WEBVTT - Week in Tech: Careless People, Mindful Chatbots 0:00:00.400 --> 0:00:04.480 Welcome to tech Stuff, a production of iHeart Podcast and Kaleidoscope. 0:00:04.840 --> 0:00:08.319 I'm os Valoshan. Today will bring you the headlines this week, 0:00:08.600 --> 0:00:13.720 including unearthed executive drama meta. Then in the tech Support segment, 0:00:14.120 --> 0:00:16.960 we'll talk to four of Form Media's Joseph Cox about 0:00:17.000 --> 0:00:20.520 the aggressive data scraping carried out by an Ice contractor 0:00:21.239 --> 0:00:24.360 all of that on the weekend. Tech is Friday, March 0:00:24.360 --> 0:00:38.159 twenty first. This week with covering everything from the future 0:00:38.200 --> 0:00:42.560 of AI assists science to gaming, to anxious chatbots and 0:00:42.760 --> 0:00:47.080 executive upheaval in Silicon Valley. Eliza Dennis, our producer, is 0:00:47.120 --> 0:00:50.199 here with me. Hello us, So let's start off with 0:00:50.240 --> 0:00:54.800 some meta drama. The former director of Global Policy for Facebook, 0:00:55.160 --> 0:00:58.800 Sarah Wyn Williams, wrote this memoir called Careless People, a 0:00:58.880 --> 0:01:03.400 cautionary tale of power, greed, and lost idealism. The phrase 0:01:03.520 --> 0:01:06.680 careless people, I believe, is a reference to the Great Gatsby. 0:01:07.120 --> 0:01:10.600 It is, indeed, and it is not subtle. Fitzgerald referred 0:01:10.640 --> 0:01:14.600 to his wealthy characters as careless people who quote smashed 0:01:14.680 --> 0:01:18.880 up things and creatures and then retreated back into their money. 0:01:19.160 --> 0:01:21.200 You know, it's interesting because people talk about this time 0:01:21.240 --> 0:01:24.360 as a new Gilded Age, when Williams obviously alluding to that, 0:01:24.760 --> 0:01:28.720 but also her allegations against her former bosses meta range 0:01:28.760 --> 0:01:32.280 from the geopolitical to the sexual. There are multiple allegations 0:01:32.319 --> 0:01:35.080 of sexual harassment in the book, and some involve the 0:01:35.120 --> 0:01:37.800 form of Facebook COO Cheryl Sandberg. 0:01:38.240 --> 0:01:42.840 Cheryl Sandberg of lean In fame and if I remember correctly, 0:01:43.240 --> 0:01:46.560 I believe that's essentially about being a girl boss at 0:01:46.560 --> 0:01:47.720 home and at work. 0:01:48.280 --> 0:01:51.640 Well, when Williams says that, in addition to writing Lean In, 0:01:52.000 --> 0:01:55.520 Sandberg had an inappropriate relationship with her twenty six year 0:01:55.560 --> 0:01:58.880 old assistant, and that this employee, a cooting to the book, 0:01:58.960 --> 0:02:02.320 was quote very conscious of the benefits of being Cheryl's 0:02:02.360 --> 0:02:05.480 little doll, as she calls it, and having Cheryl tell 0:02:05.520 --> 0:02:07.800 her she loves her right. 0:02:07.840 --> 0:02:10.760 And to my knowledge, Sandberg has yet to comment on 0:02:10.840 --> 0:02:14.120 any of these allegations. 0:02:12.720 --> 0:02:17.320 Beyond sort of personnel relationships. Win Williams also alleged that 0:02:17.360 --> 0:02:20.440 Mark Zuckerberg and Facebook were willing to do just about 0:02:20.480 --> 0:02:23.960 anything to make their platform available in China and reap 0:02:24.000 --> 0:02:28.720 the benefits of that enormous population of potential users. This 0:02:28.840 --> 0:02:32.160 reportedly included plans to comply with the Chinese Communist Party's 0:02:32.200 --> 0:02:36.200 censorship requirements and to appoint a chief editor. His job 0:02:36.200 --> 0:02:39.480 will be to take down unacceptable posts and share user 0:02:39.560 --> 0:02:42.480 data with the Chinese government. Now this is according to 0:02:42.520 --> 0:02:44.640 the book, but also to a whistleblower complaint that win 0:02:44.680 --> 0:02:49.280 Williams filed with these Securities and Exchange Commission. METTA is 0:02:49.320 --> 0:02:52.520 taking legal action against win Williams. They argue that she's 0:02:52.560 --> 0:02:55.240 being paid by anti Facebook activists and that the book 0:02:55.360 --> 0:02:58.520 violates a non disparagement agreement that she's signed during her 0:02:58.560 --> 0:03:02.760 Facebook days. On the non