Latest / Consumerpedia / 48 - How will Scammers Use AI?
Transcript
- 0:02Everyone's talking about artificial intelligence.
- 0:04This supercomputing technology has been around for decades, but
- 0:08the world changed in November with the introduction of
- 0:10ChatGPT. The A I app that lets anyone
- 0:13harness the power of this amazing technology.
- 0:17A I is being used to write term papers, create songs, and clone
- 0:20voices. And you can be sure criminals
- 0:22around the world are already looking for ways to use the
- 0:25power of a I to scam you. I'm Her Weisbaum, A contributing
- 0:29editor at checkbook.org. Welcome to
- 0:32consumerpedia@checkbook.org. We're the nonprofit that helps
- 0:36consumers select services, avoid trouble, and save money.
- 0:39Because we don't accept any advertising or take money from
- 0:42any business, we recommend you can rely on checkbook.org to be
- 0:46completely independent and objective.
- 0:48Now here's the host of Consumer Pedia America's consumer expert,
- 0:53the consumer man Herb Weisbaum. Fax machines, e-mail, text
- 0:57messages. The Internet criminals always
- 1:00find ways to use new technology to their benefit.
- 1:03So what will they do with the power of a I We have 3 experts
- 1:07ready to talk about this. Chester Wisneski is field chief
- 1:10technology officer for Applied Research at Sophos.
- 1:13That's the global digital security company.
- 1:16Chester is based in British Columbia.
- 1:18Hey, Herb. Good to talk to you again.
- 1:19Lori Cranor is the Director of the Cy Lab Security and Privacy
- 1:22Institute at Carnegie Mellon University.
- 1:24I first met Lori when she was the Chief Technologist at the
- 1:27Federal Trade Commission. Lori's joining us from
- 1:29Pittsburgh. Hello.
- 1:31And Brett Johnson, a former criminal who specialized in
- 1:34identity theft. At one time he was on the Secret
- 1:37Service 10 Most Wanted list. They called Brett the original
- 1:40Internet godfather. After serving about six years in
- 1:43prison, Brett decided to turn his skill set into fighting
- 1:47fraud. He's now a consultant, public
- 1:49speaker and podcast host. Brett lives in Alabama.
- 1:52Howdy, Herb. How are you?
- 1:54Good to talk to you again. So Brett, you told me that
- 1:57during your life of crime you stole about $7,000,000 and
- 2:01clearly you use technology to do that.
- 2:04Is artificial intelligence going to help fraudsters do a better
- 2:07job? Oh, it absolutely will.
- 2:09It already is. I mean we're already seeing a I
- 2:11being used and a couple of different services being offered
- 2:14in Worm, GPT and also fraud. G PT1 of them cost $60.00 a
- 2:18month. The other one costs $90.00 a
- 2:20month. One is concentrated on scam
- 2:22emails, phishing emails, writing codes, stuff like that.
- 2:26I would expect that as a I continues to improve.
- 2:30The ability of criminals will continue to increase as well.
- 2:33The scalability, things like that.
- 2:35I mean, they the sky is absolutely the limit as far as
- 2:38AI goes with benefiting cyber criminals.
- 2:41You think he would have been more effective at stealing money
- 2:43if you had had AI at your controls?
- 2:45I think I would have been able to scale it up better is the
- 2:48issue you figure now that talking about the growth of
- 2:51cyber crime, if cyber crime were a country today, it would have
- 2:55the third largest economy on the planet.
- 2:57All right, so it's already huge as it is, but you've got a lot
- 3:00of individuals that are engaged in cyber crime these days that
- 3:03are not really technically sophisticated.
- 3:05They don't know how to write code.
- 3:06They're not very good about coming up with phishing emails
- 3:08or anything else like that. Well, A I will absolutely lend
- 3:12to that. Not only will it make that
- 3:13easier for those unsophisticated criminals, but.
- 3:15That it will take a lot of the heavy lifting away from the more
- 3:20experienced criminals. You figure that it takes a lot
- 3:22of time to launch these scams or financial cyber crimes that
- 3:25you're trying to perpetrate. If you can hand off a lot of
- 3:28those functions to an A I type system and let that do it for
- 3:31you until it's ready to hand it over to a human attacker, that's
- 3:34going to be involved. It makes it much more scalable,
- 3:37much easier to commit, much more profitable at the end of the
- 3:39day. So Laurie, as a cyber security
- 3:42researcher and also someone who worked at the Federal Trade
- 3:45Commission, I would assume you agree with Brett that this is
- 3:47going to just make life easier for the criminals of the world.
