’all, let me tell you something: this was not just another room full of AI buzzwords. This was fraud leaders, payment experts, financial institution professionals, fintech voices, marketplace operators, and risk leaders all sitting in the same place asking one very real question.
What happens to fraud prevention when artificial intelligence changes everything?
The future of AI fraud is not some far-off thing we are all fixin’ to deal with later. It is already here. We are seeing AI change the speed, scale, and complexity of fraud operations right now. We are seeing it show up in synthetic identity fraud, AI scams, AI identity verification, agentic AI fraud, fraud detection automation, governance conversations, and the pressure fraud teams are already carrying every single day.
But this episode is not about panic.
It is about realism.
It is about what fraud fighters are seeing, what institutions are building, where the gaps still are, and how we keep humans at the center while fraud keeps accelerating around us.
What you’ll hear in this episode:
- A practical conversation about the future of AI fraud and what it means for fraud operations
- How AI fraud prevention is changing across banking, credit unions, fintech, payments, marketplaces, and risk teams
- Why synthetic identity fraud, AI scams, and identity fraud prevention are becoming major areas of concern
- How AI in fraud prevention can help investigators analyze data, identify patterns, and reduce noise
- Why AI governance in fraud prevention cannot be treated as something we clean up later
- How agentic AI fraud, know your agent, and KYA fraud prevention are becoming part of the next conversation
- Why human-in-the-loop fraud detection still matters, even when fraud analyst AI tools are getting better
- How fraud fighters and AI can work together without replacing the people who know this work best
- A reminder that responsible AI risk management has to include governance, empathy, collaboration, and practical controls
Who should listen:
- Financial institution leaders and fraud professionals
- Risk, compliance, and cybersecurity teams
- Fraud operations teams and investigators
- Credit union and community bank leaders
- Banking fraud prevention teams
- Credit union fraud prevention teams
- Fintech, payments, and marketplace risk teams
- BSA, AML, KYC, and identity teams
- Regulators and policy advisors
- Industry advocates and victim support professionals
- Media professionals covering scams, fraud, AI, and financial crime
- Anyone trying to understand how AI-driven fraud affects real people, real institutions, and real fraud teams
This conversation is for the fraud fighters who are not just trying to check a compliance box. It is for the teams trying to protect members, customers, and communities while also figuring out how to use AI-powered fraud prevention without creating new risk.
Episode notes:
AI is already changing the future of fraud operations
AI is not just changing one corner of fraud prevention. It is changing fraud operations, AI identity verification, scam tactics, governance expectations, investigator workflows, and the way financial institutions think about risk.
Throughout these conversations, we talk about how criminals are using AI to move faster and scale attacks. We also talk about how financial institutions are trying to use AI in fraud prevention in a way that is responsible, controlled, and actually useful.
Because holy moly, this cannot be about using AI just to say we are using AI.
It has to solve a real problem.
Fraud teams are finding practical AI use cases
This is where I felt encouraged.
Yes, the risks are real. AI scams are real. Synthetic identity fraud is real. Agentic AI fraud is coming into the conversation fast.
But fraud teams are not sitting still.
They are looking at AI-powered fraud prevention, behavioral tools, fraud detection automation, data analytics, investigator workflows, and stronger collaboration as ways to better identify and respond to fraud.
We talk about how AI can help fraud fighters analyze large datasets, spot patterns, surface connections across systems, support risk assessments, improve alert review, and give investigators better context faster.
Fraud teams do not need more noise.
They need better signals.
Fraud actors are moving fast, and institutions have to move responsibly
Now, here is the challenge.
Criminals are not waiting on governance committees. They are not worried about model risk, privacy expectations, regulatory review, or whether risk was brought in early enough.
Financial institutions do have to care about those things. And they should.
This episode comes back to a really important point: AI implementation and AI governance in fraud prevention have to happen together. We cannot build first and ask risk later. That is not responsible innovation, especially when the people on the other side of these scams are real members, real customers, and real families.
AI risk management cannot be an afterthought.
It has to be part of the design.
AI-driven scams still affect real people
This part matters deeply to me.
When we talk about AI scams, investment scams, romance scams, synthetic identity fraud, manipulated media, and identity fraud prevention, we cannot talk about them like they are just trend lines on a dashboard.
These are crimes that affect real people.
People are being emotionally manipulated. They are being pressured. They are being deceived by organized criminals using tools that make their work faster, more convincing, and harder to spot.
And I want to be very clear: victims should not be shamed. They should not be blamed. They should be supported.
Empathy is not optional.
Responsible AI still needs human judgment
One of the biggest themes in this episode is that AI is not replacing fraud fighters.
Let me just assure you of that.
AI fraud detection can help identify patterns. Fraud analyst AI tools can gather data, prioritize work, summarize case context, and help investigators see connections they may not have had time to find manually.
But experienced fraud professionals still matter. Frontline staff still matter. Investigator judgment still matters.
Human-in-the-loop fraud detection is not just a compliance phrase. It is how we keep context, empathy, and common sense in the process.
There will always be gray areas in fraud prevention, and those gray areas require institutional knowledge, lived experience, and that gut feeling so many fraud fighters know all too well.
Stronger fraud prevention depends on collaboration
Fraud does not live in silos, and neither should our solutions.
This episode points toward a more practical path forward: better intervention systems, responsible AI governance, stronger data practices, empathetic response protocols, better collaboration, and clear use cases for AI-powered fraud prevention.
We talk about investigator workflows, alert prioritization, identity fraud prevention, AI identity verification, agentic AI fraud, know your agent, and KYA fraud prevention.
But underneath all of that is a bigger truth.
A rising tide lifts all boats.
If we want to keep up with the future of AI fraud, we have to keep sharing, keep learning, and keep building together.
Key takeaways:
- Awareness matters, but awareness alone is not enough.
- The future of AI fraud requires stronger technology, better governance, and deeper human judgment.
- AI fraud prevention can help teams scale, but it has to be implemented responsibly.
