Latest / Consumerpedia / 88 - Cyberthieves Weaponize AI to Perfect Online Scams
Transcript
- 0:02Artificial intelligence is now. A part of our everyday.
- 0:05Lives often. In ways we barely.
- 0:06Notice AI powers the voice assistants that understand and
- 0:10respond to our commands. It guides navigation apps,
- 0:13predicting traffic and suggesting faster routes.
- 0:16It curates social media feeds, recommends what to watch, and
- 0:20even influences what we buy. But as AI technology advances,
- 0:24it's becoming harder than ever to tell what's real and what's
- 0:27Fake. I'm Herb.
- 0:29Weisbaum, the Consumer Man, a contributing editor at
- 0:31checkbook.org. Welcome to Consumerpedia at
- 0:35checkbook.org. We're the nonprofit that helps
- 0:37consumers select services, avoid trouble, and save money.
- 0:41Because we don't accept any advertising or take money from
- 0:44any business, we recommend you can rely on checkbook.org to be
- 0:48completely independent and objective.
- 0:50Now here's the host of Consumerpedia America's consumer
- 0:54expert, the consumer man Herb Weisspaum.
- 0:57Criminals here in the US and around the globe are weaponizing
- 1:00AI to create scams that are more convincing and more.
- 1:03Harmful than ever. AI is being used to create bogus
- 1:06celebrity endorsements, fake images, and even write books by
- 1:10made-up authors who don't exist. In this episode, we'll explain
- 1:13how fraudsters are using AI to supercharge their deception and
- 1:17share some tips on how you can protect yourself.
- 1:20We have two experts in the field of AI generated fakes joining us
- 1:23today. Colina Colti is a senior
- 1:25researcher and trainer at Bellingcat, a nonprofit
- 1:28investigative journalism group in the Netherlands.
- 1:30I understand you're based here where I am in Washington state,
- 1:33yes. I am and we get to both enjoy
- 1:35the lovely sunny days when we do get to have them.
- 1:38And Ben Winters, director of AI and privacy at the nonprofit
- 1:41Consumer Federation of America, which is located in the other
- 1:44Washington, the District of Columbia.
- 1:46Hi, Ben. Hello.
- 1:47Hello. So less.
- 1:48Than two years ago, October 2023, we recorded a Consumer
- 1:51Pedia podcast episode #48 that asked the question how will
- 1:55scammers use? AI and it was.
- 1:57The conclusion? Of my expert guess that.
- 1:59Fraudsters were already using this supercomputing power to
- 2:02scam people to create realistic looking e-mail and text messages
- 2:05and write malicious. Code so the question.
- 2:07For me, for both of you. Is.
- 2:09How important is AI to fraudsters right now?
- 2:11It's come a long way in the last year and a half.
- 2:14Kalina, let's start with you. I think it's incredibly powerful
- 2:17tool. Like if I wanted to switch
- 2:18careers, become a fraudster or a scammer, I would 100% be using a
- 2:23variety of AI tools in my repertoire.
- 2:25It's made everything, I think, so much, much easier for them
- 2:28and being able to scam people in a variety of different scams,
- 2:31it's made the ability for us to catch them, I think a lot
- 2:34harder. Ben.
- 2:35Yeah, I think that what it mainly does is amplify any of
- 2:39the bad behaviors that we see from scammers and have seen for
- 2:42years. It makes it easier, harder to
- 2:44detect and a little bit more personalized.
- 2:47So I think especially like some of the scariest, most damaging
- 2:51types of scams that create urgency or make it super
- 2:54personal about someone's family or life, that is easier to do
- 2:58using a lot of the commercially available AI tools we've seen in
- 3:00the last few years. So clearly AI is helping improve
- 3:03the scams that have been around forever.
- 3:05Is it enabling criminals to create new cons?
- 3:07Do they come up with anything new thanks to AI yet?
- 3:09You know, certainly the sort of live deepfake video scams where
- 3:14you could impersonate someone while you are on a Zoom or Teams
- 3:20with them, that creates this level of what consumers may
- 3:25traditionally have thought. You know, if I'm able to get on
- 3:27the phone, the person, if I'm able to get on a video chat with
- 3:29someone, I can trust that the voice or face I'm seeing is
- 3:34actually that person. And it's not just some machine.