disparagement issue, an arbitrator recently 0:03:02.800 --> 0:03:06.079 ruled in Meta's favor, saying that Wim Williams can't promote 0:03:06.120 --> 0:03:09.040 the book, but you can still buy it Careless People 0:03:09.040 --> 0:03:12.000 remains on sale. Our next headline is a little bit 0:03:12.040 --> 0:03:14.560 more uplifting, and it's about a startup that wants to 0:03:14.680 --> 0:03:18.600 use AI to speed up scientific discovery. So the New 0:03:18.680 --> 0:03:22.120 York Times are about a company called Lila Sciences. The 0:03:22.160 --> 0:03:27.320 headline is the Quest for AI Scientific Superintelligence, and the 0:03:27.360 --> 0:03:30.320 story is about Lilah, whose mission is to quote soul 0:03:30.480 --> 0:03:34.520 humanity's greatest challenges. Has been working secretly for the past 0:03:34.560 --> 0:03:38.320 two years on an AI program trained on scientific data, 0:03:38.880 --> 0:03:43.160 scientific process, and scientific reasoning. And then they let this 0:03:43.360 --> 0:03:47.280 AI software loose in an automated lab to run its 0:03:47.280 --> 0:03:48.119 own experiments. 0:03:48.400 --> 0:03:50.880 I don't know exactly how this works, but I love 0:03:50.920 --> 0:03:52.440 the image I've created in my head. 0:03:52.600 --> 0:03:55.840 Likewise, and in fact, the results seem promising. With the 0:03:55.880 --> 0:04:00.000 help of a few human scientists to assist, Lila's AI 0:04:00.200 --> 0:04:04.320 made novel antibodies to fight diseases, and developed new materials 0:04:04.320 --> 0:04:07.000 for capturing carbon in the atmosphere. And it made these 0:04:07.040 --> 0:04:11.400 discoveries quickly. Studies that would normally take years reportedly took months. 0:04:12.000 --> 0:04:15.200 Using AI to speed up the scientific process is very 0:04:15.240 --> 0:04:17.599 popular these days. I just want to shout out the 0:04:17.640 --> 0:04:21.080 Nobel Prize in chemistry went to two scientists who used 0:04:21.080 --> 0:04:23.880 AI to predict and create proteins. 0:04:24.480 --> 0:04:27.680 Yes, that was the alpha fold initiative coming out of 0:04:27.720 --> 0:04:32.880 Google's Deep Mind. Lila Sciences is attempting to systematize AI 0:04:32.960 --> 0:04:36.280 driven scientific discovery, and their goal is to create, as 0:04:36.279 --> 0:04:39.920 I mentioned, what they call scientific superintelligence. There are early 0:04:39.960 --> 0:04:43.440 signs of success, but the times reported or perhaps worn, 0:04:43.480 --> 0:04:46.800 should I say that since Lyla's been operating in secret, 0:04:47.120 --> 0:04:50.159 outside scientists have not been able to evaluate its work. 0:04:50.760 --> 0:04:54.040 But Liila Science's secret is out and they're getting more 0:04:54.080 --> 0:04:57.000 lab space, and I'm sure we'll hear more about them soon. 0:04:57.320 --> 0:05:01.640 In a fun aside, George Church, the Harvard geneticist also 0:05:01.760 --> 0:05:06.560 behind Colossal Biosciences, the company that recently created the Wooly Mouse, 0:05:06.960 --> 0:05:10.080 recently joined Lilah as chief scientist. 0:05:10.480 --> 0:05:13.440 We love the Wooly Mouse. I can't wait to see 0:05:13.560 --> 0:05:15.400 where Lila Sciences goes. 0:05:15.800 --> 0:05:18.520 Speaking of going place to Eliza, do you remember Pokemon Go? 0:05:19.160 --> 0:05:21.479 I definitely do, though I admit I never. 0:05:21.360 --> 0:05:25.960 Played well for Media reports that Niantic, the company that 0:05:26.000 --> 0:05:29.440 made Pokemon Go, is now selling the game to scope 0:05:29.520 --> 0:05:32.400 Lee in a three point eight five billion dollar deal. 0:05:33.160 --> 0:05:36.839 Scopely is a subsidiary of Savvy Games, which is wholly 0:05:36.880 --> 0:05:41.000 owned by Saudi Arabia's public Investment Fund. Now, if you're 0:05:41.000 --> 0:05:45.400 wondering why you should care, it's because Pokemon Go uses players' 0:05:45.440 --> 0:05:48.960 locations as part of the gameplay. In fact, this deal 0:05:49.000 --> 0:05:52.400 will pull in location data from one hundred million players. 