- Synthetic identity fraud, AI scams, identity fraud prevention, and agentic AI fraud are major areas to watch.
- AI governance in fraud prevention must happen alongside implementation, not after the fact.
- Human-in-the-loop fraud detection remains critical for context, empathy, and sound decision-making.
- Fraud analyst AI tools should reduce noise and help investigators focus on higher-quality work.
- Credit union fraud prevention and banking fraud prevention teams need practical, realistic AI use cases.
- Know your agent and KYA fraud prevention will become increasingly important as AI agents act on behalf of consumers.
- Collaboration across financial institutions, fintechs, marketplaces, solution providers, and fraud professionals is one of the strongest tools we have.
This episode is ultimately about more than AI.
It is about how we protect people in a moment where fraud is changing fast. It is about how we build responsibly, respond with empathy, and make sure the people doing this work know they are not alone.
Fraud is not slowing down, and neither are we.
Stay vigilant, stay informed, and keep moving fraud forward.
Episode transcript
Hailey Windham00:05
What is up, fraud fighters? Welcome back to Fraud Forward. Today's episode is another different one, because this is one that was recorded live from Safeguard's AI Deep Dive Retreat at the Broadmoor in Colorado Springs. And this event stood out. It really did. Not because it was packed with buzzwords or overproduced AI hype, but because it brought together the most diverse group of fraud leaders, payment experts, financial institutions, fintechs, marketplaces, and operators that I've ever seen in at a single conference, but who are all trying to answer the same question. And that is, what happens to fraud prevention when artificial intelligence changes everything? Throughout this episode, you're going to hear short conversations that I recorded throughout the conference with some incredibly smart leaders, again, across banking, fraud operations, identity payments, and risk. And what I loved most about these weren't those theoretical conversations. These were practical discussions about, you know, synthetic identities, scams, AI agents, governance, operational pressure, investigator workflows, and the very real challenge of keeping humans at the center while fraud accelerates around us. One theme that came up over and over again is that AI is absolutely changing fraud, but that doesn't mean it's replacing people. No, but instead what you'll hear is that it's actually helping fraud fighters scale, adapt, and make better decisions faster. So today I want to take you inside those conversations that were happening in the hallways, the breakout rooms, and networking sessions at Safeguard. And the first up is Angela Diaz. You know, I sat with her to talk about the industrialized scale of fraud and why governance can't be treated as an afterthought anymore. So here's what she had to say. What is the biggest AI-driven fraud risk that financial institutions should be paying attention to right now?
Angela Diaz02:20
The scale. The scale, and the ability for if the fraudsters get it right, because we know it's about them getting it right, figuring out that path, if they get it right at the scale that AI can provide for them, it will be just a massive hit of fraud, and our ability to react to that quickly and catch up is going to be challenging. So for me, and that's kind of a broad answer, but that's my concern, more at scale.
Hailey Windham02:47
Yeah. No. Well, you know, Erin West has said it best when it comes to the industrialized scale that these scams are targeting, you know, our customers, our members. And so being able to truly find a way to combat that effectively, that's going to be our challenge. Yeah.
Angela Diaz03:07
They are running like large corporations. Yes, they are running like large corporations. Now, I've been talking about it a lot on some panels I've been on. It's intense, and it's going to keep getting worse.
Hailey Windham03:18
So yeah. Okay. So would you say, or in your opinion, is AI giving the advantage to the fraudsters or the fraud fighters?
Angela Diaz03:25
Both. Yeah, both. And I think that the biggest opportunity is for us to accept that and be aware of exactly how they're using it and what they're doing, but be very, very focused on how we can use it to our advantage to get better at some of the things that we maybe have had challenges with, and get ourselves in real time up to scale, and something that we can continue to adapt, tweak, enhance very, very quickly. We need to focus on everywhere that we can use it in a controlled environment, of course. Add that supervisory agent layer, please, for your risk manager. But absolutely, I think that it is both, and I think it's okay to acknowledge that. But I think our focus needs to really be on, okay, we need to know enough about what they're doing, but let's let's buckle down, let's hustle, let's get to work on building what we need to build.
Hailey Windham04:19
Yeah. So I didn't pre-plan this question with you, but it's just one that's come to the top of mind. What advice would you give to a fraud professional that is looking to implement some type of AI that needs to bridge that conversation with their risk leader? What would you say should their first step be if they want to talk to their risk partner?
Angela Diaz04:39
The first step is always, and this is kind of a callback to our episode together, is like, you should already have a high level process of flow, and you should have some controls in place that are applicable to your current environment. Be able to say to your risk manager, this is what I'm thinking about, how I would like to use AI within my processes. This is my rationale behind why I think that it would be a good fit. This is how I think it might change my risk, and I would like to control for it and get their support. Now, if you are in a place where you want to introduce it into your environment, but you have no idea how it would change your risk or how to control for it, that's okay too. That's what we're there for. So simply approach them with I want to use it. I think there's a benefit. This is how I'd like to use it. Can you please help me with now how to understand what risks I need to be paying attention to, and either what new controls I need to put in place, or how are my regular controls applicable to that? What can I enhance there as well? That should always happen in parallel to each other. A theme that I'm picking up from a lot of the conferences is implementing AI, and then next year we should expect to talk about governance controls and risk. No, that's something I've been trying to change in the industry for years. We always should be working in parallel with each other. If you are changing a process, if you are introducing a process, if you are implementing a new solution, if you're changing your technology or your data, you need to be assessing the risk at that point, and in parallel, changing what you need to change as part of your change management process. You do not want to clean it up later, especially in an area like fraud.
Hailey Windham06:18
Yeah, no, I appreciate that perspective. And just for anyone listening, that's episode 97, I was checking on my phone to make sure, so check out that episode where Angela and I really dive deeper into that particular topic. So you also teed up the next question, I think, perfectly. But how can credit unions and banks realistically use AI without over complicating things?
