1 00:00:10,614 --> 00:00:15,294 Speaker 1: You're listening to a Muma Mia podcast. Mamma Mia acknowledges 2 00:00:15,334 --> 00:00:18,174 Speaker 1: the traditional owners of land and waters that this podcast 3 00:00:18,214 --> 00:00:24,854 Speaker 1: is recorded on Hi. I'm Claire Murphy. This is Mumma 4 00:00:24,934 --> 00:00:28,614 Speaker 1: MIA's twice daily news podcast, The Quickie. You know how 5 00:00:28,654 --> 00:00:31,014 Speaker 1: if you're out in public and someone sees you and 6 00:00:31,054 --> 00:00:33,054 Speaker 1: wants to find out who you are, they'd have to 7 00:00:33,134 --> 00:00:35,374 Speaker 1: launch some kind of tracing operation on the internet to 8 00:00:35,374 --> 00:00:37,974 Speaker 1: see if anyone can help track you down. But what 9 00:00:38,054 --> 00:00:42,654 Speaker 1: if that process became as easy as one quickly snapped pick. 10 00:00:43,414 --> 00:00:46,494 Speaker 2: You know, we really think that we've really broken the 11 00:00:46,534 --> 00:00:48,294 Speaker 2: sound barrier for facial recognition. 12 00:00:48,974 --> 00:00:52,054 Speaker 1: Today we're looking into how an Australian man created a 13 00:00:52,094 --> 00:00:54,534 Speaker 1: program that could mean the end to privacy as we 14 00:00:54,574 --> 00:00:56,934 Speaker 1: know it, and how he took your face and I 15 00:00:56,974 --> 00:00:59,214 Speaker 1: don't mean that in a broad like you, as in 16 00:00:59,254 --> 00:01:02,694 Speaker 1: you the public, but you listening to this right now, 17 00:01:03,094 --> 00:01:06,574 Speaker 1: your face and fed it to his machine. But before 18 00:01:06,574 --> 00:01:09,694 Speaker 1: we look into Clearview AI and it's Ossie Founder. That's 19 00:01:09,694 --> 00:01:12,214 Speaker 1: the latest from The Quickie Newsroom, Thursday November twenty one. 20 00:01:12,854 --> 00:01:15,614 Speaker 1: Royal Lotus, Troy Savann and Missy Higgins were all big 21 00:01:15,654 --> 00:01:18,454 Speaker 1: winners at last night's Arias, with Savann running out of 22 00:01:18,454 --> 00:01:20,934 Speaker 1: people to thank as he took to the stage three times. 23 00:01:21,254 --> 00:01:24,094 Speaker 1: Royal Otus took out four awards, including Best Group and 24 00:01:24,214 --> 00:01:27,334 Speaker 1: Best Rock Album for Pratts and Payin. Missy Higgins won 25 00:01:27,374 --> 00:01:29,654 Speaker 1: the public voted award for Best Live Act for the 26 00:01:29,694 --> 00:01:32,334 Speaker 1: Second Act Tour twenty twenty four, as well as being 27 00:01:32,334 --> 00:01:35,494 Speaker 1: inducted into the Arias Hall of Fame. Savann won three 28 00:01:35,534 --> 00:01:37,694 Speaker 1: trophies on the night for his album Something to Give 29 00:01:37,734 --> 00:01:40,654 Speaker 1: each Other, after taking out Best Solo Artist and Best 30 00:01:40,694 --> 00:01:43,494 Speaker 1: Pop Release. Savann took to the stage after taking out 31 00:01:43,494 --> 00:01:46,254 Speaker 1: Album of the Year saying what the hell? I literally 32 00:01:46,254 --> 00:01:48,414 Speaker 1: don't have anything to say. I said it all before. 