- 3:36But now there are programs that enable that sort of live video
- 3:41and live audio to talk to people.
- 3:44And so it creates that level of intimacy, for lack of a better
- 3:48word, that would have people's guard down in ways that they
- 3:52hadn't been able to before. Let's look at some of the ways.
- 3:54AI can be used. Ben, in your report you suggest
- 3:58that AI will be used by criminals to supercharge common
- 4:00scam such as the tech scam, the tech repair scam.
- 4:03How's AI going to help make that con more effective?
- 4:05The increase of the use of AI agents more broadly in the
- 4:09economy are going to get people's guards down to not
- 4:12worry about the very concept of screen sharing, which is one of
- 4:15the main ways in which these tech support scams get your
- 4:18information. I think another way is that they
- 4:20are able to use audio and messaging services that don't
- 4:25have the common errors that might tip people off, whether
- 4:28that be language errors, grammar errors, the way e-mail is
- 4:33formatted. So the tech support scams are
- 4:35just like a lot of those other scams.
- 4:37It increases the sort of possibility of the content that
- 4:40you are giving to the person you're trying to scam.
- 4:43Kalina, you looked at how AI is being used to sell products
- 4:45online by creating bogus images. Tell me about the coffee mugs
- 4:49that looked awesome in the photos, but not exactly what the
- 4:52customers were expecting. I am mildly obsessed with this
- 4:56coffee mug phenomenon. I first started seeing these
- 5:00mugs online. People were posting these images
- 5:02of these really fantastical looking mugs that look like they
- 5:06were made out of crystals or minerals, really beautiful, in a
- 5:09variety of colors, variety of prices, and that people then
- 5:13shared the images of what they actually received.
- 5:16After they got the package, the mail.
- 5:18It did not look anything like those images.
- 5:20Nothing like those images. Nothing like those images.
- 5:24I mean, it was I guess technically a mug or some sort
- 5:28of like drinking vessel, but I would not even guarantee that it
- 5:31would be safe to drink from. This was sort of highlighting a
- 5:34huge issue that we particularly saw during the holiday season as
- 5:37a lot of people are shopping for gifts online.
- 5:40So from mugs to Christmas ornaments.
- 5:42And in case of my father, he ordered a book that was
- 5:44completely AI generated. And I think we're in this like
- 5:47really interesting point in time when we think about AI
- 5:50generation, images and videos and text and relationship to
- 5:55commerce to how we purchase items online.
- 5:57There's a few different things that are happening here from
- 6:00one, I, I don't think everyone has sort of the base of the
- 6:02ability to look at an image and sort of immediately say, hey, I
- 6:05think this might be AI. There's something that seems off
- 6:08with this image. I think we still take a lot of
- 6:09things at face value, particularly when we're shopping
- 6:12online. And granted, for a long time,
- 6:14people can might buy something online and not receive the
- 6:16product that they get. And oftentimes this is maybe
- 6:18using someone else's image or someone else's product and then
- 6:22shipping out a lesser quality of the item.
- 6:24But we're in this phrase where you can now create a whole new
- 6:27image, something, an item that didn't exist before, like mugs
- 6:30made out of crystals and minerals to be able to share
- 6:34something like, oh, there's nothing else like this out
- 6:36there. This is a brand new type of
- 6:37item. It looks amazing.
- 6:38Let me purchase it. And really that I cannot even
- 6:41exist in reality once you look at it, once you really start
- 6:43looking at the details. And so I think we have this big
- 6:45digital literacy component with that with being able to identify
- 6:48what an image is possibly AI generated.
- 6:51And I worry for even in the future because AI image
- 6:53generators are only going to get better or more sophisticated.
- 6:56Or maybe people are going to be better at maybe only use AI
- 6:59editing and edit a portion of the image.
- 7:01Make the image more trustworthy. So let's talk about that
- 7:03cookbook that your father got. There are a lot of books now,
- 7:06cookbooks and diet books and who knows what else that are being
- 7:09written by AI. You posted a screenshot of one
- 7:11on Amazon for the Renal Diet Cookbook for Beginners written
- 7:15by Ethan New Life. Tell us what you found when you
- 7:17investigated the book. Yeah, I wish I got the listing
- 7:20before my dad purchased it, but knowing that he's a cheap guy,
- 7:23probably would buy like cheapest book.
