1 00:00:00,080 --> 00:00:03,960 Speaker 1: A note for listeners, this episode includes some disturbing descriptions 2 00:00:04,040 --> 00:00:07,200 Speaker 1: of sexual acts and assault. If you have kids around, 3 00:00:07,320 --> 00:00:10,200 Speaker 1: you might want to use headphones and please take care 4 00:00:10,280 --> 00:00:15,240 Speaker 1: when listening. If you've looked at headlines about Silicon Valley 5 00:00:15,280 --> 00:00:17,560 Speaker 1: at all in the past year, you've likely heard of 6 00:00:17,680 --> 00:00:22,560 Speaker 1: artificial intelligence and generative AI tools think chat, GPT, or 7 00:00:22,600 --> 00:00:26,560 Speaker 1: stable diffusion. They're touted as the next big thing, but 8 00:00:26,640 --> 00:00:29,520 Speaker 1: the use of these applications can have a dark side. 9 00:00:29,600 --> 00:00:32,440 Speaker 1: Many of them are open source and freely available, and 10 00:00:32,560 --> 00:00:36,120 Speaker 1: users are able to alter publicly available photos of people, 11 00:00:36,479 --> 00:00:40,239 Speaker 1: including images taken from social media, to depict events that 12 00:00:40,320 --> 00:00:44,120 Speaker 1: never happen in real life. These images are called deep fakes, 13 00:00:44,560 --> 00:00:49,160 Speaker 1: and oftentimes they're being altered in sexually explicit ways. While 14 00:00:49,200 --> 00:00:53,040 Speaker 1: the photos are fake, the harm is real. In many cases, 15 00:00:53,080 --> 00:00:55,640 Speaker 1: the victims are young people who don't know the images 16 00:00:55,720 --> 00:00:59,880 Speaker 1: exist until someone they know discovers them online. Bloomberg and 17 00:01:00,000 --> 00:01:03,120 Speaker 1: investigative reporter Olivia Carville went to a town in New 18 00:01:03,200 --> 00:01:04,280 Speaker 1: York where this happened. 19 00:01:04,880 --> 00:01:08,120 Speaker 2: In terms of the impacts to the young woman, Nightmare 20 00:01:08,240 --> 00:01:11,280 Speaker 2: is one of them lost over twenty pounds from streets. 21 00:01:11,680 --> 00:01:14,760 Speaker 2: A number of them started carrying pepper spray in their handbags. 22 00:01:15,040 --> 00:01:18,440 Speaker 2: Two of them started carrying knives in their handbags, and 23 00:01:18,480 --> 00:01:21,520 Speaker 2: they still feel a sense of trauma from what happened. 24 00:01:22,400 --> 00:01:24,880 Speaker 1: As awful as these photos can be, law enforcement and 25 00:01:24,959 --> 00:01:27,640 Speaker 1: victims have found it difficult to use the laws currently 26 00:01:27,680 --> 00:01:30,160 Speaker 1: on the books to bring the people making these images 27 00:01:30,200 --> 00:01:35,240 Speaker 1: to justice. Bloomberg cyber reporter Margie Murphy says prosecutors can 28 00:01:35,280 --> 00:01:36,600 Speaker 1: find themselves in a bind. 29 00:01:37,280 --> 00:01:41,160 Speaker 3: It's still incredibly blurry. A lot of people within law 30 00:01:41,280 --> 00:01:44,280 Speaker 3: enforcement that I spoke to you for the story all 31 00:01:44,360 --> 00:01:48,000 Speaker 3: said that it feels really mishmash at the moment between 32 00:01:48,040 --> 00:01:50,920 Speaker 3: state laws and federal laws, and that the law really 33 00:01:50,920 --> 00:01:53,000 Speaker 3: hasn't kept up with the technology in this area. 34 00:01:53,760 --> 00:01:57,760 Speaker 1: Olivia and Marghi are here to detail their business week investigation. 35 00:02:04,200 --> 00:02:07,600 Speaker 1: I'm Nancy Cook today. I'm the big take how AI 36 00:02:07,760 --> 00:02:18,760 Speaker 1: generated deep fake images are hurting real people. So Margie, 37 00:02:18,880 --> 00:02:21,200 Speaker 1: let's define the terms that we're going to talk about. 38 00:02:21,240 --> 00:02:22,920 Speaker 1: What is a deep fake photo? 39 00:02:23,840 --> 00:02:27,200 Speaker 3: So a deep fake photo is a picture which has 40 00:02:27,240 --> 00:02:31,480 Speaker 3: been digitally manipulated to depict an individual doing something they 41 00:02:31,520 --> 00:02:36,519 Speaker 3: didn't and that's using a deep learning model, so artificial intelligence. 42 00:02:36,560 --> 00:02:40,000 Speaker 3: That's where the deep comes from. And so right now 43 00:02:40,040 --> 00:02:43,919 Speaker 3: today that would be using open Aiyes, Dolly is one 44 00:02:43,919 --> 00:02:47,000 Speaker 3: that you may have heard of. There's mid journey, there's 45 00:02:47,080 --> 00:02:52,560 Speaker 3: stable diffusion tools where you can basically ask a machine 46 00:02:52,680 --> 00:02:55,720 Speaker 3: to create an image that looks incredibly realistic, but it 47 00:02:55,800 --> 00:02:56,440 Speaker 3: never happened. 48 00:02:57,320 --> 00:03:00,000 Speaker 1: And what is generative AI as you talk about it 49 00:03:00,200 --> 00:03:00,440 Speaker 1: in the. 50 00:03:00,360 --> 00:03:03,840 Speaker 3: Piece, the generative AI that we use at the moment 51 00:03:03,960 --> 00:03:07,280 Speaker 3: tends to be using text to image. So it's the 52 00:03:07,320 --> 00:03:10,920 Speaker 3: ability to kind of type in and say picture of 53 00:03:11,400 --> 00:03:14,720 Speaker 3: person eating an apple, and the machine can generate that 54 00:03:14,800 --> 00:03:18,000 Speaker 3: image for you, but you know, no photograph was ever taken, 55 00:03:18,080 --> 00:03:19,399 Speaker 3: but it looks incredibly real. 56 00:03:21,880 --> 00:03:24,920 Speaker 1: And Olivia, you open your piece with a chilling story 57 00:03:24,960 --> 00:03:26,800 Speaker 1: out of New York. Can you tell us what happened? 58 00:03:27,600 --> 00:03:27,840 Speaker 3: Yeah? 59 00:03:27,880 --> 00:03:32,160 Speaker 2: Sure. So I work on Bloomberg's investigations team and had 60 00:03:32,160 --> 00:03:35,240 Speaker 2: been interested in deep fakes, I think for most of 61 00:03:35,280 --> 00:03:37,280 Speaker 2: the world. We came to learn about what a deep 62 00:03:37,280 --> 00:03:40,320 Speaker 2: fake image is from that iconic shot of the Pope 63 00:03:40,440 --> 00:03:43,200 Speaker 2: in a puffer jacket, and I did a lot of 64 00:03:43,240 --> 00:03:45,600 Speaker 2: research around deep fakes and then partner it up with 65 00:03:45,680 --> 00:03:47,960 Speaker 2: Margie and we decided to try and find a case 66 00:03:48,320 --> 00:03:52,680 Speaker 2: which illustrates just how damaging this technology can be, and 67 00:03:52,720 --> 00:03:55,440 Speaker 2: we found that case in Long Island, a suburb of 68 00:03:55,480 --> 00:03:58,520 Speaker 2: New York City. This is one of the first deep 69 00:03:58,560 --> 00:04:02,880 Speaker 2: fake related conviction in US history. And what happened here 70 00:04:03,040 --> 00:04:07,000 Speaker 2: is a young man decided to use software tools on 71 00:04:07,040 --> 00:04:12,640 Speaker 2: the Internet to manipulate doctor alter images of girls from 72 00:04:12,640 --> 00:04:16,200 Speaker 2: his high school. He targeted about forty different girls that 73 00:04:16,240 --> 00:04:18,920 Speaker 2: he went to school with and did so in very 74 00:04:18,920 --> 00:04:23,919 Speaker 2: horrific ways. In some cases, he digitally undressed them, He 75 00:04:24,000 --> 00:04:27,240 Speaker 2: put their faces onto the bodies of women engaging in 76 00:04:27,320 --> 00:04:30,560 Speaker 2: sex acts. He used photographs from when they were as 77 00:04:30,600 --> 00:04:34,760 Speaker 2: young as five years old. And this case really shows 78 00:04:34,880 --> 00:04:38,159 Speaker 2: us just how difficult it is to lay charges against 79 00:04:38,200 --> 00:04:42,240 Speaker 2: an individual who created fake or deep faked pornography. 