Angela Diaz06:41
Start with baby steps. Start with it, maybe not in your live production environment. Something I always recommend, and probably a lot of people that are more further along in AI might kind of shrug their shoulders at it, but again, baby steps, baby steps is data. I use Microsoft Copilot all the time to help me with analyzing raw data, where I might not have the data fields I want, but I want to know trends and checks within the data, and it's a lot of line items, and I'm not a data scientist, data analyst, not a data scientist. Another good option, especially for assessments, is if you're trying to do a risk assessment, you can give Copilot a couple of simple prompts that help you walk through a risk assessment by feeding into it some procedures and standards you're trying to be compliant with, some process maps, and it'll do a risk assessment for you if you give it the right prompts. It's very easy to treat yourself, and I promise I don't work for Microsoft Copilot. That's just what all the banks are, because we are, you know, that's one of our easiest wins for us to use, right?
Hailey Windham07:50
Love that. Okay. Last, last question, then I'll let you get back to it, because I know that we've got some pretty cool breakthrough round tables and individual conversations going on right now. But what's one misconception people have about AI and fraud prevention?
Angela Diaz08:04
That it will harm us, that we will not be able to control for it. That is not true, and we can't shy away from it because of that fear. Fraud as an industry is one that is constantly changing. We are constantly evolving. We have dealt with technology changes, the way that scams happen. The internet itself changed what fraud looks like, right? It's not just like transaction monitoring. Digital banking changed the way fraud looks like. So this is just another instance where the industry is facing a change, a forced evolution. We should be good at that. That is what we need to do. It keeps us on our toes. If you're not in a position to be able to be open to, okay, what is next? What do I need to do? How do I elevate myself? How do I tackle this change? If you're not willing to do that work, if you're not willing to keep up, then you're going to struggle.
Hailey Windham08:56
So Angela brought up something incredibly important there, and that's this idea that AI implementation and governance has to happen together. But another major theme throughout the conference was collaboration, because no institution is solving this alone. So next I spoke with Melinda Szabo from the operational reality many financial institutions are facing right now, especially as scams become more sophisticated and emotionally manipulative. When you think about Safeguard, or why you wanted to come to Safeguard, what is it that actually drove you to come?
Melinda Szabo09:34
I believe my biggest goal here is to get ahead of all these scammers and fraudsters, which obviously I think it's a never-ending fight. And I learned this yesterday from one of the sessions that how AI is so sophisticated these days, and how what they can do with it. But at the same time, I think gathering industry insight, building connections, and making sure that collaboration is happening. Yes, and it's a must right now, not just to have to, I guess we have to. I think this whole conference brings a lot of people together, and I love it.
Hailey Windham10:07
I agree. There's been a really great mix, I think, of like e-commerce, banks, credit unions, solution providers, and it hasn't felt real salesy, at least from, well, they're not trying to sell me anything. But have you felt the same way? Has it been a little too pushy, or has it really just felt kind of more collaborative?
Melinda Szabo10:30
I didn't find it pushy at all. I actually enjoyed this 15-minute meeting. It seems a little daunting, but at the same time, I think it's very, very useful, and 15 minutes is more than enough to get that insight from the other person. I did not get any feeling of being, also together like more collaborative. I had very, very good sessions, very good meetings, like we did yesterday. I think we had a really, really good meeting. I agree. So no, I'm looking forward to the second day.
Hailey Windham11:02
I am too. I am too. I believe you said you had eight meetings here.
Melinda Szabo11:06
I believe so. Eight or six, something like that.
Hailey Windham11:10
And I know that General Law does too, so it does feel kind of daunting. I think that the only thing I would probably want is maybe just a little extra buffer in between our meetings, yeah, because I'm trying to write down those last minute notes, you know, and then move over, and then I feel like I'm starting the next one going, "Oh, sorry, let me, let me finish this one little note, and then we can start our meeting." Okay. So I love that what you wanted obviously from Safeguard was to try to get a better understanding of where you are, where you can go from here. Is there any one particular insight, not like a solution, but is there anything that you've taken away so far, like from day one, or anything you're hoping to grab from day two?
Melinda Szabo11:56
Absolutely. What I realized, and this thing has been in my mind for a few months, is that I believe the collaboration between financial institutions should be more close. That's the only way to stop the. I had a conversation yesterday with somebody that we talked about, that we don't reach, I think, the target audience of these scammers as much as we want to, and maybe media would be a very, very helpful tool. And also, again, as I mentioned, collaboration between the financial institutions, because how do you stop the money if I have to send in the letter by me, that's not going to be helpful, you know, when you have to stop a wire immediately. So I don't know if there is any way to get the banks more involved, build something that we can have a more, better collaboration here, because at the end of the day, I mean, the scammer is winning. And we learned from the session yesterday that, you know, it's an entire generation's money is gone from the US.
Hailey Windham13:10
Yes. When she said that, it's been a little over a year since I've heard Erin speak in person, and you know, every time she makes that statement that a generation's worth of wealth has left this country, it really puts it into perspective that if we don't do something now, like how much worse can it actually get, you know? And I look at this room full of, you know, 600 people, and I go, we are really trying to make that difference. But it's so hard whenever we've got the rules and regulations that we have to follow, where the scammers don't.
Melinda Szabo13:41
Right. I agree.
Hailey Windham13:46
So what's the biggest AI-driven fraud risk that financial institutions should be paying attention to right now?
Melinda Szabo13:51
I still do believe that that's the scams, the investment and romance scams. It's very, very scary. What they can do with AI, and how much you cannot even differentiate these days between if it's an AI or a non-AI video that you see, and people believe it. So I believe that will be the major focus. And again, that's my question to you, Hailey. How would you get media to help us?
Hailey Windham14:16
Yeah, I know that a part of it is having to, at least in my mind, you know, you have to have people that you know. I remember I have a local news anchor who I went to middle school with. I reached out to her and I said, hey, can we talk about this, and I had a five minute segment on her news show, and I'll forever be grateful for that. But how else do we do it, you know? And I know that like scams and fraud, there's like two ways that I think that media portrays it. One is, you know, the stupid victim. Oh my god, they fell for this. The other is that it's like the sex sells, right? They try to tell the sexy part of the story. And I'll never forget when CBS ran their special, I believe maybe even sometime last year, later last year, I was so mad. I was like, what kind of journalistic integrity, like I get not giving away your source, but you're going to walk into one of these scam places and, you know, let them have their 15 minutes of fame instead of instead of helping the victim like that. Oh God, I still get mad about it. So I think there's a way to do it. I think that media right now is not getting it right, but I absolutely agree with you. We have to get this information in front of people.
