WEBVTT - Thanks to AI, Fraud Has Never Been Easier 0:00:00.080 --> 0:00:04.640 People should assume and fortunately like, assume something's going to 0:00:04.720 --> 0:00:08.360 go wrong and kind of work backwards from there. Right, 0:00:08.560 --> 0:00:11.320 you know, assume that some emergency be like, Okay, thank 0:00:11.360 --> 0:00:12.720 you very much, I'm going to call you back. 0:00:13.400 --> 0:00:13.600 Right. 0:00:13.720 --> 0:00:17.560 Your bank will never call you. Right, your bank calls you, 0:00:17.600 --> 0:00:19.520 hang up the phone, look in the back of your card, 0:00:19.600 --> 0:00:24.080 call the bank. Right. If you think something's a scam, 0:00:24.480 --> 0:00:24.959 probably is. 0:00:29.440 --> 0:00:32.080 I'm Laur Siegel and you're listening to Mostly Human, a 0:00:32.159 --> 0:00:37.159 tech podcast through a human lens. So Johnny, I'm so 0:00:37.200 --> 0:00:39.760 happy to have you here today. We had a brief 0:00:39.800 --> 0:00:44.000 conversation in Davos about the future of identity and fraud 0:00:44.240 --> 0:00:47.919 and it terrified me. But you are you are the 0:00:47.920 --> 0:00:50.879 person that people go to to talk about this, and 0:00:50.920 --> 0:00:53.040 you know a lot about the bad guys on the 0:00:53.080 --> 0:00:56.320 internet and what they're doing and how they're doing it. 0:00:56.440 --> 0:01:01.520 So without sounding like major CEO elevator pitch, which explain 0:01:01.640 --> 0:01:04.399 to us common folk what it is that you do. 0:01:04.640 --> 0:01:07.600 So, sir, Cure has built the largest identity verification, fra 0:01:07.600 --> 0:01:10.279 aull prediction and any money laundering platform in the world. 0:01:10.400 --> 0:01:16.679 So we build AI on offense to enable enterprises fintech 0:01:16.720 --> 0:01:22.920 financial services insurance, gaming, governments, marketplaces really give them the 0:01:22.959 --> 0:01:26.480 tools and algorithms to be able to combat folks that 0:01:26.520 --> 0:01:32.280 are committing scams, stealing identities, perpetuating different types of fraud. 0:01:33.080 --> 0:01:35.080 And so we've been at it for a little over 0:01:35.120 --> 0:01:38.200 ten years, so we've been studying these problems pretty intimately 0:01:39.040 --> 0:01:41.200 and have I think a pretty unique seat at the 0:01:41.240 --> 0:01:46.480 table at onboarding, at log in and account takeover and 0:01:46.680 --> 0:01:51.240 money movement, call center, digital in branch. We see all 0:01:51.280 --> 0:01:54.680 different types of attacks, and yeah, I'm excited to be 0:01:54.720 --> 0:01:57.280 with you today and share a little bit about what 0:01:57.320 --> 0:01:58.040 we see in the world. 0:01:58.120 --> 0:02:02.080 I am curious you mentioned kind of for offense. We 0:02:02.200 --> 0:02:04.120 hear over and over a lot of tech folks say, 0:02:04.160 --> 0:02:06.640 okay AI is going to counter AI. AI can do 0:02:06.680 --> 0:02:08.040 a lot of bad, but it can also do a 0:02:08.040 --> 0:02:10.720 lot of good. So talk about how you're using machine 0:02:10.800 --> 0:02:14.760 learning and artificial intelligence to catch to prevent fraud and 0:02:14.880 --> 0:02:17.880 why that's a lot better than the old fashioned methods. 0:02:18.000 --> 0:02:19.200 Can explain it to us. 0:02:19.240 --> 0:02:22.240 I would say, like, we are, you know, not in 0:02:22.360 --> 0:02:25.160 the room at at the company. 0:02:24.919 --> 0:02:26.760 With a lot of security. So I know you're not 0:02:26.800 --> 0:02:28.640 in the room at the company. 