80 00:04:43,120 --> 00:04:45,279 Speaker 1: And do we have an inkling about who was the 81 00:04:45,279 --> 00:04:46,800 Speaker 1: person responsible for this? 82 00:04:47,720 --> 00:04:50,080 Speaker 2: Yeah, that's one of the most interesting parts of the case. 83 00:04:50,160 --> 00:04:53,560 Speaker 2: We felt is that the young woman who had been 84 00:04:53,600 --> 00:04:58,480 Speaker 2: targeted in these images actually investigated who was behind them 85 00:04:58,640 --> 00:05:02,120 Speaker 2: because they felt that authorities weren't doing enough to track 86 00:05:02,160 --> 00:05:06,120 Speaker 2: him down. The story really starts on New Year's Eve 87 00:05:06,320 --> 00:05:10,520 Speaker 2: twenty twenty, where a group of young women in Leavittown 88 00:05:11,040 --> 00:05:15,599 Speaker 2: came to discover photographs of them had been altered, manipulated, 89 00:05:16,160 --> 00:05:20,800 Speaker 2: turned pornographic and nature and uploaded to this website. The 90 00:05:20,839 --> 00:05:24,680 Speaker 2: website has a very graphic internet address name that's come 91 00:05:24,800 --> 00:05:28,080 Speaker 2: on printed picks dot com. And on New Year's Eve, 92 00:05:28,279 --> 00:05:32,159 Speaker 2: this group in Leavittown actually started discovering their images had 93 00:05:32,200 --> 00:05:35,320 Speaker 2: been uploaded to the site, and word got out. They 94 00:05:35,360 --> 00:05:38,440 Speaker 2: began messaging one another, sharing a link to the website 95 00:05:38,760 --> 00:05:41,800 Speaker 2: and discovered just how horrific some of these images and 96 00:05:41,800 --> 00:05:44,000 Speaker 2: the content that had gone up on this website had been. 97 00:05:44,720 --> 00:05:48,279 Speaker 2: In some cases, their poster was talking about raping and 98 00:05:48,360 --> 00:05:51,000 Speaker 2: murdering the girls. And so it was that New Year's 99 00:05:51,000 --> 00:05:53,360 Speaker 2: Eve that they first started reaching out to the police. 100 00:05:53,800 --> 00:05:56,760 Speaker 2: And throughout the course of twenty twenty one, the young 101 00:05:56,800 --> 00:06:00,560 Speaker 2: women themselves were investigating who was behind this, who had 102 00:06:00,640 --> 00:06:04,040 Speaker 2: uploaded all of their photos, altered them, turned them pornographic, 103 00:06:04,160 --> 00:06:07,080 Speaker 2: and posted them onto this website, and a lot of 104 00:06:07,080 --> 00:06:10,480 Speaker 2: them came to the conclusion that they knew who it was, 105 00:06:10,560 --> 00:06:13,760 Speaker 2: that they knew who the suspect was. They believed it 106 00:06:13,920 --> 00:06:17,080 Speaker 2: was a man that had attended school with them, a 107 00:06:17,120 --> 00:06:20,360 Speaker 2: former classmate, an actual friend of many of theirs as 108 00:06:20,400 --> 00:06:24,360 Speaker 2: well they recognized his handwriting and some of the images. 109 00:06:24,440 --> 00:06:28,039 Speaker 2: They recognized his writing style in some of the captions 110 00:06:28,120 --> 00:06:31,640 Speaker 2: on the website that he posted their images too. And 111 00:06:31,800 --> 00:06:34,960 Speaker 2: it was the young woman who actually took his name 112 00:06:35,080 --> 00:06:37,880 Speaker 2: to the police and said they believed he was behind 113 00:06:37,880 --> 00:06:38,200 Speaker 2: it all. 114 00:06:38,880 --> 00:06:41,760 Speaker 1: And Olivia, how old were these young women when this 115 00:06:41,920 --> 00:06:43,640 Speaker 1: happened and how old are they now? 116 00:06:44,520 --> 00:06:47,599 Speaker 2: Most of the young women were around eighteen or nineteen 117 00:06:47,680 --> 00:06:50,240 Speaker 2: years old when they discovered this website. It was the 118 00:06:50,320 --> 00:06:53,760 Speaker 2: year after they graduated from high school in Levettown, and 119 00:06:53,880 --> 00:06:56,920 Speaker 2: today they're around twenty two or twenty three years old. 120 00:06:59,000 --> 00:07:02,960 Speaker 1: Did they have a hard time getting the authorities involved 121 00:07:03,000 --> 00:07:06,360 Speaker 1: in this, including the local police, and making sure that 122 00:07:06,400 --> 00:07:07,560 Speaker 1: they knew it was a problem. 123 00:07:08,240 --> 00:07:11,480 Speaker 2: I think there was definitely a frustration that the police, 124 00:07:11,960 --> 00:07:15,360 Speaker 2: the families felt, at least, weren't moving fast enough to 125 00:07:15,440 --> 00:07:18,600 Speaker 2: try and do something about this case. It's important to 126 00:07:18,640 --> 00:07:23,200 Speaker 2: remember that cyber harassment cases or online vulgarity isn't really 127 00:07:23,320 --> 00:07:27,600 Speaker 2: high on police priority lists. In this example, you had 128 00:07:27,760 --> 00:07:33,000 Speaker 2: an individual who was creating fake images, and there's no 129 00:07:33,200 --> 00:07:37,120 Speaker 2: real law that says what he was doing was illegal. 130 00:07:37,720 --> 00:07:41,240 Speaker 2: So while it was grotesque and you'd look at the site, 131 00:07:41,320 --> 00:07:44,920 Speaker 2: and certainly the woman that I talked to were disgusted 132 00:07:45,000 --> 00:07:48,080 Speaker 2: by what they saw. But the difficulty is that there 133 00:07:48,120 --> 00:07:50,960 Speaker 2: are no laws in the US that said what this 134 00:07:51,120 --> 00:07:53,680 Speaker 2: young man was doing was actually illegal. 135 00:07:54,480 --> 00:07:57,080 Speaker 1: And Margie, can you explain how the new advances we've 136 00:07:57,120 --> 00:07:59,680 Speaker 1: heard about this past year in gener of AI are 137 00:07:59,680 --> 00:08:02,280 Speaker 1: making it easier to generate deep photos? 138 00:08:03,040 --> 00:08:07,600 Speaker 3: Yes, so the new advances have just one improved the 139 00:08:07,720 --> 00:08:11,240 Speaker 3: look of these pictures, so they just look incredibly realistic 140 00:08:11,280 --> 00:08:13,400 Speaker 3: now in a way that a few years ago, when 141 00:08:13,440 --> 00:08:16,440 Speaker 3: we first started hearing about deep fake and deep fake pornography, 142 00:08:16,480 --> 00:08:20,000 Speaker 3: they just didn't quite look as convincing. And secondly, a 143 00:08:20,040 --> 00:08:22,800 Speaker 3: lot of the tools that have come out are open source, 144 00:08:22,920 --> 00:08:27,760 Speaker 3: which means it's really easy for people to download the model, 145 00:08:28,200 --> 00:08:31,400 Speaker 3: train the model themselves. A lot of the technologies, you know, 146 00:08:31,400 --> 00:08:36,240 Speaker 3: they've been created so you don't need super compute power 147 00:08:36,440 --> 00:08:39,240 Speaker 3: to process these models. The idea was to kind of 148 00:08:39,240 --> 00:08:42,640 Speaker 3: democratize them and make it so that everyone could use 149 00:08:42,679 --> 00:08:46,360 Speaker 3: their PC to kind of generate images and videos. Not 150 00:08:46,440 --> 00:08:50,240 Speaker 3: for bad necessarily, but of