Melinda Szabo15:31
Yes. And I think that would reach the target audience, because you know it can be people who are sitting at their home, 60, 70 years old, watching local news, local TV, maybe not even in English.
Hailey Windham15:44
Yeah, yeah. Completely agree. So would you say is AI giving the advantage to fraudsters or to the fraud fighters?
Melinda Szabo15:53
I think both, but again, you can't fight AI without AI. Yeah, you have to have the same weapon, right? You can't fight a gun with a sword.
Hailey Windham16:05
Yes, you can try, but the only way you'll win is maybe if the gun doesn't have any bullets.
Melinda Szabo16:13
Exactly. So no, I believe it gives more advantage to fraudsters at the first place, because they have the capacity, and you know, the time and the energy and the money at this point, right, to build and think the ways to mistreat people. I think the other problem we are facing is the resources. Yes, the money and the time. Yeah, 100%.
Hailey Windham16:38
Okay. I know that you mentioned that, and I've talked to other credit unions and banks that they are willing to look at AI and see how they can use it, but maybe they haven't quite adopted it yet. But I'd love your opinion. You know, how do you think credit unions and banks realistically can use AI without over complicating things? Is there a particular scenario in which AI could actually really help that you've seen?
Melinda Szabo17:04
I think there are a lot of solutions here. I think what kind of happens is because of this, this is new, and we are a highly regulated industry, right? So it's hard to adapt. And I think everybody is worried to not be compliant. There are also certain tools out there, like we use BioCatch, that is AI driven, so it's an agentic AI-driven solution that looks at behavior, and I think that will be a very, very good first step for the adoption. But again, you can see solutions here that are bridging the gaps and collecting data for you to provide you to be able to make your decisions in the right way. So I think that would be a second step, definitely. 100%.
Hailey Windham17:51
Okay. Last one. What do you think is one misconception people have about AI and fraud prevention?
Melinda Szabo18:00
I believe a lot of people have a misconception about AI, what it can or cannot do, that it can just act by itself. It's not like in The Terminator, like then they will come and be alive and, you know, will kill entire humanity. It's basically a learning pattern. It learns whatever you feed into it. It's not going to think, it's not going to create something. It's not innovative like the human mind. I think that's what it is, and that's something to be concerned about.
Hailey Windham18:29
Yeah, I am. Just a last parting thought, I had one time someone asked me when we were talking specifically about fraud and AI, and I said, you know, the one thing that the fraudsters using AI, what AI doesn't have that the fraud fighters have is a gut. You have that gut instinct that tells you something's off, and it can just be a random, you know, I can see the same alert for two different members or customers, and I look at it and I go, this one just doesn't feel right, and it can match another one, but there's just this gut feeling that we fraud fighters have embedded within us that I feel AI will never have. But at the same time, you know, I use these LLMs, and sometimes they're my biggest hype woman. You know, I'll say, what do you think about this idea or this concept, and they'll, and it responds back, and it's like, oh girl, this is so you, and I'm like, okay, are you my friend? Like, what is this? I feel like I'm becoming friends with my AI.
Melinda Szabo19:31
I think you're absolutely right. I think it leverages your research. So it brings you the data, and you are still the one, the human has to make fraud or not fraud. And yes, we have the gut, and that's, it's not innovative, it's not a human, and it doesn't feel, or it doesn't have feelings, emotions that we have.
Hailey Windham19:51
Kind of like whenever the kids yell back at Alexa, and she's like, okay, well, I don't understand what you're saying, but have a nice day or something, and it's like, okay, I tell my kids, like, tell Alexa sorry.
Melinda Szabo20:03
No, I think it needs to be used as an additional tool, and it helps with our research. It pulls the data, but you have to have good data first. I think that's what it starts with.
Hailey Windham20:17
I love that. I love that so much. Thank you so much for joining me. Absolutely. Thank you. So one thing Melinda touched on that kept surfacing throughout Safeguard was the sheer scale of financial harm happening throughout, you know, scams and identity abuse. And when you start talking about AI-enabled identity fraud, the conversation quickly shifts towards synthetic identities. So next up is a conversation with Nyla Cortes, who I just had on the podcast also, who shared why synthetic identity fraud may be one of the biggest challenges financial institutions face over the next several years.
Nyla Cortes20:52
I really think that it's synthetic identities. I, it's a really, really huge fraud risk right now. And it's also very scary and very hard for financial institutions to really tell. Plus, we also really don't have the tools to help us really identify it. And tools are being created, tools are coming out, but I feel that right now it's so fresh and so new that only really the big banks are getting into that, where us smaller guys, it's really, really hard for us.
Hailey Windham21:21
I think too it's hard to get a good pulse read on where synthetics are right now, because so much of it is hidden in credit locks, right? So completely agree. Okay, so is, in your opinion, is AI giving the advantage to fraudsters or the fraud fighters?
Nyla Cortes21:41
That's really hard. I would say if you ask me right now, it's the fraudsters.
Hailey Windham21:46
Yeah.
Nyla Cortes21:46
You ask me in probably two years, I'll say the fraud fighters. I feel like sometimes fraud fighters, we can be kind of behind the curve, unfortunately. But we can only respond to what fraudsters are doing at the end of the day. So I think too it is a lot to do with the regulations around what we can use, how much PII we can share or not share, and then in what ways we are allowed to utilize the AI. Whereas the fraudsters truly have this open, open door plan. If it works, great. If it doesn't, we'll pivot. And there's no no harm, no foul for them, right?
Hailey Windham22:26
I know.