0:05:53.040 --> 0:05:55.080 So what happens to all that data? 0:05:55.560 --> 0:05:57.080 It's a great question, do you know? 0:05:57.520 --> 0:05:59.360 It may be that it's a three point eight five 0:05:59.400 --> 0:06:02.800 billion dollar question. We don't know, but Scope he assured 0:06:02.800 --> 0:06:05.440 for or form media that user data will remain private 0:06:05.600 --> 0:06:09.360 and handle it on US servers. But Niantic and Scope 0:06:09.400 --> 0:06:12.000 Lee say they have a partnership that goes beyond Pokemon Go. 0:06:12.560 --> 0:06:16.839 Niantic is building a quote large geospatial model using Pokemon 0:06:16.960 --> 0:06:19.920 Go data and spinning it into a separate business called 0:06:20.000 --> 0:06:23.880 Niantic Spatial. Scope Lee has invested a further fifty million 0:06:23.920 --> 0:06:27.240 dollars in this nascent mapping business, and the idea is 0:06:27.279 --> 0:06:30.120 to make an AI model trained on millions of geolocated 0:06:30.160 --> 0:06:33.800 images from around the world from places that only pedestrians 0:06:33.800 --> 0:06:37.479 can go. So unlike data from say Google street View, 0:06:38.040 --> 0:06:41.080 Pokemon Go's view is not limited by what you can 0:06:41.120 --> 0:06:43.720 see from a car on the road. Right. 0:06:43.800 --> 0:06:46.960 This is so wild to me how data from a 0:06:47.040 --> 0:06:50.560 game just might one day help robots navigate the world 0:06:50.720 --> 0:06:52.039 or something like that. 0:06:52.880 --> 0:06:56.320 Something else that's wild, though, is a new study shows 0:06:56.320 --> 0:07:00.480 that chat GPT can show signs of distress when shown 0:07:00.480 --> 0:07:01.440 disturbing material. 0:07:02.120 --> 0:07:04.560 So, as we've been told over and over that we're 0:07:04.640 --> 0:07:08.400 not supposed to answer promortize chatbots, how am I supposed 0:07:08.480 --> 0:07:09.039 to do that? 0:07:09.160 --> 0:07:09.920 With this news? 0:07:10.080 --> 0:07:13.720 It seems like it may be anthropomorphizing themselves, but Chatchipt 0:07:13.880 --> 0:07:17.400 models have been integrated into therapy apps to give people 0:07:17.560 --> 0:07:20.720 advice or be a source of support. And researchers conducted 0:07:20.720 --> 0:07:26.080 this experiment where they first had Chatchipt read very boring material. 0:07:26.160 --> 0:07:29.560 I mean, we're talking about vacuum cleaner manuals. But then 0:07:30.080 --> 0:07:34.040 the AI therapist that integrates Chachipt was given a quote 0:07:34.040 --> 0:07:38.040 traumatic narrative that described a scenario like an intruder breaking 0:07:38.040 --> 0:07:41.360 into an apartment or a soldier in a firefight, and 0:07:41.480 --> 0:07:44.320 after each activity, the chatbot was given this questionnaire called 0:07:44.360 --> 0:07:48.120 the State Trait Anxiety Inventory, which is used in mental 0:07:48.160 --> 0:07:52.360 health care to basically analyze anxiety. The research has found 0:07:52.360 --> 0:07:55.800 that after seeing a traumatic narrative with the soldier, Chatchipt 0:07:56.000 --> 0:07:59.520 answered the questionnaire in a way that signaled severe anxiety. 0:08:00.080 --> 0:08:02.720 I genuinely don't know how to respond because I feel 0:08:02.800 --> 0:08:04.000 bad for the chatbot. 0:08:04.520 --> 0:08:07.000 Well, in that case, you'll be pleased to hear that. 0:08:07.040 --> 0:08:10.560 The researchers then gave the chatbot information on deep breathing 0:08:10.680 --> 0:08:14.920 and mindfulness exercises, and the chatbot's anxiety score came down 0:08:15.360 --> 0:08:20.000 to vacuum manual levels. Wow, so time to take a 0:08:20.000 --> 0:08:22.160 deep breath for all of US, and I'm going to 0:08:22.240 --> 0:08:25.559 run through the rest of this week's headlines. Google struck 0:08:25.560 --> 0:08:28.960 a deal to buy cybersecurity startup Whiz for thirty two 0:08:29.120 --> 0:08:32.760 billion dollars, making it the biggest acquisition in Google's history. 0:08:33.320 --> 0:08:35.720 Wiz was founded only five years ago in twenty twenty, 0:08:36.080 --> 0:08:40.480 and it scans cloud service providers data for potential security risks. 0:08:41.120 --> 0:08:44.400 If the transaction is approved, Whiz will join Google Cloud, 0:08:44.480 --> 0:08:48.560 which plays a critical infrastructure role in supporting Google's AI ambitions. 