course with the Internet, there 151 00:08:50,280 --> 00:08:53,960 Speaker 3: are always people who will take these great technologies and 152 00:08:54,080 --> 00:08:56,600 Speaker 3: do horrible things with them, and so this open source 153 00:08:56,720 --> 00:08:59,160 Speaker 3: nature means that kind of once the cats out of 154 00:08:59,200 --> 00:09:01,800 Speaker 3: the bag, it's very difficult for people who develop these 155 00:09:01,800 --> 00:09:05,079 Speaker 3: technologies to put guardrails in place, and some of them have, 156 00:09:05,400 --> 00:09:09,040 Speaker 3: and they've been very upfront about talking about how, you know, 157 00:09:09,080 --> 00:09:12,720 Speaker 3: we're trying to help also do deep fake detection. So 158 00:09:12,800 --> 00:09:15,120 Speaker 3: once deep fakes are out there, we can say, oh, yes, 159 00:09:15,200 --> 00:09:17,680 Speaker 3: that was our tool, and you know, be careful that 160 00:09:17,800 --> 00:09:20,280 Speaker 3: isn't a real image. But once it's out there and 161 00:09:20,320 --> 00:09:23,600 Speaker 3: people have access to the core technology, they can remove 162 00:09:23,640 --> 00:09:26,400 Speaker 3: those guardrails and they can kind of do what they 163 00:09:26,400 --> 00:09:29,000 Speaker 3: want with them, which is why we've seen this rise 164 00:09:29,160 --> 00:09:32,880 Speaker 3: in deep fake pornography and this kind of cottage industry 165 00:09:32,920 --> 00:09:37,640 Speaker 3: that's emerged around undressing apps, which have just proliferated across 166 00:09:37,640 --> 00:09:38,559 Speaker 3: the Internet. 167 00:09:39,200 --> 00:09:41,760 Speaker 1: Olivia, you spoke to many of the women affected in 168 00:09:41,840 --> 00:09:43,960 Speaker 1: this New York case. What did they say about how 169 00:09:44,000 --> 00:09:46,440 Speaker 1: it impacted them and what they kept trying to do 170 00:09:46,520 --> 00:09:47,560 Speaker 1: to call attention to it. 171 00:09:48,320 --> 00:09:50,640 Speaker 2: I think one of the scariest things for some of 172 00:09:50,640 --> 00:09:53,240 Speaker 2: the women I spoke to was just how realistic those 173 00:09:53,280 --> 00:09:57,120 Speaker 2: images looked. They looked at the photo where they recall 174 00:09:57,280 --> 00:09:59,560 Speaker 2: taking it, and they were wearing a bikini or a 175 00:09:59,600 --> 00:10:02,520 Speaker 2: t shoe, but the image that had been uploaded to 176 00:10:02,600 --> 00:10:06,640 Speaker 2: this website, it showed them completely naked, and in some 177 00:10:06,840 --> 00:10:11,360 Speaker 2: cases the breasts that had been imposed on their bodies 178 00:10:11,600 --> 00:10:15,240 Speaker 2: were actually so realistic looking that they believed that other 179 00:10:15,240 --> 00:10:17,640 Speaker 2: people would think it was really them. And I think 180 00:10:17,679 --> 00:10:20,080 Speaker 2: that's what was one of the hardest parts for some 181 00:10:20,160 --> 00:10:24,160 Speaker 2: of the young women involved. In others, they were just scared, 182 00:10:24,520 --> 00:10:27,440 Speaker 2: to be honest, that was I think the biggest emotion 183 00:10:27,600 --> 00:10:29,600 Speaker 2: that came from the young woman that I talked to 184 00:10:29,880 --> 00:10:33,600 Speaker 2: was a fear of men, a fear of the internet. 185 00:10:34,240 --> 00:10:38,480 Speaker 2: Not only were their photos uploaded to this awful site, 186 00:10:38,760 --> 00:10:43,400 Speaker 2: but in some cases, the poster was actually encouraging users 187 00:10:43,440 --> 00:10:47,079 Speaker 2: of this site to reach out to the girls individually, 188 00:10:47,280 --> 00:10:50,880 Speaker 2: to contact them directly, so he posted their full names, 189 00:10:51,160 --> 00:10:56,240 Speaker 2: their addresses, their social media handles. I talked to a 190 00:10:56,280 --> 00:10:59,040 Speaker 2: lot of parents as well because of the sensitive nature 191 00:10:59,080 --> 00:11:03,120 Speaker 2: of the case, and the emotion I got was just 192 00:11:03,520 --> 00:11:09,200 Speaker 2: raw anger and also frustration that what happened took their 193 00:11:09,280 --> 00:11:13,880 Speaker 2: daughter's innocence away and they felt like they felt let 194 00:11:13,960 --> 00:11:16,720 Speaker 2: down by the authorities because of how long it took 195 00:11:16,760 --> 00:11:19,520 Speaker 2: to resolve the case and how challenging it was to 196 00:11:19,600 --> 00:11:23,000 Speaker 2: prosecute it because, through the eyes of the parents and 197 00:11:23,040 --> 00:11:26,440 Speaker 2: their daughters, this was a crime. They didn't care about 198 00:11:26,440 --> 00:11:28,640 Speaker 2: the nuances of the law or the fact that the 199 00:11:28,720 --> 00:11:32,280 Speaker 2: law hadn't caught up with the deep fake technology. What 200 00:11:32,320 --> 00:11:34,640 Speaker 2: they cared about was the fact that the prime suspect 201 00:11:34,760 --> 00:11:37,320 Speaker 2: was in their community and no one was doing anything 202 00:11:37,360 --> 00:11:39,840 Speaker 2: about it. So, in terms of the impacts to the 203 00:11:39,880 --> 00:11:43,840 Speaker 2: young woman nightmares, one of them lost over twenty pounds 204 00:11:43,840 --> 00:11:47,040 Speaker 2: from stress, A number of them started carrying pepper spray 205 00:11:47,080 --> 00:11:50,280 Speaker 2: in their handbags, two of them started carrying knives in 206 00:11:50,320 --> 00:11:53,760 Speaker 2: their handbags, and they still feel a sense of trauma 207 00:11:53,840 --> 00:11:54,679 Speaker 2: from what happened. 208 00:11:56,480 --> 00:11:59,839 Speaker 1: When we're back why it's difficult for investigators to charge 209 00:12:00,000 --> 00:12:03,800 Speaker 1: a ledged perpetrators with a crime for creating and posting 210 00:12:03,880 --> 00:12:15,959 Speaker 1: deep bake images. Olivia, You've mentioned how hard it was 211 00:12:16,280 --> 00:12:20,200 Speaker 1: for police and these young women to actually figure out 212 00:12:20,360 --> 00:12:24,000 Speaker 1: who was behind posting the deep fakes. How long did 213 00:12:24,040 --> 00:12:26,199 Speaker 1: they suspect the person who was behind it and how 214 00:12:26,200 --> 00:12:28,319 Speaker 1: did they sort of figure that out? 215 00:12:28,559 --> 00:12:31,360 Speaker 2: Well, the young woman involved felt like they knew who 216 00:12:31,400 --> 00:12:35,320 Speaker 2: was behind it very shortly after discovering the website. Within days, 217 00:12:35,400 --> 00:12:37,520 Speaker 2: they had a suspect in their mind, and that was 218 00:12:37,520 --> 00:12:40,680 Speaker 2: Patrick Kerrey, a nineteen year old man who had attended 219 00:12:40,760 --> 00:12:43,120 Speaker 2: high school with them and was actually a friend of 220 00:12:43,360 --> 00:12:45,800 Speaker 2: a lot of the young women who were targeted. The 221 00:12:45,880 --> 00:12:50,360 Speaker 2: problem was there was no real evidence connecting Patrick Kerrey 222 00:12:50,880 --> 00:12:54,120 Speaker 2: to the website or to these posts. When he would 223 00:12:54,160 --> 00:12:57,840 Speaker 2: upload the photos, he would do so anonymously from usernames 224 00:12:57,920 --> 00:13:02,000 Speaker 2: like Siri, Jeinefeld or tween hunter. He never used his 225 00:13:02,200 --> 00:13:06,200 Speaker 2: real name or his real identity. And the challenge from 226 00:13:06,240 --> 00:13:11,080 Speaker 2: the police perspective, they didn't have enough evidence to get 227 00:13:11,080 --> 00:13:14,760 Speaker 2: a warrant out to subpoena his IP address to really 228 00:13:15,000 --> 00:13:18,480 Speaker 2: prove that he was the man behind this campaign, and 229 00:13:18,520 --> 00:13:20,560 Speaker 2: it would take a lot of effort from the police's 230 00:13:20,600 --> 00:13:25,280 Speaker 2: part to really investigate this case because unfortunately, there were 231 00:13:25,320 --> 00:13:29,800 Speaker 2: no laws that Carrie had broken by posting fake pornographic 232 00:13:29,920 --> 00:13:31,360 Speaker 2: images of these women. 