Nyla Cortes22:26
But for us, unfortunately, we are kind of bound by those rules and regs.
Hailey Windham22:32
Yes. Okay. So how can credit unions and banks realistically use AI without over complicating things?
Nyla Cortes22:39
I think that's the real question. I think that's something that banks and credit unions are really trying to figure out right now. Where's the balance? Where's the balance between what we actually utilize or integrate as AI and what we keep as a human-driven process? It's a precarious balance, I think.
Hailey Windham23:03
I love that you said human-driven process. I think that's one thing that no matter how advanced we get, right, no matter how much AI we're allowed to use, there's always going to be that gut reaction that a fraud fighter has that AI won't have. Although I was mentioning earlier in another interview that AI is getting pretty good because mine knows me pretty well. It's like my hype woman sometimes. I've never typed into Google and said, "Hey, what do you think about this?" And it goes, "That is perfect for you."
Nyla Cortes23:37
It might now.
Hailey Windham23:40
Thank you so much.
Nyla Cortes23:41
So true. So true.
Hailey Windham23:43
Last question. What's one misconception people have about AI and fraud prevention?
Nyla Cortes23:49
That it's going to replace humans. I think that's the biggest thing. I think that like you said, you can't replace the gut feelings, the reactions. The, literally, human bias. I know we talk about human bias in the negative a lot, but I think in some regards, human bias is important. And I think what a lot of people misunderstand or choose not to think about is that AI learns from humans. And so if a human doesn't teach it, it doesn't know. It can't develop the knowledge itself. So AI may catch up to us, but it's always going to be one step behind.
Hailey Windham24:32
Wow, that was nice.
Nyla Cortes24:34
Thank you.
Hailey Windham24:34
I like that. We didn't even practice that. I love that. Okay. Any parting thoughts? Any for those that maybe are having FOMO from missing Safeguard this year? What would you say?
Nyla Cortes24:47
Oh, come next year. Truly, like this has been such a wonderful experience. Getting to meet people in this type of environment has really, I think, changed the game for conferences. And I think we're going to actually probably see this more as we go forward in the future after other event organizers really see how successful this was. But I also think that something that this conference really did well is that it didn't solely focus on the business side. It really focused on the personal connection side, which I think is really what the fraud industry needed. We go to a lot of conferences, a lot of amazing conferences, but when you leave the conference, it can really feel like you were just talking about it and not being about it. And being here and meeting with people one on one, getting to know new people, really being forced to expand your network and meet people and talk about what's really going on has really successfully driven the knowledge gaps. The just overall sense of we are all working on the same thing. And we're all working towards the same goal. And there are so many other things out there that you just truly don't know about until you talk to the people and you're forced to. Truly.
Hailey Windham26:18
For sure. Well, thank you so much for joining me today.
Nyla Cortes26:21
Thank you so much for having me.
Hailey Windham26:21
So Nyla made an important point there. Fraudsters currently have fewer constraints than financial institutions do. But that doesn't mean that fraud teams are standing still. In fact, one of the more encouraging conversations throughout Safeguard was around how AI can actually help fraud teams scale faster than ever before. So next I spoke with Cady Frankel to talk about how AI can help analysts identify patterns, trends, and connections humans simply can't process at scale anymore.
Cady Frankel27:55
AI can increase the volume in which fraudsters can attack, because basically what you're doing is you're removing a lot of the manual and administrative labor that a coordinated and organized fraud program has to go through, right? Whether it's an individual or group, and you're taking away all of those complications, and you're reducing it so much that the volume in which you can, an institution or a fintech or a payment processor, can get hit at is going to increase, if it hasn't already increased, you know, tenfold with AI.
Hailey Windham28:30
It's making us, it's hitting us at scale.
Cady Frankel28:32
Yes, absolutely. And I think that, you know, a lot of institutions are prepared for it to some degree, and a lot are going to be playing catch up very quickly.
Hailey Windham28:43
Yeah. And so it's one of those that you, it's no longer an if, it's a matter of when, and making sure that you've done your due diligence to ensure that you partner with the right people, yes, because otherwise you really will just be a sitting duck.
Cady Frankel28:59
Yes, and can you scale quickly too, right? And I think that that's what AI enables us to do in return when we use it, is we can scale just as much as they can, because it's super easy to add on more agents, and it's not very easy to add in more fraud operations analysts on a moment's notice. So.
Hailey Windham29:15
100%. And it leads up perfectly to my next question. Would you say that is AI giving the advantage to fraudsters or the fraud fighters?
Cady Frankel29:25
Oh, that is a good question. I think at the moment, as with any kind of financial crime tool or exploitation that comes about, the winning hand is always going to be with the fraudsters at the beginning, because we have to catch up with what they're doing. I think that this is probably one of the times that we are going to be closer to being in lockstep than other types of financial crimes that have historically happened over the last decades, like with whether in money laundering or fraud, it takes us a lot longer to catch on. But AI also enables us to catch on quicker and to get up to speed faster with how we're combating it. So I think for the first time in a while, we have access to very similar tools, and so we're going to have a better response.
Hailey Windham30:10
Yes, someone said earlier, whoever uses the AI first, absolutely, if we can make it first, we'll finally get one up, right?
Cady Frankel30:15
Right. Right. Absolutely.
Hailey Windham30:19
Yeah. Okay. How can credit unions and banks realistically use AI without over complicating things?
Cady Frankel30:19
I think the easiest win to get started with is using it to make connections and to see the patterns better, faster, stronger, in a way that, you know, a human can look at something and it takes a lot of cognitive effort to come to a conclusion, which is absolutely right and sound. AI, you can dump a lot of the information into it, and it can say, hey, I've noticed a trend you might want to see, right? And that can help point out a fraud ring originating from an individual or some connection, whether you know it's with external banks or things like that. So I think even if you know banks haven't moved to like an agentic model, they can still use it to process data and come to conclusions that they can then send to humans much faster.
Hailey Windham30:59
I love that. And I love that truly, whenever you think about that manual process, if you've got five sets of data to look at, right, you're gonna take that meticulous time as a true fraud fighter would.