- 3:50I think it will make life easier for the criminals, but it will
- 3:53also help the defenders who will be able to use AI to help detect
- 3:57these sorts of crimes. And we're going to talk about
- 3:59more about that later in the podcast because that's a very
- 4:01interesting topic. Chester, you actually
- 4:04demonstrated at a cyber conference how criminals might
- 4:08be able to use artificial intelligence.
- 4:10Can you tell us about that? Some of my colleagues presented
- 4:13at the Artificial Intelligence Village at the Defcon conference
- 4:16in Las Vegas, and what they were kind of showing is how you can
- 4:20use this technology to kind of create an entire scam ecosystem
- 4:24from front to back with all the supporting stuff, right?
- 4:27You can have a I write the website that you use for the
- 4:30attack. You can have a I.
- 4:32Generate fake LinkedIn profiles that make the people that
- 4:35allegedly run the business look legitimate.
- 4:37You can use AI to help you collect the credit card
- 4:40information for the fake ecommerce site that you had.
- 4:43AI right for you, right? The idea is you can kind of
- 4:46generate an entire scam environment without really
- 4:48having any skill or time invested to do so.
- 4:52In a world where we already struggle with what is
- 4:54authenticity, this is. Kind of not a great thing.
- 4:58I would agree with that. You had mentioned how does it
- 5:01help the good guys compared to how does it help the bad guys
- 5:04understand that the bad guys and I say we because I used to be
- 5:06one, we are not reeled in by existing systems or regulations,
- 5:12privacy concerns, management, things like that.
- 5:14So we've got an open playing field with which we can use some
- 5:17sort of AI type system, all right.
- 5:20And we can use it for any number of things.
- 5:21But the good guys, they're, they're having to kind of
- 5:23shoehorn in these AI systems into.
- 5:27Existing fraud controls that are in place, not only fraud
- 5:29controls, but privacy concerns, regulations, things like that.
- 5:33As such, the way that it benefits criminals is much more
- 5:36effective. So while the good guys are
- 5:38having some difficulty implementing a I systems, the
- 5:41bad guys really have an open playing field with which to do
- 5:44that, and I think they're going to be, at least at the
- 5:46beginning, much more successful than the good guys are.
- 5:48I totally disagree with that actually, because when I look at
- 5:51it, the the reality is all the data scientists that are
- 5:53required to build the sophisticated tools like.
- 5:57ChatGPT or any of the other ones requires millions of dollars of
- 6:02data scientists, not to mention access to huge quantities and
- 6:05ability to collect and process data in order to train a lot of
- 6:09these, especially the generative AI models like ChatGPT.
- 6:12And the reality is the criminals don't have access.
- 6:14They have access to the technology, they don't have
- 6:16access to the knowledge and they don't have access to the
- 6:18expertise. And they generally will do the
- 6:21cheapest, simplest thing to commit a crime, as opposed to
- 6:24overarchitecting some complicated engineering in order
- 6:27to develop their own AI models. So what we see today available
- 6:30to criminals simply lets them accelerate and economize what
- 6:35they're already doing. It doesn't make better attacks.
- 6:37It just makes attacks more voluminous or gives more
- 6:41sophistication to people who may lack it.
- 6:43It allows more people to gain entry to the ability to do
- 6:47crimes, but the crimes aren't anymore sophisticated.
- 6:49Than the existing crimes. It's just sort of automating it,
- 6:52a colleague said to me once. It doesn't sharpen the spear, it
- 6:56broadens the head like we're going to see more attacks, but
- 6:59they're not going to be any more sophisticated than what we
- 7:02already see. Whereas the good guys with the
- 7:04data scientists and access to all this information.
- 7:06Are able to use this to empower less skilled individuals to be
- 7:10more able to detect crime using these technologies in my
- 7:14opinion. I respect that opinion.
- 7:16I can't really disagree with that opinion.
- 7:17But I would say that as the victim count increases, it kind
- 7:21of evens out on both sides of the fence with that, right?
- 7:23I mean, sure, you couldn't educate people on the good guys
- 7:25side as much as you want to, but if you broaden that spear and
- 7:28hit even more victims at the end of the day.
- 7:30The good guys can only cope with so many victims at a time.