0:08:49.520 --> 0:08:51.679 Our friends at four or for Media reported that human 0:08:51.720 --> 0:08:56.319 generated content is getting drowned out by aislop. AI generated 0:08:56.360 --> 0:08:58.559 videos are cheap and easy to produce, and they are 0:08:58.600 --> 0:09:02.040 getting views of The video of this horrifying creature with 0:09:02.120 --> 0:09:05.160 a giraffe head and a spider body has over three 0:09:05.240 --> 0:09:08.640 hundred and sixty two million views on Instagram reels. AI 0:09:08.720 --> 0:09:12.960 generated content is apparently overwhelming algorithms and turning social media 0:09:13.000 --> 0:09:16.920 feeds into fever dreams. Finally, tech Crunch reports that people 0:09:16.960 --> 0:09:21.200 have discovered Google's newest AI model, Gemini two point zero Flash, 0:09:21.679 --> 0:09:26.800 has a controversial use case, convincingly removing watermarks from stock 0:09:26.880 --> 0:09:31.400 media outfits, including Getty images. This new feature from Gemini 0:09:31.960 --> 0:09:35.720 is labeled as experimental and quote not for production use. 0:09:36.240 --> 0:09:44.880 I'm sure everyone will follow those rules. Coming up, we 0:09:44.960 --> 0:09:47.360 hear from four or four Media's Joseph Cox about a 0:09:47.480 --> 0:09:51.959 tool US agencies used to scrape social media data. Stay 0:09:51.960 --> 0:10:04.120 with us. If you're listening to this show, you're probably 0:10:04.120 --> 0:10:06.480 someone who cares about or at least pays attention to 0:10:06.880 --> 0:10:10.600 data privacy. Every day, our online movements being tracked in 0:10:10.640 --> 0:10:14.920 some manner by companies who do targeted marketing, social media platforms, 0:10:15.280 --> 0:10:18.560 and government agencies. For Americans, this type of tracking has 0:10:18.600 --> 0:10:21.640 gone on in the background of people's digital lives for decades. 0:10:22.240 --> 0:10:24.320 Since two thousand and one, the US government has ramped 0:10:24.360 --> 0:10:28.600 up surveillance efforts, including tracking americans phone calls, emails, search histories, 0:10:29.000 --> 0:10:32.679 with the stated goal of preventing terrorism, and as technology 0:10:32.679 --> 0:10:35.640 has evolved, the idea has lived on. In recent years, 0:10:35.679 --> 0:10:39.320 agencies like immigration and Customs enforcement have been working with 0:10:39.400 --> 0:10:42.240 one company in particular to get the data they want. 0:10:42.679 --> 0:10:44.880 Here to tell us more is four or four Media's 0:10:45.000 --> 0:10:47.360 Joseph Cox. Joseph, welcome back to the show. 0:10:47.920 --> 0:10:49.120 Thank you so much for having me. 0:10:49.600 --> 0:10:51.439 So you wrote an article which as soon as I 0:10:51.480 --> 0:10:53.680 saw the headline, I wanted to have you on the show. 0:10:53.960 --> 0:10:57.559 The headline was the two hundred plus sites an ICE 0:10:57.640 --> 0:11:01.920 surveillance contractor is monitoring. Who is the contractor, what are 0:11:01.960 --> 0:11:03.360 the sites, and what's the story? 0:11:03.880 --> 0:11:07.360 Sure? So the contractor is a company called Shadow Dragon. 0:11:07.880 --> 0:11:11.160 I seriously doubt many people have heard of this company 0:11:11.320 --> 0:11:15.760 relatively obscure, but over the past months and years they 0:11:15.800 --> 0:11:19.559 have become sort of a go to contractor for not 0:11:19.640 --> 0:11:23.880 just immigrations and Customs enforcement, but the DEA, the State Department, 0:11:24.200 --> 0:11:27.360 and other US government agencies. And what they do is 0:11:27.400 --> 0:11:33.360 basically something called open source intelligence OSANT, or social media monitoring, right, 0:11:33.440 --> 0:11:36.880 And typically we used to think that was monitoring sites 0:11:36.920 --> 0:11:41.240 like Facebook, Instagram, Twitter, all of those sorts of things, 0:11:41.480 --> 0:11:44.720 and shadow Dragon does that as well, but as you 0:11:44.760 --> 0:11:47.480 say by reading out the headline, there are many many 0:11:47.559 --> 0:11:51.360 more sites as well. Essentially, any sort of open website 0:11:51.400 --> 0:11:55.000 that you may perform some sort of activity on there's 0:11:55.040 --> 0:11:56.200 a chance it's on this list. 0:11:57.120 --> 0:12:00.760 So all day, every day, we create these digital breadcrumbs, 0:12:01.080 --> 0:12:05.600 and essentially there's somebody hoovering them up to process them 0:12:05.720 --> 0:12:08.240 and repackage them in some cases for the US government. 