0:02:28.280 --> 0:02:31.480 That's do you have a I guess I imagine you 0:02:31.520 --> 0:02:33.840 would have a lot of security given what it is 0:02:33.880 --> 0:02:34.399 that you do. 0:02:34.880 --> 0:02:36.600 I was a little I was a little tardy to 0:02:36.720 --> 0:02:38.600 this our start here because I was on the phone 0:02:38.639 --> 0:02:41.680 with our c sale going through some more recent changes 0:02:41.720 --> 0:02:44.560 that we're making. But yeah, it's a it's pretty lock 0:02:44.960 --> 0:02:45.760 pretty locked down. 0:02:46.040 --> 0:02:49.560 Are people trying to I mean because of what you do? 0:02:49.720 --> 0:02:50.440 And I promise I. 0:02:50.440 --> 0:02:52.520 Won't go on this whole thing, but it is fascinating 0:02:52.560 --> 0:02:55.760 because you you literally protect some of the largest companies 0:02:55.840 --> 0:02:57.960 from fraud, uh and. 0:02:57.919 --> 0:03:00.919 All sorts of attacks. Like do people try to attack you? Guys? 0:03:01.240 --> 0:03:03.560 I think most companies that have something of value are 0:03:03.639 --> 0:03:07.320 under attack all the time. Fraudsters we often say they're 0:03:07.480 --> 0:03:10.000 they're like water. They find the path of the least resistance. 0:03:10.560 --> 0:03:13.919 They're very good at like testing weak points, right, they 0:03:14.000 --> 0:03:16.359 just will perpetually test and test and tests, try to 0:03:16.400 --> 0:03:19.959 find policy policy gaps, try to find weak employees. You know, 0:03:20.000 --> 0:03:22.600 it's unbelievable, you know, Laurie, if you're just started so 0:03:22.720 --> 0:03:26.800 cure tomorrow, like you'd immediately be attacked with gift cards, 0:03:26.840 --> 0:03:31.720 CEO requests, inbounds, phishing attacks. They're very, very, very good 0:03:31.720 --> 0:03:34.040 at like saying like, Okay, we see someone on how 0:03:34.040 --> 0:03:38.360 to LinkedIn change boom attack, it's it's it's remarkable, but 0:03:38.440 --> 0:03:41.640 so anyway, so on the solution set, so we do 0:03:41.760 --> 0:03:45.680 lots of red teaming. So part of the a on 0:03:45.840 --> 0:03:49.520 offense is we create a lot of synthetic identities ourselves. 0:03:49.840 --> 0:03:53.280 We mature a lot of synthetic identities ourselves. We're trying 0:03:53.320 --> 0:03:57.120 to build and create synthetic identities that kind of permeate 0:03:57.120 --> 0:04:01.720 throughout the organ kind of ecosystem. We are creating deep 0:04:01.760 --> 0:04:04.880 fakes ourselves. We are running injection attacks. We have an 0:04:04.960 --> 0:04:07.600 internal red team that basically is perpetually trying to figure 0:04:07.640 --> 0:04:10.560 out how to beat our own systems. We also use 0:04:10.720 --> 0:04:16.320 multiple external teams to simulate similar types of attacks. They 0:04:16.360 --> 0:04:20.719 go create fake identities, manipulated identities, They go higher forgers 0:04:21.160 --> 0:04:25.920 to go actual go build these fake documents themselves. It's 0:04:25.960 --> 0:04:28.200 a perpetual game of trying to stay two or three 0:04:28.279 --> 0:04:31.280 or four steps ahead in terms of how we are 0:04:31.400 --> 0:04:34.000 kind of dog fooding our own capabilities. 0:04:34.080 --> 0:04:34.640 It has to be. 0:04:34.680 --> 0:04:37.080 Every day is kind of its own little adventure where 0:04:37.120 --> 0:04:41.480 you are. Could you take us into your day to 0:04:41.600 --> 0:04:44.800 day what types of customers do you have, and what 0:04:44.880 --> 0:04:48.599 kind of threats are you seeing. Obviously they've evolved over 0:04:48.640 --> 0:04:54.039 the last years. But take us behind these very locked 0:04:54.120 --> 0:04:57.120 and secured doors and tell us what you're seeing. 0:04:57.480 --> 0:05:01.920 All customers that move money, right, or something of value 0:05:02.000 --> 0:05:07.360 or under attack all the time. Fake identities, stolen identities, 0:05:07.400 --> 0:05:10.919 manipulated identities, synthetic identities are being used to try to 0:05:10.960 --> 0:05:16.360 open new accounts, new bank accounts, new investment accounts, anything 0:05:16.400 --> 0:05:19.240 that can be used to like move money. It could 0:05:19.279 --> 0:05:22.400 be folks that are perpetuating promo abuse. Set one of 0:05:22.440 --> 0:05:25.320 the sportsbook or prediction markets, right, it could be one 0:05:25.360 --> 0:05:27.880 of the e commerce and