- 7:33When it's probably important to differentiate when I'm my
- 7:36headspace is in protecting enterprises, not protecting
- 7:39individuals. So when I look at professional
- 7:41IT people with access to these tools, they're going to be more
- 7:44effective than they were before. When we're talking about average
- 7:47people who are getting scammed day in, day out, they're
- 7:49helpless. They don't have access, they
- 7:50don't have any filter in front of them.
- 7:52And so I think it will disproportionately affect normal
- 7:55individuals than it will affect businesses.
- 7:57Laurie, since you brought up this topic, how do you think
- 7:59good guys are going to use AI to help fight the bad guys?
- 8:02I think there are a few things that good guys can do.
- 8:05Part of it is having the good guys also scale.
- 8:08Just as the criminals are going to scale, so can the good guys.
- 8:11We have a shortage of cybersecurity workers right now,
- 8:15but if we can use AI to scale our ability to perform tasks,
- 8:21that's really going to help. I think we may also be able to
- 8:25use AI so that we can put in a home computer user.
- 8:31Web browser or phone or whatever, AI that's going to
- 8:35help them detect scams, You know, that's not something that
- 8:38we do in a very sophisticated way today.
- 8:41We're basically just doing like keyword filtering.
- 8:43But it seems like we should be able to to scale that with AI
- 8:47into something much more sophisticated.
- 8:49It won't happen tomorrow, but it should be able to happen soon.
- 8:52Laurie, Federal Trade Commission Other consumer advocates always
- 8:55say there are things you can do now to spot the frauds.
- 8:58For instance, bad syntax, bad grammar with testimonials or
- 9:02reviews if they use the same words over and over again, etc.
- 9:05But that's going to change with AI, because the AI can write
- 9:08perfect English or perfect whatever language the scammers
- 9:11want to use, and can also create an individual review or
- 9:14testimonial, every single one. It doesn't have to cut and
- 9:16paste. Am I right with that?
- 9:18Yeah, there's a lot of different advice that's been given to
- 9:21consumers about how to detect scams.
- 9:24And I was actually never super excited about the advice that
- 9:28said, oh, you know, scammers just.
- 9:30You know, have typos and spelling errors and look
- 9:33professional because even without AI, a scammer can just
- 9:36make a perfect copy of an existing website or existing
- 9:40mail from a vendor and send it to you, right?
- 9:42And you know, just change the URL.
- 9:44So they were already doing that without AI.
- 9:47So in the In the work that I did on training consumers, that was
- 9:51never what I trained them to do. What about better fake reviews
- 9:54and fake testimonials? I'm not sure how much consumers
- 9:58really rely on that. It's actually very.
- 10:00Easy for people to fall for scams.
- 10:02They don't have to be very sophisticated for people to fall
- 10:05for them. People just act very quickly.
- 10:08Somebody who's taking the time to really read through the
- 10:12reviews and testimonials is already like thinking a lot more
- 10:16than most people who are falling for the scams like very quickly.
- 10:20The scammers haven't had to do sophisticated things because
- 10:23they can scam people without being sophisticated.
- 10:26So how do you navigate a world where everything can be faked?
- 10:30Are there ways for you to protect yourself?
- 10:32We're going to talk about that next.
- 10:33I'm Herb Weisbaum, the consumer man, and this is Consumer Pedia,
- 10:37powered by checkbook.org. If you like what you hear, we
- 10:41hope you'll consider being a consumer pedia supporter by
- 10:44using the link at the bottom of the show notes to make a small
- 10:47contribution each month. This is Consumer Pedia.
- 10:52We're back. Lori.
- 10:53Are we already at the point where it's hard to trust what we
- 10:56hear and see? Yeah, I think we are.
- 11:01I think, you know, we have fairly primitive ways of faking
- 11:06things that people seem to fall for.
- 11:09So yeah, it's hard to know what's real, but it's going to
- 11:12get easier with AI and maybe you can attack bigger surfaces,
- 11:15meaning instead of going after you and me, you could go after
- 11:18some corporation and get a whole lot of money all at once.
- 11:21Yeah, I think it's only gonna get worse.
- 11:24There are various proposals to kind of watermark things to try
- 11:29to address this problem, but I haven't heard a proposal that I
- 11:34think is actually workable because if you watermark all of
- 11:39the real stuff. And then the bad guys go and
- 11:43remove the watermark. Then they'll try to make it look
- 11:45like the real stuff is actually fake, and then the fake stuff is
- 11:48real. And I I just, I'm not entirely
- 11:51sure how we tackle this, Brett. I mean, it's a problem.