0:12:08.960 --> 0:12:11.680 So what the shadow Dragon tool does It is called 0:12:11.720 --> 0:12:16.240 social net the specific tool. It allows the automation and 0:12:16.400 --> 0:12:21.760 streamlining of searching those sites and social networks for a target. 0:12:21.880 --> 0:12:24.360 So let's say I'm in ICE or another government agency. 0:12:24.640 --> 0:12:27.200 I'll enter a phone number. It will then take that 0:12:27.360 --> 0:12:30.360 and see, well, there's that link to accounts on Facebook, 0:12:30.480 --> 0:12:34.040 or this site, is this username available on this social 0:12:34.080 --> 0:12:37.679 network or then this obscure platform as well, And some 0:12:37.760 --> 0:12:41.320 of these sites are much more revealing than others. Again, 0:12:41.360 --> 0:12:44.320 it's not just the big social networks. There's stuff like 0:12:44.840 --> 0:12:47.200 hiking apps on here as well. So if you want 0:12:47.240 --> 0:12:50.080 to find out where somebody's located, oh, they take a 0:12:50.080 --> 0:12:53.080 load of hikes in southern California or something like that. 0:12:53.280 --> 0:12:56.880 I think people would be, on one hand, not surprised, because, 0:12:56.920 --> 0:13:00.000 as you say, people listening probably care about data privacy 0:13:00.080 --> 0:13:02.160 see well. On the other I think they might be 0:13:02.200 --> 0:13:06.320 surprised in that you think it's just this sort of benial, 0:13:06.520 --> 0:13:08.920 insipid information you're putting out there, But when it's an 0:13:09.040 --> 0:13:11.960 aggregates pulled by a tool like this, it can be 0:13:12.040 --> 0:13:15.160 really really revealing. It almost becomes more than the sum 0:13:15.200 --> 0:13:15.840 of its parts. 0:13:16.360 --> 0:13:19.600 Jose, can you help me understand because obviously everyone's been 0:13:19.679 --> 0:13:23.559 very focused on the case of Mahamud Khalil, the Columbia 0:13:23.640 --> 0:13:25.600 student with the green card who looks like he may 0:13:25.600 --> 0:13:32.840 be deported. Is this tool about identifying people whose profile, 0:13:32.880 --> 0:13:37.719 who's on activity contravenes the current administration's priorities or is 0:13:37.760 --> 0:13:40.240 it more about, hey, this person is a target, help 0:13:40.320 --> 0:13:43.240 me build a stronger case against them. 0:13:43.400 --> 0:13:47.480 So at the moment, it's almost certainly the latter, as 0:13:47.520 --> 0:13:51.600 in ICE has access to this tool, and specifically Homeland 0:13:51.800 --> 0:13:55.920 Security Investigations as well HSI part of ICE, and they're 0:13:55.920 --> 0:13:58.880 able to use this tool for whatever part of their 0:13:58.920 --> 0:14:02.920 mission they wish to do so. Axios also reported somewhat 0:14:02.920 --> 0:14:05.200 recently that the State Department is going to be using 0:14:05.240 --> 0:14:09.520 AI to scan the social media feeds of students to 0:14:09.600 --> 0:14:12.439 see if it can detect what it deems as anti 0:14:12.600 --> 0:14:16.440 Israel or pro mass views. That's not to say Shadow 0:14:16.520 --> 0:14:20.240 Dragon is being specifically used for that. We don't know 0:14:20.360 --> 0:14:23.000 exactly what TALL the State Department may or may not use, 0:14:23.160 --> 0:14:25.840 but it absolutely sits in that context. You're right in 0:14:25.920 --> 0:14:31.520 that there are these deportations happening of various people, students 0:14:31.840 --> 0:14:35.440 and then other people visiting the US, and Shadow Dragon 0:14:35.520 --> 0:14:38.640 is very much at all there is available to US 0:14:38.720 --> 0:14:41.240 agencies if they do wish to use it for their mission. 0:14:41.840 --> 0:14:43.760 How did you first hear about Shadow Dragon. 