marketplaces that are like get 0:05:27.920 --> 0:05:30.120 one thousand dollars, Laurie, if you want to become a 0:05:30.160 --> 0:05:33.640 new driver, sign up for this new you know, e 0:05:33.680 --> 0:05:37.920 commerce subscription, you know, get your first six months free. 0:05:37.960 --> 0:05:41.560 Any of those things. Fraudsters are trying to steal that 0:05:41.680 --> 0:05:45.080 thing of value, right, Give twenty bucks, Laurie. If you 0:05:45.160 --> 0:05:49.040 recommend a friend, right, well, if that fake account then 0:05:49.120 --> 0:05:53.200 goes and recommends fifty fake friends, right, there's may one 0:05:53.200 --> 0:05:58.320 thousand bucks. And so you have. All of this is 0:05:58.360 --> 0:06:00.320 like a game of kind of cat and mouth all 0:06:00.360 --> 0:06:02.960 the time. So i'd say, like I think to maybe listeners, 0:06:03.760 --> 0:06:06.599 there's just a tax going on all the time. Fraud 0:06:06.680 --> 0:06:11.839 rings in mass are hitting most of the consumer applications 0:06:11.880 --> 0:06:15.760 that that you deal with. Maybe one of the interesting 0:06:15.800 --> 0:06:17.720 things that has been more recent i'd say over the 0:06:17.800 --> 0:06:22.040 last kind of twelve months, has been the North Korean 0:06:22.560 --> 0:06:27.320 you know it worker attack. That has been something that 0:06:27.800 --> 0:06:31.599 we've seen a huge prolific kind of acceleration of in 0:06:31.720 --> 0:06:36.640 terms of employment. We are seeing a like enormous some cases, 0:06:36.680 --> 0:06:40.600 like eighty ninety percent of all job applications are fake 0:06:40.839 --> 0:06:43.960 or fraudulent organized crime ry. 0:06:45.160 --> 0:06:47.200 Say that again, I mean, what do you mean like 0:06:47.279 --> 0:06:50.640 coming into wear remote jobs people. So you're if I'm 0:06:50.720 --> 0:06:53.200 understanding crack and please correct me if I'm wrong. So 0:06:53.240 --> 0:06:56.680 you're saying this is a new deep fake type of 0:06:56.720 --> 0:07:00.919 scam where you would have someone from North Korea trying 0:07:00.960 --> 0:07:03.560 to apply for a job here, say in the United States, 0:07:04.120 --> 0:07:07.839 taking on the identity of is taking on a synthetic identity, 0:07:07.920 --> 0:07:10.760 so a completely different identity using deep fakes? 0:07:10.800 --> 0:07:13.240 Is that correct? Am I understanding that? 0:07:13.360 --> 0:07:16.720 Yeah? North Crared is estimated about six thousand workers are 0:07:16.760 --> 0:07:22.560 operating inside US technology companies right now under fake or 0:07:22.640 --> 0:07:26.240 synthetic identities. Now, the synthetic identities can be you know, 0:07:26.280 --> 0:07:29.160 it can be a manipulated version of LORI like on 0:07:29.200 --> 0:07:34.160 a resume, it could be a deep fake of LORI, 0:07:34.840 --> 0:07:40.000 right that's controlled by someone else. It could be any 0:07:40.080 --> 0:07:44.240 combination of these things. Because you get paid, You get 0:07:44.240 --> 0:07:47.200 a paycheck every two weeks, right, and you can use 0:07:47.240 --> 0:07:49.520 AI to do a lot of these jobs, like certainly 0:07:49.520 --> 0:07:52.480 lower level jobs. Many cases, folks are obviously trying to 0:07:52.480 --> 0:07:57.360 get access to trade secrets and IPE and in many cases, 0:07:57.400 --> 0:07:58.920 you know, folks are trying to get access to to 0:07:59.560 --> 0:08:03.240 code baits or create back doors for some of these 0:08:03.240 --> 0:08:07.000 foreign regimes. And so yeah, I mean, unfortunately, the money 0:08:07.000 --> 0:08:10.240 then goes back to some of these foreign adversaries that 0:08:10.360 --> 0:08:13.960 are funding weapon programs, that are funding wars, that are 0:08:14.000 --> 0:08:18.040 funding obviously folks that are not super friendly to the 0:08:18.920 --> 0:08:22.000 Western world. And we're really seeing it across the board. 