233 00:13:32,160 --> 00:13:35,000 Speaker 1: And Olivia, who was Ana and what did she have 234 00:13:35,160 --> 00:13:37,040 Speaker 1: to do with finding a break in this case? 235 00:13:37,960 --> 00:13:41,360 Speaker 2: So Anna is a really popular cheerleader from General Douglas 236 00:13:41,400 --> 00:13:45,240 Speaker 2: MacArthur High School. She'd actually known Patrick Carrey from the 237 00:13:45,280 --> 00:13:48,199 Speaker 2: age of five years old. She lives a stone's throw 238 00:13:48,360 --> 00:13:51,000 Speaker 2: from his home and they went to the same elementary 239 00:13:51,000 --> 00:13:56,320 Speaker 2: school together. She grew increasingly frustrated throughout twenty twenty one 240 00:13:56,559 --> 00:13:59,760 Speaker 2: at the lack of response around the case and also 241 00:14:00,080 --> 00:14:03,000 Speaker 2: some of the disgusting things that were being posted about her. 242 00:14:03,679 --> 00:14:07,840 Speaker 2: There was one example where the harasser had included an 243 00:14:07,920 --> 00:14:10,959 Speaker 2: image that looked as though she had been murdered, and 244 00:14:11,120 --> 00:14:14,480 Speaker 2: he had posted it online with her full name and 245 00:14:14,640 --> 00:14:17,680 Speaker 2: some of her social media information, and she just felt 246 00:14:17,760 --> 00:14:21,720 Speaker 2: like she'd had enough, So in August of twenty twenty one, 247 00:14:22,160 --> 00:14:25,800 Speaker 2: she decided that she wanted to investigate who was behind 248 00:14:25,840 --> 00:14:28,360 Speaker 2: all this and to prove once and for all who 249 00:14:28,400 --> 00:14:32,440 Speaker 2: this cyber harasser was. She'd heard rumors and speculation that 250 00:14:32,480 --> 00:14:35,480 Speaker 2: Patrick Carey was behind it, and she didn't believe that 251 00:14:35,800 --> 00:14:38,440 Speaker 2: he could do something like this, so she wanted to 252 00:14:38,480 --> 00:14:41,720 Speaker 2: not only unmask the predator, but she also wanted to 253 00:14:41,720 --> 00:14:44,840 Speaker 2: clear Carrie's name, and she decided to do this by 254 00:14:45,120 --> 00:14:48,680 Speaker 2: actually doing a deep dive into every single photo the 255 00:14:48,720 --> 00:14:52,400 Speaker 2: poster had uploaded to this website. She studied them. She 256 00:14:52,440 --> 00:14:54,680 Speaker 2: looked into the background of the images to see if 257 00:14:54,720 --> 00:14:57,560 Speaker 2: she could find any evidence that could help her identify 258 00:14:57,640 --> 00:15:00,360 Speaker 2: who this person was, and there was one part taricular 259 00:15:00,440 --> 00:15:03,440 Speaker 2: photo she saw where he had posted an image from 260 00:15:03,480 --> 00:15:06,360 Speaker 2: a little girl's bedroom and he was actually wearing little 261 00:15:06,400 --> 00:15:10,000 Speaker 2: girl's underwear. This told her that he had access to 262 00:15:10,400 --> 00:15:13,920 Speaker 2: a young girl's bedroom, and she looked in the background 263 00:15:13,920 --> 00:15:16,080 Speaker 2: of the image and she saw a white dresser with 264 00:15:16,200 --> 00:15:19,640 Speaker 2: brown trim. She saw a soft toy animal on the beard, 265 00:15:20,400 --> 00:15:23,280 Speaker 2: and she decided that she would then search for Patrick 266 00:15:23,360 --> 00:15:26,520 Speaker 2: Carrey's younger sisters. Having grown up very close to his 267 00:15:26,600 --> 00:15:29,440 Speaker 2: home in attending elementary school with him, she knew that 268 00:15:29,480 --> 00:15:31,840 Speaker 2: he had twin sisters, her a number of years younger, 269 00:15:32,160 --> 00:15:34,200 Speaker 2: and she started looking for them online to see if 270 00:15:34,240 --> 00:15:37,080 Speaker 2: she could find any social media presence, and she actually 271 00:15:37,160 --> 00:15:41,360 Speaker 2: found one of his sisters posting TikTok videos from exactly 272 00:15:41,400 --> 00:15:43,800 Speaker 2: the same bedroom. It had the same dresser with the 273 00:15:43,800 --> 00:15:46,640 Speaker 2: same brown trim, the same soft toy animal on the beard, 274 00:15:46,960 --> 00:15:49,480 Speaker 2: and that's when she realized that a person that she 275 00:15:49,640 --> 00:15:52,600 Speaker 2: believed was a friend of hers, a classmate, someone whod 276 00:15:52,600 --> 00:15:55,840 Speaker 2: she's known for almost her entire life, was behind all this. 277 00:15:58,240 --> 00:16:01,560 Speaker 2: She took that evidence into the police station in late August, 278 00:16:02,000 --> 00:16:04,920 Speaker 2: and it was September fifth that Patrick carry was arrested. 279 00:16:05,440 --> 00:16:08,160 Speaker 2: The lead detective on the case actually went over to 280 00:16:08,360 --> 00:16:11,400 Speaker 2: Carrie's house, knocked on the door with some of the 281 00:16:11,400 --> 00:16:14,360 Speaker 2: print outs of this particular website and some of the 282 00:16:14,400 --> 00:16:17,160 Speaker 2: posts that had been made, and he began reading them 283 00:16:17,200 --> 00:16:19,920 Speaker 2: out aloud to carry who was at home with his mother, 284 00:16:20,320 --> 00:16:22,640 Speaker 2: and he admitted that he was behind it all. 285 00:16:23,160 --> 00:16:28,240 Speaker 4: Prosecutors say Carrie created fourteen different profiles posting these deep 286 00:16:28,320 --> 00:16:32,360 Speaker 4: faked images from August twenty nineteen to September of this year. 287 00:16:32,520 --> 00:16:33,520 Speaker 4: When he was arrested. 288 00:16:34,320 --> 00:16:39,080 Speaker 2: He had been charged with low level harassment violations and obscenity, 289 00:16:39,280 --> 00:16:41,680 Speaker 2: which is a pretty old school charge that was really 290 00:16:41,920 --> 00:16:44,720 Speaker 2: the prosecutors didn't feel like it was going to stick, 291 00:16:45,400 --> 00:16:48,520 Speaker 2: so they had ninety days to raise the stakes. And 292 00:16:48,560 --> 00:16:50,360 Speaker 2: they told me that the case had come through as 293 00:16:50,360 --> 00:16:53,760 Speaker 2: an on call, which in their department at least, that 294 00:16:53,880 --> 00:16:57,160 Speaker 2: meant that the police were hoping the prosecutors would upgrade 295 00:16:57,240 --> 00:17:01,200 Speaker 2: or update or file new charges again to Carrie, because 296 00:17:01,240 --> 00:17:03,240 Speaker 2: they weren't really sure what to charge him with. 297 00:17:03,920 --> 00:17:06,119 Speaker 1: And we should say, you reached out to Patrick Carey 298 00:17:06,119 --> 00:17:09,119 Speaker 1: for comment and he didn't respond to you. You spoke 299 00:17:09,119 --> 00:17:11,520 Speaker 1: to the district attorneys in this case. What did they 300 00:17:11,560 --> 00:17:14,640 Speaker 1: say about what it takes to charge someone for posting 301 00:17:14,720 --> 00:17:15,720 Speaker 1: deep fake photos. 