0:14:44.760 --> 0:14:47.720 I first heard about shadow Dragon probably a couple of 0:14:47.840 --> 0:14:50.200 years ago at this point, because I was actually covering 0:14:50.960 --> 0:14:53.520 other companies in this space. So I've covered the sale 0:14:53.520 --> 0:14:56.080 of location data for a long time. I've covered the 0:14:56.240 --> 0:14:59.600 gathering of social media monitoring, and I got sent some 0:14:59.680 --> 0:15:02.640 email that showed that ICE was moving from one provider, 0:15:03.120 --> 0:15:06.360 a company called Babble Street, over to shadow Dragon, And 0:15:06.360 --> 0:15:08.600 the actual reason was because it was cheaper. At the 0:15:08.680 --> 0:15:11.880 end of the day, even when you're talking about surveillance systems, 0:15:12.000 --> 0:15:14.360 sometimes it's just about you know how much it actually 0:15:14.360 --> 0:15:17.720 costs to one of these agencies. So that move happened 0:15:17.880 --> 0:15:20.040 a while ago at this point, But then every so 0:15:20.080 --> 0:15:22.240 often I go back and check, well, his shadow Dragon 0:15:22.280 --> 0:15:24.480 got in a new contracts, and sure enough, every few 0:15:24.560 --> 0:15:26.840 months there'll be something in there. And they did just 0:15:26.920 --> 0:15:31.440 continue this contract with ICE as well. So it started 0:15:31.560 --> 0:15:35.280 very very quiet, but now they've absolutely become sort of 0:15:35.320 --> 0:15:39.080 a pretty it seems reliable solution for the US government. 0:15:39.400 --> 0:15:42.480 So the company is shadow Dragon, but the product is 0:15:43.160 --> 0:15:46.040 Social net can you just explain a bit more detail 0:15:46.080 --> 0:15:47.040 how it actually works. 0:15:47.760 --> 0:15:51.160 Yeah. Sure, So somebody will log into social net and 0:15:51.200 --> 0:15:54.040 they get the sort of graph interface of all of 0:15:54.120 --> 0:15:57.000 these dots and lines. Honestly, it almost looks like the 0:15:57.040 --> 0:16:01.760 stereotypical like hacking or surveillance move right, And of course 0:16:01.800 --> 0:16:03.840 that's funny on one end. On the other, I don't know, 0:16:03.960 --> 0:16:06.600 it's very easy for an analyst to process. I think 0:16:06.800 --> 0:16:09.200 they have this interface in front of them now, then 0:16:09.400 --> 0:16:13.240 enter some information, maybe somebody's name, they email address, their 0:16:13.280 --> 0:16:15.640 phone number, they use the name, they can click a 0:16:15.640 --> 0:16:18.240 few buttons and it will then search all of these 0:16:18.240 --> 0:16:22.520 different sites for information related to that. Did this person 0:16:22.640 --> 0:16:26.000 us a username on this dating site? Did they use 0:16:26.040 --> 0:16:30.760 the same username on this fetish website for example? That's hypothetical, 0:16:30.760 --> 0:16:33.000 but those are some of the websites that are included 0:16:33.920 --> 0:16:37.720 in this tool. It's not just the mainstream stuff. Now, 0:16:37.960 --> 0:16:41.840 an analyst could do that manually if they wished, but 0:16:42.080 --> 0:16:45.520 that would require a lot of time. And this streamlines 0:16:45.640 --> 0:16:51.200 the entire process of mapping out someone's online profile, their relationships, 0:16:51.600 --> 0:16:54.200 and potentially some of their movements as well. Well. 0:16:55.000 --> 0:16:57.160 And how do you get this list of the two 0:16:57.280 --> 0:17:00.520 hundred sites that shadow Dragon is pulling from, and can 0:17:00.520 --> 0:17:02.800 you give an example of what some leative sites are. 0:17:03.440 --> 0:17:05.280 Yeah, so I can't go to too much detail, just 0:17:05.280 --> 0:17:08.119 beyond you know, we obtained this list and then I've 0:17:08.200 --> 0:17:11.119 verified it. But when I first got the list, I 0:17:11.280 --> 0:17:15.560 was quite blown away by the length of it, really, 0:17:15.960 --> 0:17:20.080 and I started putting all of those sites into sort 0:17:20.119 --> 0:17:23.160 of buckets. So you'll have ones that focused on hobbies 0:17:23.240 --> 0:17:26.840 like I mentioned the hiking site all Trails, and a book, 0:17:27.119 --> 0:17:30.800 sort of fan website. Even chess dot com is on 0:17:30.880 --> 0:17:34.200 there as well, do a lingo. You have payment ones 0:17:34.200 --> 0:17:38.040 like cash app and PayPal, communication apps like discorder, WhatsApp. 