0:08:22.400 --> 0:08:24.960 I mean, there's been some public attacks with some of 0:08:24.760 --> 0:08:29.800 the large crypto exchanges, but we've seen infiltration to some 0:08:29.880 --> 0:08:33.080 of the major banks, We've seen infiltration to some of 0:08:33.120 --> 0:08:39.640 the support centers, and then certainly really widespread the number 0:08:39.640 --> 0:08:43.920 of fake job applicants that everyone is seeing is growing 0:08:44.120 --> 0:08:45.400 very very very fast. 0:08:45.679 --> 0:08:48.959 You know, it is I keep hearing over and over 0:08:49.000 --> 0:08:53.440 again workplace job applicants fraud. So it certainly seems like 0:08:53.480 --> 0:08:55.920 this is something that's just like as you're saying, over 0:08:55.960 --> 0:08:56.640 the last year. 0:08:56.800 --> 0:08:59.520 So it's just becoming a completely new attack. 0:09:00.040 --> 0:09:03.160 Walk us through how companies can even look out for that, 0:09:03.320 --> 0:09:06.240 how how people even know and is this just fake 0:09:06.480 --> 0:09:10.560 applicants or is it actually deep fakes as in I'm 0:09:10.600 --> 0:09:13.240 interviewing for a job, but it's not really me. I 0:09:13.280 --> 0:09:16.240 have it's a video, a deep fake video. I mean, 0:09:16.280 --> 0:09:18.640 can you explain kind of very in a granular way 0:09:18.720 --> 0:09:20.880 what that looks like and how that plays out, especially 0:09:20.880 --> 0:09:21.920 are these remote jobs. 0:09:22.120 --> 0:09:24.520 Yeah, you know, I'll kind of frame it because it's 0:09:24.520 --> 0:09:27.040 it's a similar attack pattern across all of these LORII. Right, 0:09:27.080 --> 0:09:30.760 So it's a you know, if you're applying for a job, 0:09:30.880 --> 0:09:35.040 you submit, you know, information on your resume. You likely 0:09:35.080 --> 0:09:39.079 had the resume auto complete the job application. Your resume 0:09:39.520 --> 0:09:44.439 likely is even fake, right, the PII is manipulated that 0:09:44.800 --> 0:09:47.880 the email doesn't really belong to Laurie or is brand new. 0:09:48.600 --> 0:09:51.360 It's you know, on the resume, it's like, hey, Laurie, 0:09:51.679 --> 0:09:55.960 somehow worked at Gemini and open AI and Anthropic in 0:09:56.000 --> 0:09:57.120 the last five years. 0:09:58.240 --> 0:10:01.839 Busy, they offer really good they're all trying to poach. 0:10:01.920 --> 0:10:07.200 It could happen, so all the best resumes, like you 0:10:07.200 --> 0:10:10.080 see these pristine resumes and they're fake. All like everyone 0:10:10.080 --> 0:10:12.000 that's a fake applicant happens to have an IQ of 0:10:12.000 --> 0:10:13.800 like one hundred and fifty and you know has had 0:10:13.800 --> 0:10:15.800 that many of the best jobs in the world. But 0:10:15.960 --> 0:10:19.200 so you know, again, I think whether it's a fintech application, 0:10:19.480 --> 0:10:23.160 or gaming application or unemployment application, you see a lot 0:10:23.160 --> 0:10:26.679 of the same patterns. There's geospoofing, so it's like Louri's 0:10:26.679 --> 0:10:29.160 in New York now of a sudden, Laurie's in Vancouver. 0:10:29.679 --> 0:10:32.000 You see emails that are new, you see phones that 0:10:32.040 --> 0:10:34.920 are manipulated, lots of m v and o's, You see 0:10:35.000 --> 0:10:39.560 domain email domains that are higher risk domains. You see 0:10:39.559 --> 0:10:44.080 fronts taking over entire domains. And then you know again 0:10:44.160 --> 0:10:47.120 as people are kind of going through the process, documents 0:10:47.160 --> 0:10:51.600 are fake. Documents are manipulated, right, maybe they're injecting a 0:10:51.640 --> 0:10:54.520 deep fake image, and so you're picking up lots of 0:10:54.520 --> 0:10:58.160 little patterns. So like all of our hands like have 0:10:58.240 --> 0:11:01.240 a little bit of a micro shake, right, and so 0:11:01.360 --> 0:11:04.000 you can actually pick up that in an accelerometer of 0:11:04.040 --> 0:11:06.120 a phone, so you can actually determine if a phone 0:11:06.200 --> 0:11:09.520 is being held. You can look at like the gimbal 0:11:09.559 --> 0:11:12.360 of a of a camera to actually determine the angle. 