- 11:55I, you know, you know, I've said for years that the perception of
- 11:58truth is much more important than the truth itself, that
- 12:00facts don't matter. The only thing that matters is
- 12:03what you can convince someone up.
- 12:05O you add Dee fakes into that equation.
- 12:06Now if you've got real time Dee fakes that are being assisted by
- 12:10AI where I can have a real time conversation with someone on
- 12:14camera or over the phone, how effective is that?
- 12:17So that's the type of problems that we've got now we can no
- 12:20longer, it's difficult to trust the news.
- 12:22It's difficult to trust what we see with our eyes, what we hear
- 12:24on the radio or anything else like that.
- 12:25And it's going to continue to get worse.
- 12:28Chester I think we sometimes approach these problems the
- 12:31wrong way, right? We can't prove a negative, so of
- 12:34course anything we see could be a fake when we're just looking
- 12:37at it visually. But we are making some progress
- 12:40in being able to prove that real things are real.
- 12:42So, for example, Sony's digital cameras now digitally sign every
- 12:46image that they take. Now that photo, when it's put on
- 12:50the cover of Time magazine or when it's put on a website,
- 12:53might have been brightened or altered to make the color
- 12:56saturation better. Or it might have been cropped,
- 12:59in which case that signature doesn't mean anything.
- 13:01But it does mean a journalist can prove I actually took this
- 13:04photo, the guy in front of the tank in Tiananmen Square.
- 13:07I can prove that was a real photo if I've taken it with a
- 13:10digital camera that does this. So I think we can improve how we
- 13:14at least allow journalism to maintain and demonstrate their
- 13:18authenticity and their integrity and then assume everything that
- 13:21hasn't BeenVerified, Well, it's probably fake.
- 13:24And I mean, I I think unfortunately, we're getting so
- 13:27pessimistic. A lot of people are.
- 13:28My my mother now is so pessimistic that when she gets
- 13:31real warnings about security stuff from her financial
- 13:34institutions, she throws them away, assuming their scams.
- 13:36So there is a real danger here like of people getting too
- 13:39cynical too so. This is a really weird tightrope
- 13:42we're going to be walking as we go forward and this may be
- 13:44critical, this watermarking as we go through the campaign
- 13:48season for election of President United States.
- 13:49There could be all kinds of deep fakes created by who knows who
- 13:52have. That's what scares me.
- 13:54Well, I mean, we've already seen that with the Ukraine war,
- 13:56right? We saw a deep fake of Zelensky
- 13:58saying they had surrendered at one point.
- 14:00So these things are absolutely in play and in use.
- 14:02And the Pope in a puffy jacket, right?
- 14:05So the news is filled with stories about how scammers are
- 14:07using AI to fake voices so they can sample a voice of your child
- 14:11or your grandchild, maybe from social media posts, and then
- 14:14they call you and the computer generated voice sounds like that
- 14:17child. It seems like a lot of extra
- 14:20work for a scam that's been working really successfully for
- 14:22the last 25 or 30 years or more. I'd like to get all three of you
- 14:26to share with me if you think this is really happening and
- 14:29whether you think we're going to see more of this, if it's really
- 14:31worth the effort. So I'm going to start with you,
- 14:33Chester. Well, I don't think.
- 14:35It's happening the way it's being portrayed as happening,
- 14:37but I also don't doubt that it could be happening in the future
- 14:41when it's easier to do. I think today the resources
- 14:43involved and the quality of the tools that are available to the
- 14:47general public is too much effort for too little reward for
- 14:51criminals to bother. And I think Brett, he very
- 14:53specifically said in real time, right, that it's not in real
- 14:56time right now, which means all you can do is maybe pre generate
- 15:00a short clip and then take over the traditional scam.
- 15:03So maybe, you know that's easy enough that it might be
- 15:06happening. But I suspect that it's time for
- 15:08FBI to justify some budget and they're just overhyping
- 15:12something that I think it's being done the oldfashioned way
- 15:14at the moment. But I don't think we should
- 15:15ignore this because I think it's only a matter of a very short
- 15:18period of time before this is. Simple enough and will be in
- 15:21real time that we should be planning for it now, even if
- 15:24it's not happening today. And Chester, you told me about
- 15:26someone you spoke to in one of these situations that we
- 15:28reported in the news. Could you share that story?