Cady Frankel31:10
And you have to gather them from different systems, and it's a lot of manual effort, which is absolutely worth it to come to good conclusions. But if we can get that information in one place and have a quick take on it much faster, point you to the things that maybe have the most bang for your buck as far as reviewing time, that's going to save, that's going to empower your analyst to come to decisions quicker, and maybe file SARs, exit customers, take actions against accounts a lot quicker.
Hailey Windham31:36
Yeah, I even had someone say earlier that it would help spot something that maybe the analyst has never come across before, because it's learning from all of us, right? And so if we, the more we teach it, the humans teach it, then those newbie fraud fighters, or level one that are moving up to level two, it's going to help empower them with knowledge that they didn't know, and then they can take it from there and research it, and you know, pick up that new trick.
Cady Frankel32:03
Right. Anybody who's ever been an analyst before knows that there is always several, several senior team members on your team that have seen everything back from the 80s to now, and they are phenomenal resources, and they know things that you have never heard, because there is a fraud scheme that might come back that was used, like I said, maybe in the 80s, and it's got a new face now, and you certainly weren't around in the 80s to see it. Yes, but Bob was there.
Hailey Windham32:29
For sure. I love that you say that. The knowledge gap is the one thing that scares me, between that wealth of information from the hands-on processing versus everything being so automated. But I think that that's also one of the things that kind of encourages me about this industry, is that we're not kicking anybody out, we're instead saying please teach us, please help teach these models, because we need that information from you.
Cady Frankel32:54
Absolutely. Could not agree more.
Hailey Windham32:56
Appreciate it. Last question, and then I'll let you go. What's one misconception people have about AI and fraud prevention?
Cady Frankel33:06
That's a good one. That's gonna stop me for a second. Yeah. A misconception people have about AI being implemented into fraud. I think that it's going to, the misconception that people might have is that it's going to reduce or eliminate people within the fraud cycle. And I think ultimately, not only are people going to have a conservative approach and always have a human in the loop, but ultimately the people who are going to make the most sound decisions are going to be your people with experience. You know, we've all been there before, right? Where you go to maybe consider filing a SAR, and ultimately you're better, you are able to find information, or to have that gut check that says this may be risky, but it's not necessarily suspicious or not necessarily exit-worthy. And you know, maybe AI may come to a different decision because it's going by the letter of the law of your policies and procedures, where people are still able to see some of the gray area that may move it in either direction. So I think people are not going to be eliminated or reduced from it, because that gray area is always, always going to exist.
Hailey Windham34:07
I agree. I also think it's important to call out the cultural differences, right? Whenever we use AI, it's completely unbiased, here's what we say in the policy, the procedures, this is what we should be looking at. But depending on what region you're in, or how you grew up, something's going to feel a little bit more suspicious. That's going to have that gut check you're talking about.
Cady Frankel34:26
Absolutely, yes. You know, people from different income levels, right, might spend differently, and you bring your own personal biases into things, right? So something might initially look suspicious to you that somebody's blowing one and a half million dollars on a home remodel, and somebody else is like, "Oh no, honey, that's Boca Raton. That's totally fine there."
Hailey Windham34:43
Completely. Yeah. Well, thank you again so much. I really appreciate your perspective, and for you joining me today on the podcast.
Cady Frankel34:47
Thank you so much for having me. Appreciate it.
Hailey Windham34:47
I really appreciated Cady's emphasis on preserving human judgment, because despite all the excitement around AI, nearly every leader I spoke with agreed on one thing. The goal is not replacing investigators. It's reducing noise and helping them focus on higher quality work. Amanda Balmer had some fantastic practical examples of what that actually looks like inside fraud operations today.
Amanda Balmer35:11
So I've listened to some of your other interviews, and I definitely agree with synthetic identities, basically stealing identities as well, true identity theft, and being able to create fake IDs and documents to support perpetrating identity theft and synthetic identity theft. However, I think one of the big things that Erin West highlighted in her presentation was the AI-driven manipulation that is going on from the fraudsters. We can deal with trying to stop these people that are coming into our credit union trying to steal money from us. But when you actually have these fraudsters using AI to support their claims that they are someone who they actually aren't, you know, it's so hard, they are able to better psychologically manipulate their victim and they get them so ingrained.
Hailey Windham36:12
This sounds horrible.
Amanda Balmer36:13
I totally agree. And there was one thing that Erin highlighted in her presentation that I feel is really important to share, is that with the use of AI, how they're pulling these images off of like Google and then making it look like they're there. Or like, even I remember being a part of a victim call with Erin for Operation Shamrock, and the guy was like, no, no, no, I know she was real because I myself looked on Google and saw these flowers were on this table, and this was a real thing. And we were like, no, that's not true. So seeing how they can manipulate these images, it's not even just the emotional manipulation of having a conversation with someone for six, eight months, a year or two years, you know, depending on how deep this game goes. But it's also the visual. I mean, you don't think that the average person, right, because you're just talking to an average person, you don't know. And then they're able to manipulate these images and make their claims just seem so true. I remember when I started out in fraud, we would ask individuals, did you actually FaceTime? Right now you can't just rely on, did you see a video of that? No. They can manipulate those images. They can fabricate those videos. And it adds to the psychological manipulation that gets that victim so ingrained in that fraud that it is so much harder to get them out.
Hailey Windham37:34
Yes, 100%. So would you say that AI right now is giving the advantage to the fraudsters or the fraud fighters?
Amanda Balmer37:41
I will say fraudsters, with the caveat that we are fighting. And conferences like this are bringing together those people that can actually make a difference in using AI the right way to help fight fraud and to help us become better than the fraudsters. So I am hopeful that we will be able to use AI better than the fraudsters one day.
Hailey Windham38:05
Yeah, for sure. I think AI is scary but hopeful, like that just kind of seems to be the same with everyone that I talk to. Okay, how do you think that credit unions and banks realistically can use AI without over complicating things?