- 7:25Look for real diet cookbook on Amazon.
- 7:27Saw a book that had good reviews.
- 7:29So there was like positive reviews on there, granted, not a
- 7:32lot, but five star reviews. And so he purchased this book
- 7:35without a second thought, you know, had all sort of the
- 7:37markers you would think without looks like maybe a legitimate
- 7:40cookbook. And so when I went to go visit
- 7:41him during the holiday season, he's like, oh, I'm having a hard
- 7:44time deciphering the cookbook and I don't really understand it
- 7:46all. I was like, sure, no problem,
- 7:47I'll help you with this. When you immediately looked at
- 7:49the book, I started flipping through it.
- 7:51You could tell that this book was not written by a human and a
- 7:54lot of and so not a lot of the markers from like just
- 7:57formatting and the content we start reading it.
- 7:59You're like, this doesn't really make sense.
- 8:01And so a few things I started off quickly was just when
- 8:05Googling the name of the author, Ethan new life, absolutely
- 8:08nothing. The only thing they came up with
- 8:10that name was actually from the Amazon listing itself.
- 8:12So that was already sort of a a red flag for me, especially
- 8:15because they said he's ready to accomplish author.
- 8:17He's written all these other things.
- 8:19So it's seemingly someone who would have an online profile
- 8:21having written other things on the Internet.
- 8:23So the author didn't exist. The photo that was used on
- 8:26Amazon, it looked AI generated to me.
- 8:28I ran it through some checks. It did seem to indicate that it
- 8:31was AI generated profile picture.
- 8:33I then took a bit of the text and ran it through several
- 8:36different AI text generation services.
- 8:38All of those services estimated a very high likelihood of AI
- 8:41generated text. And so suddenly there was all
- 8:44these other components to it. So there was nothing in this
- 8:47book from the author to the content to the images or
- 8:51anything about it that seemed to be legitimate or trustworthy.
- 8:54And when you then investigated the listing, you looked at the
- 8:57reviews. These accounts here would just
- 9:00write positive reviews on a whole wide range of books.
- 9:03So you can tell that these reviews were probably not likely
- 9:05to be trustworthy and probably also false as well.
- 9:08So there was multiple failings here.
- 9:10And not to say the blame my father because he's very old and
- 9:13it'll take him a while. But there's a lot of clues here
- 9:16that we can look to in our own sort of ability to be able to go
- 9:19and decipher whether or not a book is trustworthy online.
- 9:21Because as you write there, it's not just cookbooks, it's
- 9:24something, it could be also like mushroom identification books.
- 9:27It could be actually children's books.
- 9:28It could be, you know, a lot of extra learning materials.
- 9:30There's a whole wide range of content out there, even if just
- 9:33in the book genre, people using AI generation images or AI
- 9:37generated text and also without disclosing it.
- 9:40So people are unaware that they're buying something that's
- 9:43generated by a technology versus by a human.
- 9:46So Ben, this is the soup to nuts for AI.
- 9:48It generates the content that's bogus and may be wrong.
- 9:51It creates the ad to put it on the Internet.
- 9:53It writes the fake reviews. This is all AI help on the bad
- 9:56guys here. That's right.
- 9:57And you know, when you are thinking about this whole
- 10:01ecosystem of where that AI generated content is being
- 10:04hosted, sold, promoted, shared, there are a lot of entities
- 10:09there, right? And there's a lot of questions
- 10:11about, you know, the checks that are, A, coming from the AI
- 10:16providers that are allowing you to make this content in the 1st
- 10:19place, but, B, the sort of checks and moderation practices
- 10:22that the companies hosting this content want to allow.
- 10:26And so that's a really interesting open question moving
- 10:28forward that will have a lot of impacts on what people actually
- 10:31see. Artificial intelligence has
- 10:33advanced so rapidly that it's now possible for anyone, but
- 10:35especially scammers and criminals.