0:11:12.720 --> 0:11:15.440 And so does the angle at that moment match the 0:11:15.480 --> 0:11:18.680 actual video file that's being submitted, And so even in 0:11:18.760 --> 0:11:22.440 session you could see these like little little shifts that 0:11:22.559 --> 0:11:26.280 could determine is there actually a manipulation of the video 0:11:26.720 --> 0:11:30.040 or image. And so it's again it really becomes a 0:11:30.120 --> 0:11:32.680 kind of a combination of all these things, whether it's 0:11:33.000 --> 0:11:37.000 a job application, a sportsbook application, you know, if someone 0:11:37.040 --> 0:11:41.280 filing their taxes, someone registering to vote, you know, any 0:11:41.320 --> 0:11:45.240 of these become you know, attack vectors where you know, 0:11:45.280 --> 0:11:47.640 these types of kind of patterns can be used to 0:11:48.120 --> 0:11:52.000 you know, unfortunately hide nefarious or in many cases like 0:11:52.040 --> 0:11:55.160 criminal activity that we're having to counteract. 0:11:55.360 --> 0:11:57.199 I mean, and you have to get more and more 0:11:57.240 --> 0:12:00.360 sophisticated as the attacks get more sophisticated. I want to 0:12:00.400 --> 0:12:04.320 paraphrase something you said a little while ago. You said, 0:12:04.360 --> 0:12:07.280 to tell a live human from a video feed was 0:12:07.360 --> 0:12:09.160 visible to the naked eye just two years ago, but 0:12:09.200 --> 0:12:12.720 it's no longer detectable that way, and establishing trust between 0:12:12.720 --> 0:12:15.640 strangers transacting online might be one of the most defining 0:12:15.920 --> 0:12:18.880 conversations in the next couple of years. It used to 0:12:18.880 --> 0:12:20.319 be in tech that data was gold. 0:12:20.360 --> 0:12:21.640 Is trust the new gold. 0:12:22.160 --> 0:12:24.960 You need a lot of data to determine trust. And 0:12:25.000 --> 0:12:28.680 that's where the machines actually can reason over way more 0:12:28.720 --> 0:12:32.719 information than our human eyes can. Right in many of 0:12:32.760 --> 0:12:36.280 these sessions, we have thirty forty to fifty models LORI 0:12:36.760 --> 0:12:41.040 running to try to have the machines determine whether they 0:12:41.040 --> 0:12:44.120 should trust the session or not. And you know, you 0:12:44.120 --> 0:12:47.120 can be looking at thousands or tens of thousands of 0:12:47.120 --> 0:12:53.400 elements in milliseconds. Yeah, which which you know again goes 0:12:53.440 --> 0:12:57.600 back to this like AI fighting AI in a world 0:12:57.640 --> 0:13:00.120 where trust, yeah, on the internet is really hard to 0:13:00.120 --> 0:13:00.840 to determine. 0:13:00.960 --> 0:13:02.040 Can you just tell me? 0:13:03.679 --> 0:13:05.280 I know, you probably have to sign a lot of 0:13:05.320 --> 0:13:07.400 papers that make it so you can't say a lot 0:13:07.440 --> 0:13:08.520 about these types of things. 0:13:09.080 --> 0:13:11.959 But like, what's the craziest thing you've seen? 0:13:12.360 --> 0:13:15.520 Like you've just seen some insane things in your career. 0:13:15.640 --> 0:13:16.640 I imagine like. 0:13:17.520 --> 0:13:19.840 When you look back at the movie of your life, 0:13:20.280 --> 0:13:22.559 what is going to be like the plot point of 0:13:23.440 --> 0:13:28.080 you at this company building this and this thing shocked you, Like, 0:13:28.120 --> 0:13:29.080 what is that scene? 