- 15:30Well, yeah. This was an older woman who had
- 15:34gotten one of these calls saying that her grandchild had been
- 15:37kidnapped and then she needed to send some ransom money right
- 15:40away or they were gonna harm her.
- 15:42And the evidence that she who was quoted in a story as being
- 15:45one of the victims who'd had the AI of her granddaughter, all it
- 15:49was, was a woman. Scream.
- 15:50And then a man came on the phone and demanded the ransom.
- 15:53So there was no ability to determine whether it even
- 15:56sounded like her granddaughter, let alone be an A I Deep fake.
- 16:00Brett, I agree with everything Chester says.
- 16:02You've got a lot of security companies that are out there
- 16:04right now that are hyping the A I fraud angle, all right?
- 16:07And the truth of the matter is, is this it's not really advanced
- 16:11enough for fraudsters to be using it.
- 16:13And Mass, All right. Especially if we're talking
- 16:15about deep fakes and especially real time.
- 16:17Real time is not even out there right now.
- 16:19Now it's coming. It will absolutely get there.
- 16:21But it's way too difficult right now for fraudsters to do that.
- 16:23The if you're looking at a scam like that.
- 16:26The purpose of the scamster, the goal is to get the potential
- 16:29victim to act out of emotion, not out of reason or logic.
- 16:33So if you call 2:00 in the morning, you're acting like that
- 16:36grandchild that you barely get to speak to three or four times
- 16:39a year, and you're trying to scare them to give that knee
- 16:41jerk reaction. You don't need a deep fake voice
- 16:45to make that successful. It's all about causing that
- 16:49potential victim to put aside logic, put aside reason.
- 16:52Now later on as a I continues to improve when we get to that.
- 16:56That real time deep fake ability, I think that we'll see
- 17:00this trickle down aspect of cybercrime that's kind of
- 17:02inherent throughout the entire cybercrime environment.
- 17:05So you take one specific type of technique, it starts in one
- 17:08vertical and as it becomes more ubiquitous, it bleeds through
- 17:11all the other fraud verticals that are out there.
- 17:14That's what we end up happening, but we're nowhere near that
- 17:16right now. And having talked to you Brett,
- 17:18criminals like quick and easy, not a lot of effort.
- 17:21And you know, I've covered this for 25 years and having a voice
- 17:24in the background off phone going.
- 17:26Oh, true. And it's like, is that you,
- 17:28Charlie? You're school or true?
- 17:30That's work to steal a whole lot of money from people.
- 17:33They don't have to go to the trouble trying to sample your
- 17:35voice and create something in real time.
- 17:37No, absolutely. And you got to understand too
- 17:38that you know there are three motivations for why cyber crime
- 17:41happens. You've got status, cash,
- 17:43ideology, 98% plus of attacks are cash based attacks.
- 17:47That's lowest hanging fruit. That's a criminal that's looking
- 17:50for the easiest access that gives the largest return on that
- 17:53criminal access. So they're not looking to spend
- 17:56a lot of money, they're not looking to take a lot of time or
- 17:58effort. They're looking for the easiest
- 18:00access. That gives them the most profit
- 18:01on that access. Laurie, could this AI, this
- 18:04creation of someone else's voice, could that be used by the
- 18:07criminals who are going for the long take the deep haul So you
- 18:10can have the boss scam where the boss is overseas and they know
- 18:14the boss is overseas and it's contacting the assistant back in
- 18:17the US and saying I'm overseas, I'm trapped.
- 18:19I need you to wire me $10,000. Would that be worth the effort?
- 18:22Is that how this might be used? I think it's definitely more.
- 18:25Likely to be used on these high value targets.
- 18:28Another point I'll make about the low value targets is I think
- 18:31in a lot of these they don't even know the name of the
- 18:33grandchild and they don't have access to a sample of the voice.
- 18:38Like in order to train the AI you need to find out who the
- 18:40grandchild is, get a sample of their voice and then you can
- 18:43train the AI to do that. I don't think they even go that
- 18:46far. If you have a high value target
- 18:49then you have a situation where it's worth taking the effort to
- 18:52find. You know the name of the boss
- 18:54and. If it's a high profile person,
- 18:56their voice is readily available to get a sample of and then I
- 19:01think there's more of a chance that it would be worth the
- 19:03effort for the bad guys to actually go through and do that.
- 19:06You never underestimate how lazy criminals are.