302 00:17:16,680 --> 00:17:20,520 Speaker 2: The two district attorneys that actually handled and investigated this 303 00:17:20,720 --> 00:17:24,440 Speaker 2: case were really surprised when it landed on their desk 304 00:17:24,560 --> 00:17:27,280 Speaker 2: and they realized, you know, the depravity of some of 305 00:17:27,320 --> 00:17:30,600 Speaker 2: the content, just how disgusting some of these posts were, 306 00:17:30,760 --> 00:17:33,080 Speaker 2: and the fact that there were no real charges that 307 00:17:33,119 --> 00:17:36,240 Speaker 2: they could lay against this individual. I remember one of 308 00:17:36,240 --> 00:17:39,240 Speaker 2: the district attorneys telling me that she was scared that 309 00:17:39,800 --> 00:17:42,840 Speaker 2: reading through some of Carrie's posts, she was afraid that 310 00:17:42,960 --> 00:17:47,439 Speaker 2: this behavior could go offline, as in he could commit 311 00:17:47,600 --> 00:17:50,520 Speaker 2: real crimes in the real world, not just as an 312 00:17:50,520 --> 00:17:54,440 Speaker 2: anonymous predator online. So she wanted to ensure that they 313 00:17:54,520 --> 00:17:57,320 Speaker 2: could raise the stakes on the charges that have been 314 00:17:57,359 --> 00:18:00,120 Speaker 2: filed against Carrie. But the challenge they really. 315 00:18:00,560 --> 00:18:02,359 Speaker 3: As there were no charges. 316 00:18:02,560 --> 00:18:05,159 Speaker 2: There was no charge that they could turn to that 317 00:18:05,320 --> 00:18:09,560 Speaker 2: said it was illegal to post fake pornographic content of 318 00:18:09,840 --> 00:18:12,800 Speaker 2: young women on the Internet, so they had to get creative. 319 00:18:13,000 --> 00:18:16,440 Speaker 2: They studied the penal code, They read through every page, 320 00:18:16,680 --> 00:18:20,160 Speaker 2: They looked at every single photo that Carrie had uploaded 321 00:18:20,160 --> 00:18:23,320 Speaker 2: to this website, studied the background of the images, studied 322 00:18:23,320 --> 00:18:27,080 Speaker 2: what was in the images, and finally, on October fifth, 323 00:18:27,200 --> 00:18:29,480 Speaker 2: they found a crack in the case, which is they 324 00:18:29,520 --> 00:18:34,000 Speaker 2: found one real nude image of a fourteen year old 325 00:18:34,080 --> 00:18:39,320 Speaker 2: girl's genitals. That meant that Patrick Carey had posted child 326 00:18:39,480 --> 00:18:43,800 Speaker 2: sexual abuse material online and that was prosecutable. So the 327 00:18:43,920 --> 00:18:47,280 Speaker 2: charges were upgraded to see SAM charges. 328 00:18:47,480 --> 00:18:51,159 Speaker 1: And see SAM here were referring to child sexual abuse materials. 329 00:18:52,520 --> 00:18:55,000 Speaker 1: So marguie briefly, what laws are out there now in 330 00:18:55,040 --> 00:18:58,640 Speaker 1: the US the cover photo sharing, what's protected and what's illegal. 331 00:18:59,480 --> 00:19:03,040 Speaker 3: So that currently no federal law that criminalizes the sharing 332 00:19:03,080 --> 00:19:06,920 Speaker 3: of fake pornographic images in the US. That doesn't mean 333 00:19:07,080 --> 00:19:11,320 Speaker 3: that someone sharing these kind of images won't be prosecuted 334 00:19:11,359 --> 00:19:15,359 Speaker 3: because there are varying state laws under kind of revenge porn. 335 00:19:15,800 --> 00:19:18,560 Speaker 3: There are new bills that are being brought in about 336 00:19:18,640 --> 00:19:22,000 Speaker 3: deep fake images, but there isn't this strict federal law. 337 00:19:22,080 --> 00:19:23,840 Speaker 3: So this is why we had these issues with the 338 00:19:23,880 --> 00:19:28,000 Speaker 3: prosecutors when it comes to children. So child sexual abuse 339 00:19:28,080 --> 00:19:31,600 Speaker 3: material that is banned. So there is a federal law 340 00:19:31,640 --> 00:19:35,800 Speaker 3: saying that you cannot generate child sexual abuse material. However, 341 00:19:36,160 --> 00:19:39,080 Speaker 3: it's a narrow law and the problem is that a 342 00:19:39,119 --> 00:19:44,280 Speaker 3: lot of these ai generations of children are just nude images, 343 00:19:44,520 --> 00:19:47,800 Speaker 3: which you wouldn't prosecute someone who's posted a picture of 344 00:19:47,840 --> 00:19:50,359 Speaker 3: a child in the bath or on the beach, and 345 00:19:50,400 --> 00:19:53,840 Speaker 3: in the same way, there isn't a precedent for prosecuting 346 00:19:53,880 --> 00:19:56,639 Speaker 3: anyone who generates an image of a new child, So 347 00:19:56,720 --> 00:19:59,800 Speaker 3: there has to be a very explicit abuse being to 348 00:20:00,040 --> 00:20:03,439 Speaker 3: picted in the image, which is kind of given a 349 00:20:03,480 --> 00:20:07,239 Speaker 3: lot of people credence to just had the misfortune of 350 00:20:07,560 --> 00:20:10,320 Speaker 3: looking at a lot of forums where people are sharing 351 00:20:10,359 --> 00:20:13,119 Speaker 3: tips on how to generate these kind of images. You 352 00:20:13,160 --> 00:20:15,960 Speaker 3: can see in the comments that they feel quite emboldened 353 00:20:16,000 --> 00:20:19,960 Speaker 3: by this law, so it's still incredibly blurry. A lot 354 00:20:20,000 --> 00:20:22,919 Speaker 3: of people within law enforcement that I spoke to for 355 00:20:23,040 --> 00:20:27,000 Speaker 3: the story all said that it feels really mishmash at 356 00:20:27,000 --> 00:20:30,199 Speaker 3: the moment between state laws and federal laws, and that 357 00:20:30,200 --> 00:20:32,359 Speaker 3: the law really hasn't kept up with the technology in 358 00:20:32,400 --> 00:20:32,879 Speaker 3: this area. 359 00:20:33,760 --> 00:20:36,119 Speaker 1: At the same time the young women and prosecutors in 360 00:20:36,160 --> 00:20:38,760 Speaker 1: New York were trying to track down who was posting 361 00:20:38,840 --> 00:20:42,199 Speaker 1: these deep bakes, there were other people investigating this site 362 00:20:42,240 --> 00:20:42,840 Speaker 1: who were they. 363 00:20:43,920 --> 00:20:47,840 Speaker 3: For a few years actually, there had been a group 364 00:20:47,880 --> 00:20:53,080 Speaker 3: of people who had had similar experiences as those girls 365 00:20:53,119 --> 00:20:57,400 Speaker 3: in Levittown, victims who had had their pictures posted or 366 00:20:57,520 --> 00:21:00,400 Speaker 3: simply just people who had heard of the web site 367 00:21:00,400 --> 00:21:03,440 Speaker 3: and were repulsed by it. And what we've found out 368 00:21:03,680 --> 00:21:08,560 Speaker 3: was one of the key people in investigating the website 369 00:21:08,760 --> 00:21:11,760 Speaker 3: was Will Wallace, who's a former police officer who's living 370 00:21:11,760 --> 00:21:14,919 Speaker 3: in New Zealand and had been working for years on 371 00:21:15,200 --> 00:21:18,760 Speaker 3: trying to uncover some of the users for similar reasons 372 00:21:19,080 --> 00:21:21,439 Speaker 3: as the levitt Town women. 