0:13:29.559 --> 0:13:31.760 I think, honestly, I think some of the like in 0:13:31.800 --> 0:13:36.120 person stuff ironically is the most brazen. Like what, Yeah, 0:13:36.160 --> 0:13:39.920 like we stopped someone a couple of weeks ago that 0:13:40.320 --> 0:13:42.960 I think they were in something like thirty nine different 0:13:43.080 --> 0:13:47.880 zip codes in like eight or nine different states, with 0:13:48.120 --> 0:13:53.839 like hundreds of different identity documents where this individual had 0:13:53.880 --> 0:13:58.080 been on all types of these bat like blocklists. So 0:13:58.120 --> 0:14:03.040 they're going into banks, getting stopped and then just going 0:14:03.160 --> 0:14:05.480 to another one. And so yeah, I think just the 0:14:05.480 --> 0:14:08.680 the brazen thiss that that we see and and and 0:14:08.760 --> 0:14:10.840 of course that's you know, that's that's in person, which 0:14:10.880 --> 0:14:13.760 I think is is is its own unique ye, but 0:14:13.880 --> 0:14:18.960 online like we've seen attacks where we will see you know, 0:14:19.080 --> 0:14:23.200 Laurie's face someone pretending to be LORI like thousands and 0:14:23.280 --> 0:14:26.880 thousands of times and a couple of like a couple 0:14:26.880 --> 0:14:31.800 of hour period across thirty forty fifty customers. Yeah. Again, 0:14:31.840 --> 0:14:33.640 I think that that this is the this is the 0:14:33.640 --> 0:14:36.880 fun cat and mouse game that we get to play 0:14:37.280 --> 0:14:39.320 all the time. And and uh yeah, I think just 0:14:39.320 --> 0:14:43.800 the sheer, the sheer sophistication of the automation and the 0:14:43.920 --> 0:14:47.160 number of different attacks people are to perpetuate today and 0:14:47.200 --> 0:14:50.960 the surface area on the Internet that they're able to, uh, 0:14:51.120 --> 0:14:54.320 you know permeate is is Yeah, to me one of 0:14:54.360 --> 0:14:56.440 the one of the most kind of fun parts of 0:14:56.440 --> 0:15:00.120 the job, because we're I often analogous our work to 0:15:00.200 --> 0:15:03.720 like special teams in sports. Uh, you know, no one, 0:15:03.840 --> 0:15:06.240 no one cares about the kickers until something goes wrong, 0:15:07.280 --> 0:15:09.800 and so you know, we we we very much are 0:15:09.840 --> 0:15:12.080 play and I think a really important special teams role 0:15:12.160 --> 0:15:15.360 and how the Internet operates right and how you know, 0:15:15.400 --> 0:15:18.200 many of these organizations you know, are are are able 0:15:18.200 --> 0:15:19.840 to provide on the other side of you know, a 0:15:19.880 --> 0:15:21.960 beautiful user experience to the good consumer. 0:15:22.120 --> 0:15:23.560 I'm speaking of sports references. 0:15:23.600 --> 0:15:25.240 You are probably one of the few tech CEOs that 0:15:25.240 --> 0:15:29.880 I've interviewed who has a background being an athlete of sourse. 0:15:30.120 --> 0:15:32.200 So, didn't I see. 0:15:32.040 --> 0:15:34.960 Something along the lines of you learned a very important 0:15:35.040 --> 0:15:40.080 lesson from totally screwing up a play in college football? 0:15:40.520 --> 0:15:41.120 Is that? Did I? 0:15:41.000 --> 0:15:43.400 I didn't make that up? That's real, right, Yeah. 0:15:43.400 --> 0:15:45.160 I think I was the only two sport athlete in 0:15:45.200 --> 0:15:48.320 the ACC at Boston College and played football and baseball, 0:15:48.360 --> 0:15:52.120 and yeah, I had an amazing, amazing experience. It's like, yeah, 0:15:52.160 --> 0:15:55.640 my my my very first punt of my very first 0:15:55.680 --> 0:15:58.360 game at BC, the kick returner was late coming on 0:15:58.400 --> 0:16:00.280 the field. I picked my head up in the snap 0:16:00.360 --> 0:16:03.320 hit me right in the chest. And so you know, 0:16:03.680 --> 0:16:06.440 as they say, you should only make some of these 0:16:06.480 --> 0:16:09.560 mistakes once in life. But yet I think a lot 0:16:09.600 --> 0:16:12.560 of the traits that you learn as an athlete, persistence 0:16:12.640 --> 0:16:17.000 and grit and hard work and delayed gratification and uh, 0:16:17.040 --> 0:16:20.360 you know, having a having a multi year plan have 0:16:20.400 --> 0:16:23.840 a Yeah, voted quite well in the transition into into 0:16:23.920 --> 0:16:24.400 this world. 0:16:38.240 --> 0:16:40.080 Can you give us an inside look at how a 0:16:40.120 --> 0:16:42.520 generative AI is reshaping fraud? 