- 10:37To create deep fake. Videos including in real time.
- 10:40We're going to talk more about that scary thought next.
- 10:43I'm her Weisbaum, the consumer man.
- 10:44And this is Consumerpedia, powered by checkbook.org.
- 10:49At checkbook.org we have the answers to your consumer
- 10:52questions 24/7. That's checkbook.org.
- 10:55We're a unique nonprofit that can help you with select local
- 10:58service providers, avoid trouble and save money.
- 11:02Because we don't accept any advertising or take money from
- 11:05any business, we recommend you can rely on Checkbook to be
- 11:08completely independent and objective.
- 11:10Our advice is always free. Ratings of local services are
- 11:13available with a subscription. If you live in one of these
- 11:16seven metro areas and haven't joined checkbook yet, check us
- 11:19out. Boston, Chicago, Philadelphia,
- 11:22Seattle, San Francisco, Minneapolis, Saint Paul and
- 11:26Washington DC Get your free 30 day subscription by going to
- 11:29checkbook.org/consumerpedia. Deep fakes have been around for
- 11:35a while, and until recently they were pretty.
- 11:37Easy to spot, unnatural. Movement lips.
- 11:39It didn't sink to the words, but not anymore.
- 11:41Cybercrime rings are actively weaponizing deep fake tech and
- 11:45dropping these videos on social media.
- 11:47Recent deep fake videos included Elon Musk appearing to promote
- 11:50fake crypto investment platforms.
- 11:52You can deposit any of the listed cryptocurrencies without
- 11:55worrying about your assets one bit.
- 11:56If you make the minimum deposit, doubling your money is 100%
- 12:00guaranteed. Tom Hanks supposedly pitching a
- 12:02dental insurance plan and. An AI generated Oprah.
- 12:05Endorsing weight loss gummies. I have nothing to do with weight
- 12:10loss gummies or diet pills and I don't want you all taken
- 12:14advantage of by people misusing my name.
- 12:17Many. People believe these bogus.
- 12:19Endorsements and testimonials when they see them on TikTok,
- 12:21Facebook or their other social media feeds.
- 12:24So a question for both of you, how good are these deep fake
- 12:27videos? It really does run the gamut of
- 12:30the mouth looks nothing like what the words are coming out of
- 12:33and you can sort of tell or if it is someone you have heard
- 12:37talking a million times, you know that cadence is just a
- 12:39little bit off. But honestly, you have to be
- 12:42like super vigilant most of the time to know that it would be
- 12:45definitely fake. In an ideal world, consumers
- 12:48should not have to be constantly vigilant about whether something
- 12:52is real or not. And I think you sort of have to
- 12:54be nowadays if you want to be able to differentiate some of
- 12:59them. The real answer is it kind of
- 13:00depends, but they're getting pretty good sometimes.
- 13:02And nobody looks at them like you guys do.
- 13:04You see something? It's, it's a little tiny thing
- 13:06on your screen or your phone. You look at it and go, oh, no
- 13:08one's questioning whether it's real or not.
- 13:10If it's if it looks real. Yeah, and they shouldn't have
- 13:13to. You're right.
- 13:13You know, you are just scrolling through and you get an ad for
- 13:16something that you do or don't care about.
- 13:19I mean, it might just get your attention because you never
- 13:22would have thought that you would see, you know, LeBron
- 13:24James or Tom Hanks or whatever, talk about this random insurance
- 13:27product. But yeah, it works to get your
- 13:30attention. And absolutely, it's really hard
- 13:32to know whether it's legitimate or not.
- 13:34I think that's an absolutely great point for encountering a
- 13:37lot more inauthentic content at a lot of different ways in ways
- 13:41that people aren't necessarily preparing themselves.
- 13:43And, you know, not that they should need to, shouldn't have
- 13:45to. There should be better
- 13:46protections about it because there are services out there
- 13:49that offer sort of AI spokespeople in which people who
- 13:53have sold their looks and the video of them and you can make
- 13:56them using AI technology, generate them to say anything.
- 13:59So it's not perfect, but it's good enough that most people
- 14:02wouldn't look deeper and look to say, hey, maybe this isn't a
- 14:07authentic person or this is an authentic video.