33 00:01:48,574 --> 00:01:52,134 Speaker 1: I'm genuinely speechless. The Bluey album dance Mode won Best 34 00:01:52,214 --> 00:01:54,494 Speaker 1: Children's Album. Song of the Year went to g Flip 35 00:01:54,534 --> 00:01:57,574 Speaker 1: for the Worst person Alive and First Nations Rap supergroup 36 00:01:57,614 --> 00:02:00,054 Speaker 1: three Percent one Best Hip Hop and Dance Release for 37 00:02:00,094 --> 00:02:03,414 Speaker 1: the album Killed the Dead. The judge overseeing the case 38 00:02:03,454 --> 00:02:06,974 Speaker 1: against Sean Diddy Combs has ordered the prosecution to destroy 39 00:02:07,094 --> 00:02:10,094 Speaker 1: their copies of handwritten notes taken from Combs as jail 40 00:02:10,174 --> 00:02:12,614 Speaker 1: cell until a decision is reached on whether they can 41 00:02:12,654 --> 00:02:16,134 Speaker 1: be used as evidence. The notes reportedly discuss mister Combs's 42 00:02:16,174 --> 00:02:20,014 Speaker 1: paying a potential witness, which prosecutors say demonstrated he was 43 00:02:20,054 --> 00:02:23,374 Speaker 1: attempting to obstruct justice, but combs as lawyers argue the 44 00:02:23,414 --> 00:02:26,774 Speaker 1: notes are subject to lawyer client privilege and are therefore protected. 45 00:02:27,134 --> 00:02:29,294 Speaker 1: The prosecutors say the notes were found as part of 46 00:02:29,294 --> 00:02:32,414 Speaker 1: a jail wide safety related suite that was planned before 47 00:02:32,454 --> 00:02:35,454 Speaker 1: Combs was even arrested, a search that was properly carried 48 00:02:35,494 --> 00:02:38,014 Speaker 1: out by an investigator and then filtered by a team 49 00:02:38,054 --> 00:02:40,974 Speaker 1: of government lawyers not connected to the case, which means 50 00:02:41,014 --> 00:02:43,574 Speaker 1: the notes should be allowed in as evidence. The jarge 51 00:02:43,614 --> 00:02:45,974 Speaker 1: will keep copies of the notes until he determines whether 52 00:02:46,014 --> 00:02:49,494 Speaker 1: they can be used or not. Under the Albanesi government's 53 00:02:49,534 --> 00:02:52,854 Speaker 1: world leading ban on children accessing social media platforms could 54 00:02:52,854 --> 00:02:55,774 Speaker 1: be fined tens of millions of dollars for breaches. The 55 00:02:55,894 --> 00:02:59,054 Speaker 1: legislation will be introduced to Parliament today, setting the minimum 56 00:02:59,054 --> 00:03:02,374 Speaker 1: age for social media to sixteen and requiring social media 57 00:03:02,414 --> 00:03:05,454 Speaker 1: platforms to take reasonable steps to stop young people from 58 00:03:05,494 --> 00:03:08,814 Speaker 1: creating accounts. Parents will also not be able to give 59 00:03:08,854 --> 00:03:12,214 Speaker 1: their consent their children to use social media companies that 60 00:03:12,294 --> 00:03:15,374 Speaker 1: systematically breach the legislation could be hit with fines of 61 00:03:15,454 --> 00:03:18,294 Speaker 1: up to fifty million dollars. The reforms will allow the 62 00:03:18,294 --> 00:03:22,134 Speaker 1: Communications Minister to exclude some services from the ban, including 63 00:03:22,134 --> 00:03:27,654 Speaker 1: messaging platforms, online games and health education platforms, but Facebook, Instagram, TikTok, 64 00:03:27,694 --> 00:03:31,534 Speaker 1: YouTube and Twitter will all be sixteen and over. Sydney 65 00:03:31,534 --> 00:03:34,374 Speaker 1: siders can expect some delays today unless a pay dispute 66 00:03:34,414 --> 00:03:37,694 Speaker 1: where the city's trained workers is resolved. Train services from 67 00:03:37,734 --> 00:03:40,494 Speaker 1: Newcastle all the way down to Wollongong and across Sydney 68 00:03:40,694 --> 00:03:43,134 Speaker 1: will see all services grind to a halt from this 69 00:03:43,214 --> 00:03:46,414 Speaker 1: morning and potentially lasting for the next three days. Amid 70 00:03:46,454 --> 00:03:49,414 Speaker 1: an escalating pay dispute between the New South Wales government 71 00:03:49,414 --> 00:03:53,214 Speaker 1: and rail workers. No agreement was reached yesterday afternoon, but 72 00:03:53,374 --> 00:03:55,974 Speaker 1: talks will continue today with the