0:16:43.400 --> 0:16:46.560 Yeah, look, I think you know, we've been fortunate, as 0:16:46.600 --> 0:16:47.920 I was alluding to, to be a part of like 0:16:47.960 --> 0:16:54.640 a number of these these waves. But the fraud rings 0:16:55.680 --> 0:16:59.280 operate like a business. They have p and ls, they 0:16:59.320 --> 0:17:02.720 make technology investments, they look at their return and so 0:17:02.880 --> 0:17:07.320 as the cost of compute and the cost of AI. Right, 0:17:07.400 --> 0:17:10.200 certainly with even like open weight models, obviously many of 0:17:10.200 --> 0:17:13.159 the new Chinese models coming out, you have a tenth 0:17:13.160 --> 0:17:15.840 the twentieth to fiftieth of the price to be able 0:17:15.880 --> 0:17:19.040 to perpetuate and utilize some of these technologies. It's really 0:17:19.680 --> 0:17:25.760 a shift in terms of the economic incentives for fraudsters 0:17:25.800 --> 0:17:27.879 because it's been comes so much cheaper. So I think 0:17:27.920 --> 0:17:30.800 it's cheaper, you actually do more of it, right, And 0:17:30.840 --> 0:17:34.560 then a lot of this white collar digital crime does 0:17:34.560 --> 0:17:38.320 it end up getting prosecuted if you're selling heroin or 0:17:38.680 --> 0:17:41.880 cocaine or you know, obviously other stuff on the streets. 0:17:42.160 --> 0:17:45.240 You're actually seeing a lot of those criminals are moving 0:17:45.920 --> 0:17:48.800 into this type of like white collar digital crime. Right. 0:17:48.880 --> 0:17:52.280 You're seeing there's been good reports I think in Kingston 0:17:52.320 --> 0:17:54.719 and Jamaica a lot of the old call centers are 0:17:54.720 --> 0:17:57.560 now becoming scam centers, right, and you see a lot 0:17:57.600 --> 0:18:00.560 of this kind of conversion of like traditional stres crime 0:18:00.960 --> 0:18:02.840 is now moving to this type of crime because it's 0:18:02.840 --> 0:18:06.920 actually so easy. So like the tools have been made 0:18:06.920 --> 0:18:10.040 available to everyone. Right, you can obviously go and create 0:18:10.240 --> 0:18:12.960 many of these accounts for free. You don't even have 0:18:13.000 --> 0:18:15.560 to put in any amount of PII, and you have 0:18:15.600 --> 0:18:20.520 access to this supercomputer, right that can perpetuate the creation 0:18:20.600 --> 0:18:23.000 of synthetic identities. Right, I can say, hey, Lourie lives 0:18:23.000 --> 0:18:25.879 on the street, give me, give me you know a 0:18:26.000 --> 0:18:29.400 million fake identities that are all in this zip code, 0:18:29.800 --> 0:18:32.760 right make here's all these fake documents I have. And 0:18:32.840 --> 0:18:36.560 so the bearer the barrier to becoming a fraudster, right 0:18:36.760 --> 0:18:40.160 is no longer about capital, and in many cases it's 0:18:40.160 --> 0:18:43.840 not even about skill. And if you need the skills, 0:18:43.880 --> 0:18:45.600 you can go pay nine ninety nine on line in 0:18:45.720 --> 0:18:48.760 a YouTube video or Instagram or Telegram and they'll send 0:18:48.800 --> 0:18:51.399 you a book of how to go run you know, 0:18:51.480 --> 0:18:54.880 these scams or attacks, and so these like fraud as 0:18:54.920 --> 0:19:00.840 a service offerings and combination with AI and combination with 0:19:00.960 --> 0:19:05.560 the digitization of really everything in our life has created this, 0:19:05.720 --> 0:19:11.840 you know, effectively unlimited surface area of potential attacks. And 0:19:11.920 --> 0:19:15.640 so you know, as I said, frosters are just testing, testing, testing, testing, 0:19:15.720 --> 0:19:19.240 and it's an accelerating kind of rewriting of the rules 0:19:19.240 --> 0:19:24.120 of fraud because of the new technology capabilities. 