- 14:10And at the end, these Instagram wheels or Tiktok videos are
- 14:12selling a product. And so it maybe seems like, oh,
- 14:15I'm giving a testimonial about this.
- 14:16I'm talking about my career. And in reality, it is AI audio
- 14:20of a real person who's then being AI edited in a particular
- 14:23way to match that audio. And I think it becomes a really
- 14:27dangerous territory when we see these videos getting, you know,
- 14:30hundreds and thousands of likes and engagement.
- 14:33And I read the comments and very, very few people, if to
- 14:36none at all, recognize that this is in a way it's a scam because
- 14:40it makes it seem like here it's a testimonial from a
- 14:42veterinarian about a product and it's not real at all.
- 14:45Ben, you showed in your report that news stories can easily be
- 14:48faked to create a tragedy. So a charity scammer could send
- 14:52you an e-mail and say there was this volcano eruption or the
- 14:55tsunami somewhere in the world and show people in a desperate
- 14:58situation and ask for your money.
- 15:00And that never happened. And what I saw from your report
- 15:02is pretty realistic. Yeah, absolutely.
- 15:04We saw this earlier this year when there were some really bad
- 15:08hurricanes and weather events throughout the southern United
- 15:11States. So in someone's head, they're
- 15:14already primed to think, OK, there is this natural disaster.
- 15:17There are going to be calls for charity.
- 15:19Often scams are praying on current events or recent events.
- 15:23And so, you know, whether it be a earthquake in, you know, Haiti
- 15:28or somewhere else, or a photo that got really viral that I do
- 15:32highlight in the report of little girl looking absolutely
- 15:35terrified with a life raft holding a little dog.
- 15:38And yeah, people on social media, through e-mail, targeting
- 15:42e-mail campaigns, people were blasted this information and
- 15:46provided a sort of link to give money to purportedly help people
- 15:51like this little girl, when in reality that money is not going
- 15:53to help anything, just towards a scammer.
- 15:55And now deep fake videos can take place in real time, which
- 15:58is sure to make romance scams more profitable for the bad
- 16:01guys. So they're going to be able to
- 16:03now talk to you on Zoom or something like that and just be
- 16:06this beautiful person on the other end of the line.
- 16:08And it's not really them, yeah. You know, there was a recent
- 16:12illustration in in a four O 4 media article of real time
- 16:16deepfake fraud where you are able to have your real
- 16:20expression sort of have a person or face that is not yours
- 16:24layered over it while also distorting your voice to match
- 16:28that. And so with romance scams, which
- 16:31is a massive loss for people, over $670 million lost just in
- 16:372024 according to the FBI. There is already a lot of AI
- 16:41used by chat bots in the texting portion on a dating app or over
- 16:45e-mail over messaging app. But now when you in the past
- 16:50thought, OK, if I get them on the phone or a video chat, I can
- 16:53verify that they're a real person.
- 16:54That is now sort of thrown into question because the increasing
- 16:58availability of this real time video and audio masking.
- 17:02So it's sort of kind of like the technology they used when Tom
- 17:04Hanks was in The Polar Express movie.
- 17:07Tom Hanks did his thing, and then the computer showed the
- 17:09conductor doing his thing. The technology's so good.
- 17:11That's what they're doing in real time.
- 17:13Yeah. Yeah.
- 17:13Kalina, I know you have a personal story about this.
- 17:16Yes. So my father who's 82, bless
- 17:19them, he is trying to do a little bit of that online
- 17:22dating. And I think he's oftentimes the
- 17:24subject of these like romance scams.
- 17:26And I think a lot of them end up using AI for different
- 17:29components of the romance scams that he gets involved in.
- 17:32And that could be from sometimes AI edited images that are sent
- 17:37to him or even sometimes what he'll try to do, he'll ask the
- 17:40women that he talks to to send them a photo of their ID, like
- 17:43their driver's license to prove that they're real.
- 17:45And then once I take a look at it, I'm like, this is Photoshop,
- 17:47this is edited. And then I sometimes will see a
- 17:50video. He's like, oh, no, they sent me
- 17:51a video of them. And very clearly to me, I can
- 17:54tell that this is AI audio and this is a image that they've
- 17:57tried to turn into a video and it's just not done very well.