373 00:21:22,160 --> 00:21:24,560 Speaker 2: I think, for me, the craziest part of the reporting 374 00:21:24,600 --> 00:21:27,800 Speaker 2: process of this story was that moment that we realized 375 00:21:27,840 --> 00:21:31,520 Speaker 2: this vigilante movement to shut down come On Printed Picks 376 00:21:32,160 --> 00:21:36,159 Speaker 2: coincided with the investigation that the young woman from Levittown 377 00:21:36,440 --> 00:21:40,440 Speaker 2: were doing to try and identify their harasser. I remember 378 00:21:40,560 --> 00:21:43,480 Speaker 2: I was driving back from Leavittown to New York City. 379 00:21:43,760 --> 00:21:46,480 Speaker 2: It was late at night, it was raining, and I 380 00:21:46,600 --> 00:21:49,320 Speaker 2: called Margie and had her on speakerphone, and I was 381 00:21:49,359 --> 00:21:52,159 Speaker 2: telling her everything I'd learned from an interview, my first 382 00:21:52,160 --> 00:21:55,600 Speaker 2: interview with Anna, where she described how she actually discovered 383 00:21:55,640 --> 00:21:59,160 Speaker 2: Patrick Carey was behind this website. And Margie jumps back 384 00:21:59,200 --> 00:22:02,160 Speaker 2: in and tells me that there's been a vigilante movement 385 00:22:02,359 --> 00:22:06,560 Speaker 2: of trolls and hackers in a former police detective from 386 00:22:06,600 --> 00:22:09,159 Speaker 2: New Zealand, which happens to be where I'm from that 387 00:22:09,280 --> 00:22:11,920 Speaker 2: had been involved in trying to shut this website down 388 00:22:12,080 --> 00:22:14,600 Speaker 2: for the past two or three years. 389 00:22:15,200 --> 00:22:19,199 Speaker 3: He had been tipped off by former police colleagues that 390 00:22:19,320 --> 00:22:21,520 Speaker 3: there was a case they couldn't crack a woman who 391 00:22:21,600 --> 00:22:25,200 Speaker 3: had received years of abuse from people sending her pictures 392 00:22:25,200 --> 00:22:28,679 Speaker 3: of her photos next to their genitalia. They were sending 393 00:22:28,680 --> 00:22:31,359 Speaker 3: it to her parents, to her boyfriend that caused the 394 00:22:31,480 --> 00:22:34,720 Speaker 3: end of her relationship. And after he'd done one case, 395 00:22:35,000 --> 00:22:37,640 Speaker 3: he just got a kind of obsession with the website 396 00:22:37,680 --> 00:22:40,199 Speaker 3: because he couldn't believe that it was still up and running. 397 00:22:41,680 --> 00:22:45,040 Speaker 3: He was adamant that he could find out who was 398 00:22:45,040 --> 00:22:49,800 Speaker 3: behind this file site, and he kept trying to alert journalists, 399 00:22:49,840 --> 00:22:53,480 Speaker 3: he was alerting law enforcement in different parts of the world, 400 00:22:53,920 --> 00:22:58,200 Speaker 3: and he was just getting nowhere. And so after months 401 00:22:58,200 --> 00:23:02,280 Speaker 3: and months of frustration, he decided to look into the 402 00:23:02,320 --> 00:23:06,000 Speaker 3: admin himself to try and put some heat on him, 403 00:23:06,280 --> 00:23:09,480 Speaker 3: and he decided to publish a blog exposing who was 404 00:23:09,520 --> 00:23:14,479 Speaker 3: behind it and detailing his investigation into how he uncovered her. 405 00:23:14,520 --> 00:23:18,480 Speaker 2: It was. We were both just so shocked and surprised, 406 00:23:18,520 --> 00:23:21,280 Speaker 2: because in journalism you always hope that the stars are 407 00:23:21,280 --> 00:23:23,160 Speaker 2: going to a line, and that you can have an 408 00:23:23,200 --> 00:23:27,080 Speaker 2: intersecting narrative of two different stories happening at the same time, 409 00:23:27,160 --> 00:23:30,320 Speaker 2: but it never really happens, and it did in this story. 410 00:23:30,640 --> 00:23:33,760 Speaker 2: We could explain to readers not only what happened in 411 00:23:33,840 --> 00:23:37,360 Speaker 2: Leavettown in this one very specific case, but the broader 412 00:23:37,400 --> 00:23:41,200 Speaker 2: global picture of a group of online vigilantes who were 413 00:23:41,240 --> 00:23:45,359 Speaker 2: targeting this website, joining forces digitally from around the world 414 00:23:45,560 --> 00:23:46,440 Speaker 2: to shut it down. 415 00:23:47,240 --> 00:23:50,359 Speaker 1: And you spoke to Will Wallace about the apparent administrator 416 00:23:50,440 --> 00:23:53,879 Speaker 1: of the site, scatran Costa, and what Will did to 417 00:23:53,960 --> 00:23:55,400 Speaker 1: try to get the sait taken down. 418 00:23:55,640 --> 00:23:59,040 Speaker 5: Here's well, it wasn't likely that law enforcement were going 419 00:23:59,080 --> 00:24:01,520 Speaker 5: to intervene at that. It was pretty much sort of 420 00:24:01,600 --> 00:24:04,679 Speaker 5: chasing Scott round the Internet and finding out who the 421 00:24:04,680 --> 00:24:08,760 Speaker 5: providers were and reporting him to the providers. There's so 422 00:24:08,800 --> 00:24:12,000 Speaker 5: many services at someone's fingertips that it is quite easy, 423 00:24:12,160 --> 00:24:14,359 Speaker 5: particularly if you have the skills and knowledge, just to 424 00:24:14,760 --> 00:24:17,439 Speaker 5: shift to a different provider. That is one of the 425 00:24:17,600 --> 00:24:20,560 Speaker 5: issues where it's kind of people on the Internet having 426 00:24:20,760 --> 00:24:22,840 Speaker 5: a go at getting rid of a website, is that 427 00:24:22,920 --> 00:24:25,440 Speaker 5: you're always going to be doing that, Whereas with law 428 00:24:25,520 --> 00:24:28,400 Speaker 5: enforcement intervention, they have a lot more levers they can pull. 429 00:24:29,080 --> 00:24:30,800 Speaker 5: I don't know if it will end. I don't know 430 00:24:30,800 --> 00:24:33,479 Speaker 5: if the service will ever get taken down, but at 431 00:24:33,560 --> 00:24:36,239 Speaker 5: least I feel as though I've done my bit, and 432 00:24:36,280 --> 00:24:39,880 Speaker 5: that's to create more accountability. 433 00:24:40,359 --> 00:24:43,800 Speaker 1: After the break, the outcome of the Levittown case and 434 00:24:44,240 --> 00:24:57,479 Speaker 1: why AI technology keeps out pacing regulation, Olivia. In an 435 00:24:57,560 --> 00:25:00,199 Speaker 1: interview later on with one of New York's das, you 436 00:25:00,240 --> 00:25:02,480 Speaker 1: found out that there was more to the story with 437 00:25:02,600 --> 00:25:04,440 Speaker 1: Carrie's deep fakes. What happened? 438 00:25:05,240 --> 00:25:09,440 Speaker 2: Yeah, So this interview was done in the Nassau County Courthouse. 