- 19:11I would not call us lazy. I would call us.
- 19:13We look for the avenue of least resistance.
- 19:16Very well put, by the way. It sure does make a good
- 19:19broadcast story when a correspondent can go to a
- 19:22digital security company and say, find my voice on line and
- 19:25then make me say all these kinds of things in here.
- 19:27Or they're like, oh, my God, the world's coming to an end.
- 19:30But the pretend criminal, the digital security company knows
- 19:33whose voice are looking for and what to do.
- 19:35And as as Laurie, as you pointed out, a lot of times they have no
- 19:38idea. They're just making these calls.
- 19:40And it's the victim who gives them the name.
- 19:41The victims, Johnny, is that you?
- 19:43Susie, Is that you? You know, they don't even know
- 19:45what voice to sample. Yeah.
- 19:47I agree. All right.
- 19:50When we come back, we're going to talk about some possible ways
- 19:53that you might be able to protect yourself as we boldly
- 19:56move into this world of artificial intelligence.
- 19:58I'm Herb Weissbaum, and you're listening to Consumerpedia,
- 20:01powered by checkbook.org. Consumerpedia Fast Facts
- 20:06Artificial intelligence is already a part of Daily life.
- 20:10Smart speakers and digital assistants like Siri are powered
- 20:13by a I. So are the map programs we all
- 20:16rely on. A I helps predict the weather,
- 20:19diagnose diseases, create new drugs.
- 20:22It's also being used to develop selfdriving cars.
- 20:25Surveys show most business executives are sold on a I they
- 20:28believe it can drive growth and cut costs.
- 20:31By one estimate, the global market for a I will top $190
- 20:35billion by 2025. All right, so how do we protect
- 20:41ourselves in this brave new world of artificial intelligence
- 20:44that is going to empower all of us, including the criminals in
- 20:47the world? I like all three of you.
- 20:49To answer this question, is AI simply going to make the
- 20:52criminals more effective at the current scams, which means we
- 20:55protect ourselves the same way we're doing now?
- 20:57Or is it going to allow them to create new scams?
- 21:00So it's just old scams, better effectiveness or they're going
- 21:03to be new scams. You think that'll be created by
- 21:05this? And Chester, we're going to
- 21:06start with you. Well.
- 21:07I don't have the creativity that.
- 21:10Some of these criminals do when I see some of the scams coming
- 21:12down the Pike sometimes. So I can't really say that they
- 21:15won't come up with new ways, but certainly from what I see
- 21:18available today with the tools that are accessible and usable
- 21:23enough that criminals would bother using them.
- 21:25I think it's just going to be a higher quantity of slightly
- 21:30more, Let's call them high definition scams, right?
- 21:32Like when foreign criminals were writing scams, there often were
- 21:36grammatical mistakes, there were spelling mistakes.
- 21:38And I don't think it was ever a great technique for deciding
- 21:41something was a scam, but it was.
- 21:42It was another tell. And that tell is now officially
- 21:45gone. It was already largely gone, and
- 21:47now it's completely gone. If you get a text based scam,
- 21:50whether it's on your WhatsApp or you're an SMS message or an
- 21:53e-mail, you can pretty much count on it being.
- 21:55Perfect American English, Canadian English, British
- 21:58English, whatever it needs to be because these tools have made
- 22:01that ubiquitously accessible to create scams.
- 22:04And so I think today we're mostly just going to see higher
- 22:07volumes of high definition scams.
- 22:10Right. So I don't like the idea of new
- 22:13scams and I'll tell you why because if you look at attacks
- 22:15that that attack landscape that's out there, 90% plus of
- 22:20every single attack uses a known exploit.
- 22:23It's not unknown vulnerabilities, it's not
- 22:25zeroday attacks. It's the stuff we've typically
- 22:27been told about for years that we're not doing anything about
- 22:30that creates that threat landscape.
- 22:32So you add in an A I type system that's better able to navigate
- 22:36that threat landscape that's already available.
- 22:38And I think at that point you start to.
- 22:40Ping new areas that are out and and maybe new ideas are coming
- 22:44to fruition from that. You take for example the
- 22:47pandemic. Government fraud was already a
- 22:50thing before the pandemic, but most of the cyber criminals that
- 22:54were out there, we're not really hitting the government sector.
- 22:58The pandemic changed that. Now you've got one of the
- 23:00primary focuses of most cyber criminals today is looking at
- 23:04government benefit fraud. That's simply because that
- 23:07opened up a new view or an insight that criminals could.