- 18:00And so the bar seems to be even really low.
- 18:03And you know, he'll say they're like, no, I talk to them on
- 18:05video, I talk to them. And then when I get to see the
- 18:07content, cuz maybe he'll try to record it on another device for
- 18:10me to prove that they're real. It's an AI image, it's AI audio.
- 18:13And it's something that's happening.
- 18:14I can tell you for sure with my own father today.
- 18:18And as much as I like to try to prove that some of these women
- 18:21are inauthentic or they're not real, it becomes really hard to
- 18:24convince someone, particularly when they're emotionally
- 18:26involved in someone. So there was this like component
- 18:28of being invested into a online relationship.
- 18:31You're like, no, I've been talking to them.
- 18:32I've seen video, I've seen images, I've heard.
- 18:34And yet I have to tell him that's not real, that's
- 18:37inauthentic. And it's like, how do you
- 18:39convince someone who's like seeing what they see?
- 18:41Yeah. And so I think it is something
- 18:43that it's it's happening today. And I think that is something
- 18:47that's really dangerous because even though my father is 82, it
- 18:50doesn't mean it's only something that's happening to senior
- 18:52citizens. It could happen to anyone of any
- 18:54age. I think there was this really
- 18:55viral story of this woman who thought she was in a
- 18:58relationship with, I think Brad Pitt with these AI, the photos
- 19:02there, and she got scammed out of a large sum of money thinking
- 19:05that he was in a hospital and needed to send him money.
- 19:07And so, you know, it could happen to anyone.
- 19:10She wasn't elderly or anything. So it's really, really something
- 19:13that I think we need to be conscientious of from classic,
- 19:16you know, scammer groups like the Yahoo Boys based in Nigeria
- 19:19that are using that technology that been mentioned from the 404
- 19:22article to a wide variety of groups where we don't even know
- 19:25what kind of technology because the technology that's available
- 19:27off the shelf, that's not something that's relegated to
- 19:29just a Hollywood movie set. That's something you can
- 19:32purchase and use today with battered low tech literacy is
- 19:35convincing to ready a large enough swath of the population
- 19:39that you'd make your investment on that technology back like 10
- 19:42or 100 times fold. It does take work, it does take
- 19:44time, and the amount of damage a singular person can do I think
- 19:48has exponentially increased with the availability of AI
- 19:51technology. And Dad, if you're listening to
- 19:53this segment, from now on, talk to your daughter before you fall
- 19:56in love with someone. OK?
- 19:57That's just my consumer man advice.
- 19:59Yes, absolutely. So how do you protect yourself
- 20:03and will? AI be.
- 20:04Used to police. AI we're.
- 20:06Going to talk about that next. I'm Herb Weisbaum, and this is
- 20:08Consumer Pedia powered by checkbook.org.
- 20:12Consumer Pedia. Fast Facts The losses to cyber
- 20:16fraud hit a new record in 2024. According to the FBI's Internet
- 20:21Crime Complaint Center, cyber criminal stole $16.6 billion
- 20:25from Americans last year, an increase of 33% from 2023.
- 20:30More than half of these losses, $9 billion, resulted in scams
- 20:35involving cryptocurrency, both bogus crypto investment schemes
- 20:39and making payments to scammers via a digital currency.
- 20:42Cryptocurrency related fraud increased 66% from 2023.
- 20:47While anyone can fall for a cyber scam, the FBI says
- 20:50individual 60 and older are greatly affected.
- 20:56AI fakes are here and they're getting.
- 20:57Harder to spot, so we need. To know how to deal with it,
- 21:00Kalina and Ben. Sometimes it's easy to spot a
- 21:02computer generated image. A hand with six figures, tables
- 21:05or chairs that are floating in the air because they don't have
- 21:07legs, Skin that looks like plastic.
- 21:09But AI is getting so much better, Kalina.
- 21:11You know how to look for shadows in the wrong place or
- 21:14perpendicular lines that don't stay perpendicular.
- 21:16But for most of us, we don't have that skill set.
- 21:18We see an image and we assume it may have been touched up a
- 21:21little bit, but it's not totally fake.