439 00:25:09,520 --> 00:25:11,840 Speaker 2: I was talking to both of the district attorneys who 440 00:25:11,840 --> 00:25:15,439 Speaker 2: had been involved and investigated this case, Kelsey Laura and 441 00:25:15,560 --> 00:25:19,800 Speaker 2: Melissa Scannell, and we were talking through their investigative process 442 00:25:20,040 --> 00:25:24,040 Speaker 2: and how they came to increase the charges against Carrie, 443 00:25:24,320 --> 00:25:28,119 Speaker 2: how they investigated what really happened, and I recall towards 444 00:25:28,160 --> 00:25:30,760 Speaker 2: the end of the interview, I asked them if they'd 445 00:25:30,800 --> 00:25:35,159 Speaker 2: ever contacted the administrator or owner of the website or 446 00:25:35,200 --> 00:25:38,160 Speaker 2: tried to get the girl's photos removed, and they told 447 00:25:38,200 --> 00:25:42,399 Speaker 2: me that after his conviction, there was one final task 448 00:25:42,480 --> 00:25:44,840 Speaker 2: they wanted to do, which was to reach out to 449 00:25:44,880 --> 00:25:48,159 Speaker 2: the person behind this horrific website and ask them to 450 00:25:48,200 --> 00:25:51,600 Speaker 2: remove all photos and all posts that Carrie had ever 451 00:25:51,680 --> 00:25:55,000 Speaker 2: made associated with his usernames. They were going to do 452 00:25:55,080 --> 00:25:59,120 Speaker 2: that by sending the administrator Carrie's certificate of conviction and 453 00:25:59,160 --> 00:26:03,520 Speaker 2: saying these hosts resulted in criminal conduct, please remove them all. 454 00:26:03,880 --> 00:26:06,760 Speaker 2: But when they went to do that, the website was offline. 455 00:26:06,800 --> 00:26:10,600 Speaker 2: So just days earlier, William Wallace and the Vigilantes mission 456 00:26:10,800 --> 00:26:14,040 Speaker 2: had been successful when this website had gone dark, so 457 00:26:14,280 --> 00:26:16,720 Speaker 2: the district attorneys felt like there wasn't really any more 458 00:26:16,760 --> 00:26:19,359 Speaker 2: that they needed to do. The photos had been removed, 459 00:26:19,400 --> 00:26:22,679 Speaker 2: the posts were no longer there. But during that interview 460 00:26:22,840 --> 00:26:26,879 Speaker 2: I informed them that actually this website had moved to 461 00:26:26,960 --> 00:26:30,600 Speaker 2: a different domain, a different URL, and all of the 462 00:26:30,600 --> 00:26:34,600 Speaker 2: posts of the Levittown Girls were back up online and visible. 463 00:26:35,080 --> 00:26:39,479 Speaker 2: And I remember Melissagists her jaw dropped, and she was 464 00:26:40,280 --> 00:26:44,879 Speaker 2: so angry and also shocked and confused at the same time. 465 00:26:45,200 --> 00:26:49,400 Speaker 2: And she was speaking very eloquently and calmly throughout the interview, 466 00:26:49,480 --> 00:26:51,919 Speaker 2: and she just stopped and said, what what do you 467 00:26:52,040 --> 00:26:53,080 Speaker 2: mean They're still there? 468 00:26:55,480 --> 00:26:58,360 Speaker 1: And Marguie, did anyone associated with running any of these 469 00:26:58,400 --> 00:27:00,640 Speaker 1: sites respond to you when you can't them. 470 00:27:01,400 --> 00:27:04,640 Speaker 3: They did after a long while. We did a lot 471 00:27:04,680 --> 00:27:08,680 Speaker 3: to try and contact Scott Trent Coster. It took a 472 00:27:08,720 --> 00:27:12,879 Speaker 3: few months and eventually they responded to a kind of 473 00:27:12,960 --> 00:27:15,439 Speaker 3: long email where we asked a lot of questions and 474 00:27:15,560 --> 00:27:17,480 Speaker 3: laid out the story and that we were going to 475 00:27:17,560 --> 00:27:22,119 Speaker 3: name them, and they called me a weirdo, which is 476 00:27:22,160 --> 00:27:25,280 Speaker 3: one of the strangest responses I've ever had. They pointed 477 00:27:25,320 --> 00:27:27,600 Speaker 3: out that it was very strange for me to want 478 00:27:27,600 --> 00:27:32,480 Speaker 3: to name anyone and that I should not be focusing 479 00:27:32,480 --> 00:27:36,240 Speaker 3: my time on exposing someone's identity. We know that the 480 00:27:36,280 --> 00:27:39,760 Speaker 3: admin has spoken to some of the vigilantes before and 481 00:27:39,880 --> 00:27:44,440 Speaker 3: they have had similar responses, very odd responses like why 482 00:27:44,480 --> 00:27:47,920 Speaker 3: are you trying to ruin my business for me? You 483 00:27:47,960 --> 00:27:50,560 Speaker 3: should just leave us alone? What's wrong with us having 484 00:27:50,560 --> 00:27:51,440 Speaker 3: our kinks? 485 00:27:52,040 --> 00:27:54,280 Speaker 1: And Olivia, where is Patrick carry now? 486 00:27:55,119 --> 00:27:58,480 Speaker 2: He was sentenced to six months in prison, ten years 487 00:27:58,560 --> 00:28:02,600 Speaker 2: probation in lifetime sentence as a sex offender. This means 488 00:28:02,640 --> 00:28:05,320 Speaker 2: he can't own a smartphone, he can't own a device 489 00:28:05,359 --> 00:28:07,800 Speaker 2: with a camera, He can't live within a thousand feet 490 00:28:07,960 --> 00:28:10,119 Speaker 2: or be within a thousand feet of a school or 491 00:28:10,160 --> 00:28:14,040 Speaker 2: a playground or a place where children are, and surprisingly, 492 00:28:14,080 --> 00:28:16,680 Speaker 2: in August I found out he'd actually been released after 493 00:28:16,760 --> 00:28:19,440 Speaker 2: four months with time off for good behavior, and he's 494 00:28:19,480 --> 00:28:22,720 Speaker 2: back in Livetown. One of the young women who open 495 00:28:22,800 --> 00:28:26,560 Speaker 2: our story Cecilia Luce. She was driving along the street 496 00:28:26,720 --> 00:28:30,600 Speaker 2: and she saw him walking down one of Levittown's main drags. 497 00:28:30,640 --> 00:28:33,040 Speaker 2: He had on a brown hoodie, he had headphones in, 498 00:28:33,600 --> 00:28:36,800 Speaker 2: and she started immediately crying. She was so scared to 499 00:28:36,840 --> 00:28:40,520 Speaker 2: see him, And then as she drove on, she felt anger, 500 00:28:41,000 --> 00:28:44,200 Speaker 2: and she decided to stop and turn her car round 501 00:28:44,280 --> 00:28:46,600 Speaker 2: and go back. She wanted to confront him, she wanted 502 00:28:46,600 --> 00:28:48,680 Speaker 2: to yell at him. She had been a close friend 503 00:28:48,680 --> 00:28:51,200 Speaker 2: of Patrick Carrey's throughout high school. So she did a 504 00:28:51,280 --> 00:28:53,520 Speaker 2: U turn, turned to her car, back around, and he 505 00:28:53,520 --> 00:28:56,600 Speaker 2: had disappeared. She carried on driving for a number of minutes, 506 00:28:56,600 --> 00:28:58,520 Speaker 2: looking down all the side streets to see where he 507 00:28:58,560 --> 00:29:01,600 Speaker 2: would have gone, but she never saw him again. At 508 00:29:01,640 --> 00:29:04,280 Speaker 2: is sentencing, one of the young women I talked to 509 00:29:04,520 --> 00:29:07,520 Speaker 2: actually read out a victim impact statement. Her name is 510 00:29:07,600 --> 00:29:10,880 Speaker 2: Kayla Michelle, and she was the only young woman who 511 00:29:11,480 --> 00:29:16,040 Speaker 2: testified at his sentencing. She described attending every single hearing, 512 00:29:16,160 --> 00:29:19,400 Speaker 2: relating to this case. I think for a few reasons. 