Amanda Balmer38:21
Yeah, some of the sessions I've attended have done a really great job of highlighting this. You're not replacing that actual individual. You can't replace that Spidey sense they call it, and you can't replace that actual decision making because there's so many different factors. There's a human factor that you need to take into consideration when you're making decisions. However, I think a lot of the speakers have highlighted that we can use AI to help and empower our team members that are fighting fraud. It can gather data from different places, and it doesn't take them an hour to go look for all of this data. It can make recommendations, and it can actually even be trained to say, hey, this is pretty similar to X, Y, and Z scam. If your fraud fighter has not come across that scam before, they can use that as a guidepost to look up, hey, what is this fraud, and then make their decision better. So I think it really is just empowering us to make better decisions.
Hailey Windham39:16
I love that too, in the sense of imagine if you've got a new level one analyst who's fresh off the teller line or fresh off the call center, who you know has a passion for fraud and wants to be in it. And now you hired them and they're great. They've got a great instinct, but they don't know that type of scam yet or something. So then whenever you have something that can prompt it, and I even think like in an investigation, I know Frank McKinnon mentioned this before in a session, maybe not here, but somewhere else, where he said, you know, we can upload the case information into whatever system and then it can say, well, hey, did you consider, and then you go down another linking chart or analysis or something. And it wasn't something that you thought about before, but it's helping to prompt so that we don't miss things. That human error happens, right, but AI can help us with that.
Amanda Balmer40:07
Absolutely, I totally agree.
Hailey Windham40:09
Love that. Okay. And then last one. What's one misconception people have about AI and fraud prevention?
Amanda Balmer40:14
Definitely that AI is going to take their job. Yeah, there is no way that AI is going to take your job. It is just going to make you be a better fraud fighter at the end of the day. That is it. It's not going to take your job. It is going to help you. 100%.
Hailey Windham40:28
One area where these conversations started getting especially interesting was around identity, authentication, and AI agents. Not just humans using AI, but AI acting on behalf of humans. And that introduces an entirely new set of operational and governance questions. So next I spoke with Sinduri about identity verification, AI-driven workflows, and why institutions need to think carefully about balancing innovation with risk management.
Sinduri Valleru41:01
I would say more around identity. So that would be the biggest challenge, because a lot of us are traditionally doing regular verifications and authentications. So I believe that's where we need to be looking for how to identify a person, improve on techniques that we currently using. So I'm sure with AI that's also helping us to do it better. So I would say identity.
Hailey Windham41:28
I love that. And you teed up the next question perfectly. And it is, is AI giving the advantage to the fraudsters or the fraud fighters?
Sinduri Valleru41:37
I would say both. It's just that who is using it first. It's definitely giving advantage to fraudsters. Before, it was not a thing everybody knew how to hack or, you know, because a lot of things are tied up with technology, they need to know how to code or do things. But now it's becoming so easy, right? So that's the advantage. But for us, like being in the fraud strategy and team to prevent the fraud, we need to also catch up the pace. But the real challenge is how soon we do it, especially because for fraudsters they don't have any regulations, compliance. We do, we have compliance, governance, risk, we need to evaluate all that. But that's where I'm looking forward to see how fast we can do those things, align with our regulations, and, you know, have some good governance also, because without understanding the risk, even just implementing AI is not a great strategy. So we need to evaluate, but we need to be in a fast paced environment.
Hailey Windham42:52
I completely agree, and love that you said that. Okay. How can credit unions and banks realistically use AI without over complicating things?
Sinduri Valleru43:02
I would say it should be a step by step process. So first is how we can improve our productivity and efficiency using AI. So it's definitely not a replacement. It's more like, okay, I'm thinking in my team, if I use the AI agent, how much productivity and efficiency I can improve, so that we can focus on other things. And that's where also catching up fast is what is determined, like not just like building the whole ecosystem with AI, like slowly. Okay, this is where I spend a lot of my time. I'll use AI agents to do it, and I'll focus on other things, so that way we can catch up the pace that we were talking about also.
Hailey Windham43:48
So I would look at it from that aspect. That way too we're not stuck in the mundane task. I remember advocating for my fraud team back in the day when optimizing our alerts, right? False positive ratios are so high that, you know, I can remember sitting in the seat going, "Oh my god, not again." You know. And so that is what leads to more human error, right? Because it's like being bored in the task. Whereas if we could, you know, get rid of some of those, or have like a level one agentic AI like process those for us, and then just elevate the other ones, I think it makes not only the efficiency happen, but then it makes the productivity and the value of what they actually produce within these alerts, and in these cases, it makes our team so much better.
Sinduri Valleru44:32
That's true. This is where I go tie back to the data, because when an analyst or investigator have to evaluate a fraud case or detect the fraud, they need to look at loads of data points. With agents, if we can gather all that, show it to them, say these are the risk factors why this particular alert has triggered, and then they can make a decision, less manual errors and faster, you know, productivity.
Hailey Windham44:57
I love that. Absolutely.
Sinduri Valleru44:58
That's just my first use case.
Hailey Windham45:00
So we don't have to go into all these other systems and pull data, it can all be right here in one spot. I love it.
Sinduri Valleru45:05
Also tell them exactly what the risk is to look for.
Hailey Windham45:08
And prompting, you know, hey, like if we can put in our summary of what a case is, or what the resolution is, and then it can spit back out and go, hey, well did you consider this? And then that's when it's like, oh yeah, let me go back and look one more time. And then it just gives us an additional level of kind of like that oversight, but it's like that unbiased oversight of, hey, make sure you consider this aspect as well. So I love that. Okay. Last question, what's one misconception people have about AI and fraud prevention?
Sinduri Valleru49:37
This is a tricky question. I mean, this is more about how do you identify a human doing things or an agent doing these things, because you are on the other side of looking at transactions coming through, payments happening, and all of that. But from a consumer perspective, they want to use AI, and like they're like, "Okay, now I can just tell my agent, go book my tickets." And for me, like, everybody is like, "Now, how do I identify that? Is it a human or an agent?" That's where everybody is more worried about it. I think that's where we can use AI as well to identify that. So I'm sure there are things which need to be evolved around that area. I don't know if everybody is at that point to identify the agents better, and if it is linked to this person, etc. But I'm sure that would be a thing, like we need to adapt. It's not going to change. Everybody is going to use AI agents to do their purchases, but we should also know how to identify those things. But that shouldn't be the biggest challenge, I would say.