0:19:24.240 --> 0:19:28.000 I just I really believed to some degree that way 0:19:28.040 --> 0:19:30.960 beyond the companies that you help, right, like these these 0:19:31.400 --> 0:19:35.600 multi billion dollar businesses that we all to some degree 0:19:35.640 --> 0:19:37.480 are going to be living fake lives on the Internet 0:19:37.520 --> 0:19:40.000 and we don't even realize it. And I came to 0:19:40.040 --> 0:19:42.880 this conclusion when I realized I had many fake lives 0:19:42.880 --> 0:19:45.119 on the Internet. And you know, you could say, well, 0:19:45.320 --> 0:19:47.600 you're real kind of a public figure, this or that. 0:19:47.800 --> 0:19:50.959 But as I started digging into my own face, I 0:19:50.960 --> 0:19:55.360 say fake lives. I found there was a Lori selling crypto, 0:19:55.880 --> 0:19:58.159 and then I connected that Lori selling crypto. It was 0:19:58.200 --> 0:20:01.600 a lot of my it was my identity, it was me. 0:20:01.800 --> 0:20:05.040 It was a whole fake profile that had the pictures 0:20:05.080 --> 0:20:07.239 of my wedding, this and that, plus a little bit 0:20:07.240 --> 0:20:10.320 of something else. And there, as I started digging, it 0:20:10.400 --> 0:20:13.359 was so far beyond fake Lori. There was fake this 0:20:13.440 --> 0:20:16.040 person and fake this person, all these fake people talking 0:20:16.040 --> 0:20:16.640 to each other. 0:20:16.920 --> 0:20:17.320 And then I. 0:20:17.280 --> 0:20:20.439 Started reverse imaging, you know, some of these images, and 0:20:20.520 --> 0:20:23.199 this person who was doing this is actually works at 0:20:23.200 --> 0:20:23.840 a police. 0:20:23.560 --> 0:20:24.639 Department in Ohio. 0:20:25.119 --> 0:20:27.840 And all of these normal people, these people who live 0:20:27.840 --> 0:20:31.120 in the real world, were part of this fake bubble, 0:20:31.160 --> 0:20:34.840 this fake life of fraud that they had no idea about. 0:20:35.240 --> 0:20:37.399 And so when I think about the future of AI 0:20:37.480 --> 0:20:39.359 and what you're just talking about, which is like the 0:20:39.359 --> 0:20:42.239 democratization of these tools that make it so easy at 0:20:42.280 --> 0:20:45.720 a time when our identities are out there for anyone, 0:20:45.880 --> 0:20:48.280 and now it's easy to do it. That's what I 0:20:48.359 --> 0:20:51.640 worry about, this idea of these fake lives and what 0:20:51.680 --> 0:20:53.480 will be you know, and the things that we don't 0:20:53.520 --> 0:20:55.960 even know about that us normal people out here don't 0:20:55.960 --> 0:20:56.639 even know about. 0:20:57.359 --> 0:20:59.840 Even the risks. I mean our team, there's like go 0:21:00.080 --> 0:21:03.399 and sourced tools, Laurie, like Kling, which is you know, 0:21:03.440 --> 0:21:05.720 there's been a lot of really good videos made on x. 0:21:06.280 --> 0:21:08.639 You know, my team was able to take a public 0:21:08.760 --> 0:21:13.879 image of me and with a free software in about 0:21:13.960 --> 0:21:19.239 twenty minutes go create me, replicating whatever it was they 0:21:19.280 --> 0:21:21.440 were doing. And so Laurie, if someone's trying to go 0:21:21.560 --> 0:21:23.960 create a fake crypto account with your image, you know, 0:21:24.040 --> 0:21:25.639 we're having to do a real time search. It's like, hey, 0:21:25.720 --> 0:21:27.919 is this an image that's a public image? 0:21:28.080 --> 0:21:28.200 You know. 0:21:28.280 --> 0:21:32.399 AI is just dramatically, you know, changed the cost and 0:21:32.520 --> 0:21:37.680 skill to commit fraud. There's obviously image and manipulation, there's 0:21:37.800 --> 0:21:42.359 voice clode manipulation. Yeah, it's it's it is really kind 0:21:42.359 --> 0:21:44.719 of changed the calculus in terms of you know, how 0:21:44.760 --> 0:21:46.679 do you trust how do you trust people on the Internet. 0:21:46.760 --> 0:21:49.280 And it's and it's hitting everything. It's hitting everything from 0:21:49.359 --> 0:21:53.960 the folks trying to implement age Assurance Right regulations, which 0:21:54.000 --> 0:21:56.600 has you know, been a sweeping change of the Internet 0:21:56.840 --> 0:21:59.280 obviously over the last twelve to eighteen months. But like 0:21:59.640 --> 0:22:02.600