- 21:22What are we to do? I think that's a great question.
- 21:25I, you know, I, I think learning the skills is an important part,
- 21:28but of course, you know, not everyone needs to do that as
- 21:31much as I would want everyone to learn those skills because I
- 21:33think if we rely solely on an AI detection tool, those tools
- 21:38could be fooled. So I think relying on a tool is
- 21:41not going to be the best way. And I would personally, and
- 21:43it's, it's a hot take, but I would love to see regulation in
- 21:46the area. I think, you know, as AI image
- 21:48generation video audio has really just exploded in the past
- 21:52few years, we are so behind on like what we should actually be
- 21:56doing. Like there's that famous quote
- 21:58from Jeff Goldblum's character in Jurassic Park where he says,
- 22:01you know, your scientists are so busy thinking whether or not
- 22:04they could, they forgot to like ask themselves that they should
- 22:06something along those lines. And just because we have the
- 22:09ability to have this technology, we shouldn't just let it go
- 22:12completely haywire, which is kind of what it feels like going
- 22:15right now. I spend a lot of my time
- 22:17thinking about like AI generated images, particularly when it
- 22:19comes to non consensual imagery, non consensual sexually explicit
- 22:22imagery. And of course, that technology
- 22:24exists and it's meant to be convincing and a lot of it's
- 22:27getting gotten a lot better just over the past year.
- 22:30But just because that technology exists and it's capable doesn't
- 22:33mean that should be a technology that is widely accessible and
- 22:36used. You know, there should be really
- 22:37tight restrictions on technology that has great potential to
- 22:41cause harm. And we just don't have any of
- 22:43that right now. In fact, there's lots of
- 22:45businesses, their whole model is creating this technology to make
- 22:48a profit off of it, right? You see these businesses out
- 22:51there offering these AI tools that are used consistently time
- 22:54and time again to either scam or harm or defraud in some way.
- 22:58And that's perfectly legal with no sort of repercussions on to
- 23:01anyone. And so I really do think as
- 23:03controversial as a take it might be to some communities, I think
- 23:06regulation, even though it's not the perfect solution, might be
- 23:09the only or some sort of consequences, some sort of way
- 23:12to help at least realist in. Otherwise people are just going
- 23:15to take a profit off of it and this harm is going to be done to
- 23:18just so many different communities.
- 23:20And you deal with people on Capitol Hill, and we're not
- 23:23exactly in a point in our country where we like
- 23:26regulations. So while I agree with everything
- 23:28Kalina said, I don't know what if there's going to be a
- 23:31regulatory fix to companies somehow have to be held
- 23:33responsible, or what are we going to do about this?
- 23:35I mean, I think it is a sort of all the above situation.
- 23:40There is definitely not a magical legislative fix.
- 23:44Even if I had a magic wand, it's really hard to get restrictions
- 23:49and things clamped down on the ability to make or send this
- 23:54stuff in general. But where we can and do need to
- 23:58beefed up with regulatory resources are things like
- 24:02providing more resources to the state attorneys general as well
- 24:05as the FBI and FTC, who are tasked with enforcing a lot of
- 24:09these scams. There needs to be better
- 24:12reporting mechanism so that when an individual is on a dating app
- 24:16or a social media platform or an e-mail provider or your text,
- 24:20when you report that junk, that it goes on to the appropriate
- 24:23authorities so that people know what the trends are.
- 24:26Education and putting it on the consumer to figure out whether
- 24:30it's AI or not is not a sustainable solution.
- 24:33I think some of the things that consumers can do is to follow
- 24:37sources that publish information about what these scams are.
- 24:41So whether that is something like Bellingcat or things like
- 24:44the AARP Scam Watch or the Better Business Bureau Scam
- 24:47Tracker, or even a subreddit called Scams on Reddit, there
- 24:52are a lot of sources where people can just see in their
- 24:57news feed or in an e-mail newsletter this type of scam is
- 25:01happening more and more so that you can be aware of it.