Hailey Windham50:54
Yes, yes, I agree. The agentic commerce is one of those that it's pretty interesting, but at the same time I'm on the hunt for a bargain always, and I don't think I could ever trust the AI agent to actually shop the way I would shop, you know. So that's one of my hesitances, but I know that others don't really have that.
Sinduri Valleru51:16
I mean, that's where some of them are so hesitant, knowing about AI, all what it does. But not people who don't know much about it. Oh, I can use this as a convenience, right, right, right. So.
Hailey Windham51:29
Yeah, one use case I was talking about with someone yesterday was the one way I would see that I would definitely use agentic commerce is if it like connected with my refrigerator and told me when my milk was running low or when it was about to expire, and it could order me more, or like the laundry detergent, because we're always running out of that too. That'd be great to have it on auto, like so it knows, alright, you're running low, we're going to go ahead and order some and have it, you know, set to the approved brand, and you know, a potential threshold for dollar amount to spend. But anyways, so I thank you so much for joining me. I really appreciated your perspective.
Sinduri Valleru52:06
Thank you. Thanks for having me.
Hailey Windham52:11
By this point in the conference, one thing became very clear. AI isn't some future state conversation anymore. It's already here. The real question now is how financial institutions operationalize it responsibly while fraud threats continue evolving in real time. And to close things out, I sat with the one and only Steve Lenderman to talk about what comes next, including AI agents, know your agent frameworks, and why fraud fighters may finally be starting to regain some ground.
Steve Lenderman52:43
Yeah, everybody's talking about agentic AI. We talked about KYC, KYB, and it's KYA, know your agent. We really need to understand who's authorized those transactions, who's doing the searches. And I've been looking at it from a little deeper perspective around synthetic identity, which is something I play in, right? Is now we can start building agents to actually start giving life to synthetic identities. So in the future, look for that coming up.
Hailey Windham53:48
Yeah, for sure. Would you say is AI giving the advantage to fraudsters or fraud fighters right now?
Steve Lenderman53:55
I think the pendulum is swinging back to the fraud fighters, because I think AI is becoming more adaptable, and governance and regulations are enabling us to do more with AI than we had been in the past. So I think we're closing the gap on the bad guys, but the pendulum was certainly swinging our way for a change. Yeah.
Hailey Windham54:12
Great. Okay. How can, in your opinion, how can credit unions and banks realistically use AI without like over complicating things?
Steve Lenderman54:21
Yeah, so first I always warn, don't use it just to say you're using it. Make sure you understand you have a use case and stay within that use case. A lot of organizations, particularly, think they can solve all their problems, and they could get scope creep, and then they can't solve any problems. So understand where you're going to use it. It could be something as simple as just gathering information, something as in writing reports, creating documents, prioritizing queues. Those things I think are really important, that you focus in on that, and don't try to boil the ocean with AI, right? Because a lot of organizations say they're using AI, and they're really not. It's just that kind of a verbiage right now.
Hailey Windham54:56
Okay. Last one. What's one misconception people have about AI in fraud prevention?
Steve Lenderman55:02
That it's gonna solve all your problems, for sure. Another big misconception is it's gonna take your job. Matter of fact, we're using it at I Solve on a daily basis, and it's creating actually more good work for us. So it's creating opportunity for more investigators, analysts, etc. Right? And so that I think is a misconception, that it's going to eliminate your job. It just makes you more efficient. It actually makes the investigator analyst able to work quality work, not some of the noise that we've been getting in the reports.
Hailey Windham55:29
They can actually dive into the stuff they like doing, something that matters. Yeah, absolutely. Well, thanks again for your time. I appreciate you coming on.
Steve Lenderman55:33
It was fun. Yeah, see you next week. All right.
Hailey Windham55:33
Okay, so that's a wrap on the interviews from Safeguard. I want to first give a huge shout out and kudos to Mittel Parmar and Brian Davis and the entire team from Safeguard for an absolutely amazing event. I honestly was blown away. The Broadmoor was absolutely beautiful, breathtaking sights. Obviously I was able to spend a lot of time with some people who mean a lot to me in the industry. I was able to meet some new people and it was just an incredible experience. I also want to shout out Sardine for having the most epic reception, dinner, whatever to welcome everyone. It was so wonderful because it was purposeful, right? There was not, it wasn't how can we sell you something? It was just, hey, let's talk. What's going on in your world? And we absolutely loved it. I loved being there. I loved seeing the Sardine logo there. Obviously it's like embedded in me now. But yeah, it was just a great experience, a great event, and just want to of course give credit where credit's due. They did such an amazing job. And then a huge thank you to everyone who took the time to sit down with me during Safeguard. And another shout out to Nyla who didn't mind rallying a few to come and have a conversation with me. Her and Angela both, they went out and grabbed a few people, so I was very grateful for that. What stood out most from these conversations though, it wasn't fear. It was realism. Fraud fighters understand the risk. We know synthetic identities are evolving. We know scams are scaling. We know AI lowers the barrier for attackers. But we also know this industry is adapting incredibly fast. And maybe most importantly, we're finally seeing more honest conversations around governance, operational pressure, investigator enablement, and responsible implementation. And that matters. Because the future of fraud prevention isn't going to be built by technology alone. It's going to be built by the people willing to collaborate, challenge assumptions, and evolve alongside it. Thank you so much for listening to Fraud Forward. And again, I just I'm so grateful for this community, for this industry, and for those of you that are willing to come on and share your insights. It's so important in this industry that we share, that we collaborate. So again, I just thank you so much.
Hailey Windham58:00
Stay vigilant, stay informed, and keep moving fraud forward.
View full transcript