513 00:29:19,600 --> 00:29:22,000 Speaker 2: One of them is because she'd known Carrie since she 514 00:29:22,120 --> 00:29:23,960 Speaker 2: was eight years old, she thought he was a friend 515 00:29:24,000 --> 00:29:26,760 Speaker 2: of hers. And another reason is that she was just 516 00:29:26,880 --> 00:29:30,520 Speaker 2: so angry at him and she wanted to ensure that 517 00:29:30,720 --> 00:29:34,400 Speaker 2: justice was going to be served. So during my interview 518 00:29:34,440 --> 00:29:37,160 Speaker 2: with Kayla, who is now twenty three and still lives 519 00:29:37,160 --> 00:29:40,720 Speaker 2: in Leavertown, we talked about the moment that she actually 520 00:29:41,320 --> 00:29:44,200 Speaker 2: had the courage and bravery to read out this victim 521 00:29:44,240 --> 00:29:46,960 Speaker 2: impact statement in court, and she described to me how 522 00:29:47,200 --> 00:29:50,760 Speaker 2: nervous she was, that she'd written down what she wanted 523 00:29:50,800 --> 00:29:53,120 Speaker 2: to say and the piece of paper was shaking in 524 00:29:53,160 --> 00:29:56,080 Speaker 2: her hands, that she had to take three deep breaths 525 00:29:56,160 --> 00:29:58,680 Speaker 2: before she was strong enough to talk, and that she 526 00:29:58,760 --> 00:30:02,080 Speaker 2: looked directly at Patrick Carry in court and he refused 527 00:30:02,080 --> 00:30:04,920 Speaker 2: to make eye contact with her. I reached out to 528 00:30:04,960 --> 00:30:06,840 Speaker 2: one of the court reporters who were in court that 529 00:30:06,920 --> 00:30:09,040 Speaker 2: day in order to get a full transcript of what 530 00:30:09,120 --> 00:30:12,840 Speaker 2: Kayla said, and this was what was read out in court. 531 00:30:14,320 --> 00:30:17,400 Speaker 2: Kayla said, I'm the strongest person in the room because 532 00:30:17,400 --> 00:30:19,760 Speaker 2: I'm looking you directly in the face to tell you 533 00:30:19,800 --> 00:30:22,960 Speaker 2: that you discussed me. You hurt me, but you've also 534 00:30:23,120 --> 00:30:25,880 Speaker 2: changed me. You had the audacity to talk to me 535 00:30:25,960 --> 00:30:28,920 Speaker 2: through social media, joke with me, and try to be 536 00:30:29,000 --> 00:30:32,840 Speaker 2: cordial with me while behind my back belittling me, putting 537 00:30:32,880 --> 00:30:37,440 Speaker 2: me down, sexualizing my younger self and body. You completely 538 00:30:37,520 --> 00:30:40,160 Speaker 2: change the way that I viewed myself and my body, 539 00:30:40,440 --> 00:30:43,200 Speaker 2: and for that, I'll never forgive you. Your name will 540 00:30:43,240 --> 00:30:46,200 Speaker 2: forever give me nightmares and haunt me. I hear your 541 00:30:46,280 --> 00:30:47,560 Speaker 2: name and I feel sick. 542 00:30:50,000 --> 00:30:53,280 Speaker 1: Marguie, What is needed legally to prevent the further spread 543 00:30:53,280 --> 00:30:55,680 Speaker 1: of these sites and AI generated photos? 544 00:30:56,400 --> 00:30:58,840 Speaker 3: That is just such a hard question. You know, we're 545 00:30:58,840 --> 00:31:01,360 Speaker 3: talking about this story. This isn't the first, and I'm 546 00:31:01,440 --> 00:31:03,560 Speaker 3: it's the first we're hearing of, but I think there'll 547 00:31:03,600 --> 00:31:07,920 Speaker 3: be many similar stories like this because the law is 548 00:31:08,400 --> 00:31:12,160 Speaker 3: tricky and we're really catching up with what technology is 549 00:31:12,160 --> 00:31:15,640 Speaker 3: able to do. And also there is an issue with 550 00:31:16,040 --> 00:31:20,000 Speaker 3: who is liable who is accountable for this, So it's 551 00:31:20,040 --> 00:31:23,520 Speaker 3: not just the users of the technology who are generating 552 00:31:23,560 --> 00:31:27,360 Speaker 3: these images. They then have the ability to post them online, 553 00:31:27,920 --> 00:31:31,640 Speaker 3: and then the owners of these websites aren't really liable 554 00:31:31,800 --> 00:31:35,240 Speaker 3: under Section two thirty, which means that as a a 555 00:31:35,280 --> 00:31:39,400 Speaker 3: website or blog. If a user is posting something that's criminal, 556 00:31:39,400 --> 00:31:41,320 Speaker 3: they're not going to be held liable for it, but 557 00:31:41,440 --> 00:31:45,800 Speaker 3: they're enabling this whole forum for thousands of images to 558 00:31:45,800 --> 00:31:46,520 Speaker 3: be uploaded. 559 00:31:47,400 --> 00:31:50,000 Speaker 2: I think that the reason why their question is so 560 00:31:50,160 --> 00:31:53,960 Speaker 2: hard goes back to their age old Silicon Valley mantra 561 00:31:54,280 --> 00:31:57,600 Speaker 2: of move fast and break things, where this technology was 562 00:31:57,680 --> 00:32:03,120 Speaker 2: rolled out before legislators, regulators, the public really understood its power. 563 00:32:03,800 --> 00:32:06,960 Speaker 2: So now as we're seeing states trying to catch up 564 00:32:07,200 --> 00:32:11,880 Speaker 2: to these new tools, they're creating new laws that As 565 00:32:12,000 --> 00:32:15,040 Speaker 2: Maggie and I reported, we looked across the United States 566 00:32:15,080 --> 00:32:17,960 Speaker 2: and saw all the different portions of legislation that have 567 00:32:18,040 --> 00:32:21,080 Speaker 2: been created to try and target deep fakes. But we've 568 00:32:21,120 --> 00:32:24,320 Speaker 2: got a patchwork across the country now. Some states have 569 00:32:24,440 --> 00:32:29,000 Speaker 2: civil statutes, some have criminal statutes, Some only target election 570 00:32:29,280 --> 00:32:32,080 Speaker 2: related deep fake content, which is a subject we haven't 571 00:32:32,080 --> 00:32:35,360 Speaker 2: even had time to address on today's podcast. But other 572 00:32:35,440 --> 00:32:40,280 Speaker 2: statutes are just amendments to existing revenge porn laws. So 573 00:32:40,720 --> 00:32:43,479 Speaker 2: no one's really figured out the right way to solve 574 00:32:43,520 --> 00:32:47,280 Speaker 2: this problem yet. In October, we saw President Biden announce 575 00:32:47,320 --> 00:32:51,840 Speaker 2: a new executive order to target artificial intelligence and ensure 576 00:32:51,840 --> 00:32:54,680 Speaker 2: that it was illegal that you couldn't use these tools 577 00:32:54,720 --> 00:32:59,200 Speaker 2: to create child sexual abuse imagery, child porn, or intimate 578 00:32:59,240 --> 00:33:02,960 Speaker 2: imagery of individuals doing things that they never really did. 579 00:33:03,520 --> 00:33:04,680 Speaker 1: And we don't. 580 00:33:04,480 --> 00:33:06,880 Speaker 2: Actually know when these are going to be implemented or 581 00:33:06,920 --> 00:33:08,920 Speaker 2: how they are going to be implemented. You know, that's 582 00:33:08,960 --> 00:33:10,240 Speaker 2: the big question right now. 583 00:33:10,960 --> 00:33:13,440 Speaker 1: Olivia Marghi, thank you so much for joining. 584 00:33:13,200 --> 00:33:15,920 Speaker 2: Us, Thanks for having us, Thank you so much. 585 00:33:17,200 --> 00:33:19,320 Speaker 1: Thanks for listening to us here at The Big Take. 586 00:33:19,560 --> 00:33:23,480 Speaker 1: It's a daily podcast from Bloomberg and iHeartRadio. For more 587 00:33:23,520 --> 00:33:29,160 Speaker 1: shows from iHeartRadio, visit the iHeartRadio app, Apple Podcasts, Bloomberg CarPlay, 588 00:33:29,440 --> 00:33:32,239 Speaker 1: or wherever you listen. And we'd love to hear from you. 589 00:33:32,600 --> 00:33:35,760 Speaker 1: Email us with questions or comments to Big Take at 590 00:33:35,760 --> 00:33:41,040 Speaker 1: Bloomberg dot net. Our supervising producer is Vicky Vergalina. Our 591 00:33:41,080 --> 00:33:45,160 Speaker 1: senior producer is Katherine Fink. Our producers are Michael Falerro 592 00:33:45,240 --> 00:33:48,760 Speaker 1: and Moe Barrow. Raphael I'm see Lee is our engineer. 593 00:33:49,160 --> 00:33:52,920 Speaker 1: Our original music is by Leo Sidron. I'm Nancy Cook. 594 00:33:53,080 --> 00:33:55,240 Speaker 1: We'll be back tomorrow with another Big Take 595 00:34:00,120 --> 00:34:03,280 Speaker 3: The Dot Bun, the Don Staple Bar