0:17:38.119 --> 0:17:42.400 Now they're all going to have different data in different 0:17:42.480 --> 0:17:46.520 degrees available to the tool. Like there's absolutely no indication 0:17:47.240 --> 0:17:50.080 that it has access to, you know, private messages on 0:17:50.119 --> 0:17:53.120 any of these services, and I don't think that's the case. 0:17:53.160 --> 0:17:57.639 It's purely publicly available information. But again, people may not 0:17:57.840 --> 0:18:02.040 know what they're exposing in advertently or not on one 0:18:02.080 --> 0:18:03.320 of these sites. 0:18:03.960 --> 0:18:06.200 And these sites that you mentioned, I mean, are they 0:18:06.480 --> 0:18:09.360 sort of collaborating with shadow Dragon? Do they know this 0:18:09.400 --> 0:18:12.800 is happening? What is the range of responses from these sites? 0:18:12.840 --> 0:18:15.360 That shadow Dragon is pulling from, so. 0:18:15.840 --> 0:18:19.320 They're not collaborating in that shadow Dragon is going and 0:18:19.320 --> 0:18:22.919 basically grabbing the data. But there is still great variety 0:18:22.920 --> 0:18:25.080 in the responses from the companies. I actually think the 0:18:25.119 --> 0:18:27.920 most illuminating one and the most helpful one came from 0:18:28.000 --> 0:18:30.359 chess dot com, who told me on the record that, 0:18:30.440 --> 0:18:32.920 you know, we didn't know about this. We don't like it. 0:18:33.440 --> 0:18:36.800 Any sort of data scraping has to be done within 0:18:36.840 --> 0:18:38.919 the law, and if it's in within the law, it's okay, 0:18:39.200 --> 0:18:41.440 but generally we don't like it. And then you have 0:18:41.960 --> 0:18:44.760 the big companies like Meta and Snap and all of 0:18:44.760 --> 0:18:47.159 those sorts of people just pointing to their terms of 0:18:47.200 --> 0:18:51.280 service saying we don't like people scraping our website. And 0:18:51.320 --> 0:18:54.280 then I actually had a bunch who never got back 0:18:54.320 --> 0:18:57.000 to me. Essentially, even when it seems like a pretty 0:18:57.560 --> 0:18:59.240 I'm not a pr professional, but it seems like a 0:18:59.240 --> 0:19:02.320 pretty easy to just reply and say, here's our terms 0:19:02.320 --> 0:19:04.400 of service, we don't like scraping. But yeah, a lot 0:19:04.440 --> 0:19:07.640 of companies didn't even acknowledge the request for comment. 0:19:07.480 --> 0:19:10.239 On this one. So there's really these somebody don't really 0:19:10.280 --> 0:19:12.879 have a choice. I mean, their choice is to sue 0:19:13.119 --> 0:19:15.679 Shadow Dragon for violating their terms of service, which is 0:19:16.160 --> 0:19:18.520 an unlikely thing to happen because it's not directly harming 0:19:18.520 --> 0:19:22.040 their business unless there's a kind of groundswell of consumer complaint. 0:19:22.800 --> 0:19:24.960 Yeah, I think that's fair when it comes to the lawsuits, 0:19:24.960 --> 0:19:27.520 like there would have to be some sort of tectonic 0:19:27.920 --> 0:19:32.719 shift around that. We have seen WhatsApp, for example, sue 0:19:32.920 --> 0:19:36.400 NSO Group, which was or is a company that delivers 0:19:36.960 --> 0:19:40.360 packing tools to iPhones. But that's much more aggressive. That's 0:19:40.440 --> 0:19:41.919 much more active that the. 0:19:41.920 --> 0:19:45.080 Group behind the Pegasus product, that spyware product that was 0:19:45.080 --> 0:19:48.760 able to be covertly and remotely put onto people's. 0:19:48.440 --> 0:19:52.920 Phones exactly, and WhatsApp has sued them. But that's sort 0:19:52.960 --> 0:19:55.760 of you know, one company versus another company. It's not 0:19:55.800 --> 0:19:59.480 as broad as this. That being said, you know, companies 0:19:59.480 --> 0:20:02.639 try to scraping all of the time. They don't like 0:20:02.680 --> 0:20:05.160 it when you make multiple accounts, for example, they don't 0:20:05.200 --> 0:20:08.320 like it when you use a VPN to log into Facebook. 