- 23:10Look at and try to exploit at that point in time.
- 23:13I think that's what AI does at the end of the day is it allows
- 23:16that that viewing angle to increase for cyber criminals.
- 23:20Until we get to the point that we're plugging those 90% of
- 23:24exploits that we already know about, that we're not doing
- 23:26anything about, I think that AI will continue to eat things
- 23:30alive. And it's like, you know, you
- 23:32call it Bitcoin, I call it Ponzi.
- 23:34You know, like it's there. It's a different version of the
- 23:36same thing, right? It's just it's now digital.
- 23:38Laurie, do the same rules. Apply as the how we protect
- 23:42ourselves. All the rules that consumer
- 23:43advocates and federal regulators have shared with us all these
- 23:46many years is that it's the same scam, just with maybe a little
- 23:50bit better Polish to it. I think for now, yeah, I think a
- 23:54lot of the same rules apply, and I agree that it's likely we're
- 23:58going to just see the old scams on steroids.
- 24:01I do have some concerns that we may see some new ways of
- 24:06composing scams that people hadn't thought of before.
- 24:10That basically the AI can run through all the possible
- 24:13combinations and maybe hit on some combinations that we just
- 24:16hadn't thought of that turned out to be like really effective.
- 24:19That may be something that we see.
- 24:21But yeah, as far as protecting yourselves, I think the biggest
- 24:25advice is to be skeptical and, you know, think before you click
- 24:29and think about is something too good to be true?
- 24:32Does something makes sense? Does it all add up?
- 24:34Like that advice still holds even with AI.
- 24:38Yeah, Chester, you have an example of how AI.
- 24:40Could be used to make criminals lives easier, the ones that
- 24:43commit the romance. Scams, sure.
- 24:45I mean, we're seeing lots of these interactive scams, often
- 24:49related to romance or sometimes cryptocurrency or other things
- 24:53like that, and it's quite a human intensive thing.
- 24:56The way it works today is you're getting a message over text
- 24:59message or WhatsApp message typically, maybe Facebook
- 25:02Messenger, this type of thing. And they're interacting with you
- 25:06to kind of build some rapport before they go for the kill to
- 25:09get some money from you. And that is quite human
- 25:12intensive. I mean a couple of our
- 25:13researchers have done a few publications on this recently.
- 25:17And like often when they, you know, reach back out to the
- 25:20romance scammer, they might not get a response for a couple
- 25:22hours because there's a human that had to, you know, reply to
- 25:26them to keep the scam going. And certainly with the automated
- 25:29nature of artificial intelligence and its ability to
- 25:32hold a conversation. Basically, if you've played
- 25:34with. Tools like the free ChatGPT on
- 25:37Open A I's website. It's a pretty plausible human
- 25:40being, so you could imagine being able to shorten the time
- 25:44scale and the human effort. And this is goes back to things
- 25:46I think we've talked about throughout this interview, where
- 25:49we're really accelerating, right?
- 25:50If there's humans behind these scams today, they can only maybe
- 25:53manage 50 victims at a time that they're interacting with.
- 25:56But if you can suddenly tie on a I into getting somebody 80% of
- 26:01the way through the scam. Then, you know, one human
- 26:04handler now can handle 1000 or 5000 victims at a time.
- 26:08And that of course just increases the scalability of
- 26:11these attacks. Hey, I'm Brett.
- 26:21Now that's that's already there and and this idea of using a
- 26:24good password you've got 80% of the population uses the same or
- 26:27similar passwords across logins on different websites.
- 26:12Will AI make it easier for criminals, cyber criminals to
- 26:16hack passwords or they already have the sophisticated
- 26:19algorithms they need? Assuming you're using a good.
- 26:31So it's it's not really. I don't think AI is going to
- 26:34substantially affect that particular dynamic of
- 26:38cybercrime. You know, we need to and we're
- 26:41moving over to pass keys now, things like that as well.
- 26:43But the issue is, is that cyber criminals have already
- 26:46anticipated things like that. You know, you can say it's a
- 26:49game of cat and mouse, but it's a game of either being reactive
- 26:52or proactive. And unfortunately, the good guys
- 26:55tend to be more reactive than proactive.
- 26:57So we're just about out of time. Let me finish by going around
- 27:00the horn and leave us with one tip, one thing you want us to
- 27:03know to help protect ourselves before we leave.