- 25:03I think the other thing that we try to tell folks is to report
- 25:07when things happen. And you might not get recourse
- 25:10for the harm that you suffered as part of a scam, but you might
- 25:14help other people watch out for it or help enforcing agencies
- 25:18catch it. And then, you know, sort of us
- 25:20would at least the the companies do have to be a little bit more
- 25:23responsible with how they roll these things out.
- 25:26They are absolutely profit making machines, regardless of
- 25:29the rhetoric about how much open AI wants to benefit all of
- 25:33humanity or whatever they say. But at the same time, they then
- 25:37release these tools with very little guardrails.
- 25:40And so there's a real choices to be made at every stage, whether
- 25:43it is a chatbot or a audio generation platform, there are
- 25:48choices about what you are going to actually respond to.
- 25:52You know, sometimes you might get a response on a chat bot
- 25:54saying, oh, I can't help you with that.
- 25:56If something seems clearly illegal, there's a lot of ways
- 25:58to get around that on audio and video, Consumer Reports study
- 26:02that was super recent illustrated that almost none of
- 26:05the commercially available audio and video generation apps
- 26:08require authentication of the actual person you are trying to
- 26:13replicate. So a lot of the ways these
- 26:14things work, you know, you could take this podcast and upload 10
- 26:17seconds of my voice to an audio generation platform and then you
- 26:21could get me to sort of say whatever I want.
- 26:24And so there are methods to require authentication from me,
- 26:29or at least efforts to do that so that someone can't just pick
- 26:32that up and do it. And then that reduces the
- 26:34likelihood of it being disseminated if it has to be
- 26:37authorized. So, you know, there are little
- 26:40things here and there that the companies have to take
- 26:42responsibility for, that the platforms that this content is
- 26:45shared on and through have to take responsibility for in
- 26:48moderation to the extent that whether it's through state
- 26:52legislation, federal legislation or enforcement agencies, some of
- 26:57these people both doing the scams but also facilitating
- 27:00them, have to be held responsible.
- 27:01And I think the bottom line and the take away from this episode
- 27:04should be slow down. Criminals always try to rush you
- 27:07to get you to act on impulse. Don't believe everything you
- 27:10see. Assume anything and everything
- 27:12can be faked and you need to check it out before you take any
- 27:15action and especially if you give away any personal
- 27:17information or any money. Want to thank both of thank you
- 27:20for joining us. Kalina Koltai is a senior
- 27:22researcher at Bellingcat and Investigative Journalism Group.
- 27:24Thank you, Kalina. Thank you for having me here.
- 27:27And Ben Winters, director of AI and Privacy.
- 27:29At the consumer. Federation of America.
- 27:31Thanks so much. I put a link to Kalina's report
- 27:33on spotting. AI Fakes and Ben's report on how
- 27:36AI will become a normal part of business for fraudsters in the
- 27:39show notes. You might also want to check out
- 27:41these other consumer pedia podcasts they deal with
- 27:44Protecting yourself from Fraud episode 11.
- 27:47How to protect Yourself from Identity Theft.
- 27:49Episode 74, Why Passkeys are better than passwords.
- 27:53And episode 76, fake reviews are everywhere.
- 27:57And with that, we come to the end of another episode.
- 28:00We hope you'll rate this one and follow us on Apple Podcasts,
- 28:03Spotify, or wherever you listen to your podcasts.
- 28:06Remember, we release new episodes every other Thursday.
- 28:10Another way you can support this show is to follow us on
- 28:12Consumerpedia on Facebook and Instagram, and at my
- 28:16Consumerpedia on X I'm Herb Weisbaum, thanks for listening.
- 28:20Consumer Pedia is a public service of checkbook.org.
- 28:24We're a unique nonprofit that helps you save money and make
- 28:26smarter choices. You can count on Checkbook to
- 28:29help you find the best services and avoid the worst with local
- 28:32ratings that are accurate and unbiased.
- 28:35If you live in or around these seven cities and haven't joined
- 28:38Checkbook yet, check us out Boston, Chicago, Philadelphia,
- 28:42Minneapolis, Saint Paul, Seattle, San Francisco, and
- 28:45Washington DC. To get your free 30 day
- 28:47subscription, go to checkbook.org slash Consumer
- 28:50Pedia Consumer Pedia. Empowering consumers to save
- 28:54money and make smarter choices.