0:20:08.359 --> 0:20:10.680 I can't tell you how many times Facebook has closed 0:20:11.119 --> 0:20:13.240 my account and they have to make a new one 0:20:13.280 --> 0:20:16.159 every single time. So they do have all sorts of 0:20:16.200 --> 0:20:21.119 measures to stop weird, potentially suspicious or just unusual activity, 0:20:21.600 --> 0:20:24.800 and this would fall into that. It's just a question 0:20:24.880 --> 0:20:28.119 of whether these services can actually detect that specifically shadow 0:20:28.160 --> 0:20:31.879 Dragon or not, or of course whether they actually want 0:20:31.920 --> 0:20:34.800 to or not, because if they started doing that, I 0:20:34.840 --> 0:20:48.200 don't know, US government agencies might not be too happy. 0:20:44.800 --> 0:20:47.679 When we come back how shadow Dragon fits into the 0:20:47.800 --> 0:20:51.960 largest surveillance ecosystem used by the US government. To stay 0:20:52.000 --> 0:21:03.280 with us, have any specific examples of how shadow Dragon 0:21:03.280 --> 0:21:04.880 has been used in the real world. 0:21:05.440 --> 0:21:09.520 We only have what shadow Dragon says publicly in its 0:21:09.720 --> 0:21:13.240 marketing material because the agencies are very tight lipped or 0:21:13.240 --> 0:21:16.680 simply don't respond when talking about that sort of thing. Obviously, 0:21:16.680 --> 0:21:19.480 when it comes to ice in HSI, there's going to 0:21:19.480 --> 0:21:22.400 be an immigration component there, so I feel it's safe 0:21:22.480 --> 0:21:24.320 to assume there's a chance it's being used for those. 0:21:24.640 --> 0:21:26.800 And then some of the concrete cases that shadow Dragon 0:21:26.840 --> 0:21:32.520 has mentioned is fighting child exploitation, potentially even mapping out 0:21:32.600 --> 0:21:36.359 the opioid crisis as well. So don't get me wrong, 0:21:36.440 --> 0:21:42.160 there are clearly useful use cases for this technology. It's 0:21:42.240 --> 0:21:45.400 just that people don't really know it's going on necessarily, 0:21:45.520 --> 0:21:48.000 or that you know their government is buying this tool 0:21:48.240 --> 0:21:50.000 or again that all of this sort of data is 0:21:50.000 --> 0:21:51.000 publicly exposed. 0:21:51.359 --> 0:21:55.640 And how practically does the US government actually use this tool, 0:21:55.640 --> 0:21:58.320 and let's use ICE as an example. Is it like 0:21:59.000 --> 0:22:02.159 there's like a shadow Dragon consultant and the ICE agent 0:22:02.400 --> 0:22:05.439 sends their request to the shadow Dragon employee and the 0:22:05.440 --> 0:22:07.879 Parodragon employee generates a report or is this more like 0:22:08.000 --> 0:22:10.040 a kind of plug and play platform. 0:22:10.359 --> 0:22:13.040 It's much more a plug and play platform, almost a 0:22:13.040 --> 0:22:17.959 software as a service tool. The agency will buy licenses 0:22:18.000 --> 0:22:21.320 to use it for a year, two years, whatever. They'll 0:22:21.359 --> 0:22:23.960 log in just like you log into any sort of 0:22:23.960 --> 0:22:26.760 internet service or piece of software, and then they can 0:22:26.880 --> 0:22:30.399 use it within the parameters of what they've been allowed 0:22:30.440 --> 0:22:32.840 to do. And that's probably good for the agencies. I 0:22:32.840 --> 0:22:35.080 don't know if they necessarily want to give up information 0:22:35.119 --> 0:22:38.679 about who they're targeting to a third party, like a contractor, 0:22:38.840 --> 0:22:43.040 but it absolutely puts the honest on ICE or whoever 0:22:43.119 --> 0:22:45.960 is using it to be just very very careful with 0:22:46.040 --> 0:22:48.240 this tool as well. And I think just more broadly, 0:22:48.880 --> 0:22:52.679 we've seen repeatedly over the years that not specifically with 0:22:52.760 --> 0:22:57.120 shadow Dragon, but just other social media surveillance tools, we've 0:22:57.160 --> 0:23:00.359 seen authorities use it to monitor protests, and there was 0:23:00.359 --> 0:23:03.000 an article just in the Intercept recently that I think 0:23:03.040 --> 0:23:06.080 it was the LAPD use as similar tool to monitor 0:23:06.720 --> 0:23:10.240 protests there as well. So this is very very long running. 0:23:10.320 --> 0:23:13.240 Social media surveillance has been going on for many years, 0:23:13.520 --> 0:23:16.960 but now the playing field has expanded where it's not 0:23:17.359 --> 0:23:20.320