- 27:06And Laurie, I'm going to start with you.
- 27:07Yeah, I think people just need to think before they react.
- 27:12Think before you click, before you text, Definitely before you
- 27:16send money. And be skeptical of what you
- 27:18read and think about. Is this plausible in this
- 27:21situation? And don't.
- 27:23Rush to give away personal information, passwords, money,
- 27:28bank account numbers, anything like that.
- 27:30Chester. I'm going to focus on the
- 27:31impersonation side of this, because I think it's one of the
- 27:34more interesting areas where people can be empowered to
- 27:37protect themselves better. Whether they're impersonating
- 27:39your boss, asking you to go buy gift cards, whether they're
- 27:42impersonating the kidnapped granddaughter, or whatever
- 27:46particular thing it is. When it's related to
- 27:48impersonation, you have two tools in your tool bag.
- 27:50One challenge the person who's claiming to.
- 27:53To be who they say they are with something that is not publicly
- 27:56available that a criminal would know.
- 27:58What did you send me for my birthday to the granddaughter
- 28:00you know last year? Or maybe if it's the boss with
- 28:03the gift cards, It's like perhaps you mention an internal
- 28:07project and and ask if it's related to that.
- 28:09Or you know something that is not easy for a criminal to know,
- 28:13the other thing you can do. Is use a second method of
- 28:16communication to verify it. If you really believe it might
- 28:18be real and you get an e-mail or a text from the boss, call the
- 28:22boss on the phone and say, hey, why do you need these gift cards
- 28:25or if it's the granddaughter calling you.
- 28:27Maybe you text her or some Facebook Messenger or or
- 28:30something, you know, FaceTime on your iPhone to confirm that it
- 28:33really is the person who says it is, and and those are things
- 28:36that the criminals really can't bypass.
- 28:38Create information so slow down and don't trust and verify.
- 28:41Absolutely. And finally, Brett.
- 28:43I mean, I like all the answers so far.
- 28:45What I would say is, is understand what a criminal is
- 28:47looking for. We look for information, access
- 28:50data or cash. If you get any type of
- 28:53unsolicited engagement that asks for any of that, you need to be
- 28:57cautious. Immediately.
- 28:58More importantly, you need to develop this idea of situational
- 29:01awareness in the real world. In the physical world, we are
- 29:04very situationally where we know when we take a wrong turn and
- 29:08end up in a bad neighborhood, or if something's going wrong in a
- 29:11store, our our hackles raise. But for some reason, that
- 29:14situational awareness is not very good at translating to an
- 29:17online environment. We need to understand that there
- 29:20are predators in these online waters.
- 29:22If we can get to that, if we can trust but verify and just
- 29:25understand that we need to protect our.
- 29:27Ourselves and always be vigilant and develop that situational
- 29:30awareness. All these other things will come
- 29:32with that. That's my firm belief.
- 29:34Laurie, Brett, Chester, thank you so much for sharing your
- 29:37expertise with us. Thank you.
- 29:38Thanks. Thank you.
- 29:40Well, that's it for this edition of Consumer Pedia.
- 29:42We hope you'll rate this episode and follow us on Apple Podcast,
- 29:45Spotify, or wherever you listen to your podcast.
- 29:49Remember, we release new episodes every other Thursday.
- 29:53Another way you can support this show is to follow us on Consumer
- 29:55Pedia on Facebook and Instagram, and at my Consumer Pedia on
- 29:59Twitter. I'm her Weisbaum.
- 30:02Thanks for listening. Consumer Pedia is a public
- 30:05service of checkbook.org. We're a unique nonprofit that
- 30:08helps you save money and make smarter choices.
- 30:11You can count on Checkbook to help you find the best services
- 30:14and avoid the worst, with local ratings that are accurate and
- 30:17unbiased. If you live in or around these
- 30:20seven cities and haven't joined Checkbook yet, check us out
- 30:23Boston, Chicago, Philadelphia, Seattle, San Francisco,
- 30:28Minneapolis, Saint Paul, and Washington DC To get your free
- 30:3230 day subscription, go to checkbook.org/consumer Pedia.
- 30:36If you like what you've heard, we hope you'll become a
- 30:38supporter by using the link at the bottom of the show.
- 30:41Notes to make a small contribution each month.
- 30:43Consumer Pedia Empowering consumers to save money and make
- 30:47smarter choices.