hopes it will be 73 00:03:56,014 --> 00:03:59,454 Speaker 1: resolved before impacting too many New South Wales travelers. The 74 00:03:59,574 --> 00:04:02,374 Speaker 1: union's secretary said it was the government's decision to stop 75 00:04:02,454 --> 00:04:05,174 Speaker 1: trains rather than meet the union's demand for twenty four 76 00:04:05,214 --> 00:04:09,214 Speaker 1: hour rail operations. To prevent the industrial action, they're calling 77 00:04:09,214 --> 00:04:11,854 Speaker 1: for a thirty two percent pay rise across four years. 78 00:04:12,094 --> 00:04:15,894 Speaker 1: The government's offering eleven percent and while we're on industrial action. 79 00:04:16,094 --> 00:04:19,414 Speaker 1: Woolworth's warehouse workers are also planning to walk off the job, 80 00:04:19,694 --> 00:04:22,894 Speaker 1: threatening to disrupt grocery supplies in the lead up to Christmas. 81 00:04:23,254 --> 00:04:26,374 Speaker 1: It's not expected to immediately impact stores, but shelves could 82 00:04:26,494 --> 00:04:29,894 Speaker 1: end up bare if the pain conditions dispute isn't settled. 83 00:04:30,174 --> 00:04:33,014 Speaker 1: Workers say they feel pressured to cut corners to do 84 00:04:33,054 --> 00:04:36,854 Speaker 1: their jobs in the distribution centers, potentially risking their own safety. 85 00:04:37,094 --> 00:04:40,974 Speaker 1: They're calling for equal pain conditions across the country's Woolards warehouses, 86 00:04:41,254 --> 00:04:43,574 Speaker 1: raising wages to at least thirty eight dollars an hour, 87 00:04:43,654 --> 00:04:47,654 Speaker 1: with percentage increases in the following years. That's what's going 88 00:04:47,694 --> 00:04:50,214 Speaker 1: on in the world today. Next, your face exists in 89 00:04:50,214 --> 00:04:53,174 Speaker 1: a database that you've probably never heard of. But that 90 00:04:53,334 --> 00:04:56,054 Speaker 1: database has implications for all of us, and it means 91 00:04:56,094 --> 00:05:15,574 Speaker 1: privacy may just be a thing of the past. On 92 00:05:15,614 --> 00:05:19,454 Speaker 1: the Friday after Thanksgiving twenty twenty two, Atlanta local Randall 93 00:05:19,494 --> 00:05:22,294 Speaker 1: Reid was driving to his mother's house just outside the 94 00:05:22,294 --> 00:05:25,494 Speaker 1: city when he was pulled over by police. They asked 95 00:05:25,494 --> 00:05:27,654 Speaker 1: for his license, but Reid had left it at home, 96 00:05:27,894 --> 00:05:30,934 Speaker 1: apologizing to the officers instead giving them his details his 97 00:05:31,054 --> 00:05:34,134 Speaker 1: name and address after asking him if he had any 98 00:05:34,134 --> 00:05:36,814 Speaker 1: weapons on him, to which he responded no. He was 99 00:05:36,814 --> 00:05:38,494 Speaker 1: asked to step out of the vehicle, where he was 100 00:05:38,534 --> 00:05:43,294 Speaker 1: handcuffed and taken into custody. He hadn't broken any traffic laws. 101 00:05:43,654 --> 00:05:46,894 Speaker 1: In fact, Reid had no idea why he was being arrested. 102 00:05:47,694 --> 00:05:50,814 Speaker 1: Police eventually told him that he had to warrants out 103 00:05:50,814 --> 00:05:53,934 Speaker 1: on him for theft in the state of Louisiana. Reid 104 00:05:53,974 --> 00:05:57,094 Speaker 1: told the officers he'd never even been there before, but 105 00:05:57,294 --> 00:05:59,574 Speaker 1: they took him to the county jail, booked him and 106 00:05:59,734 --> 00:06:02,934 Speaker 1: ordered an extradition. The twenty nine year old had been 107 00:06:02,934 --> 00:06:06,374 Speaker 1: accused of using stolen credit cards to buy luxury designer 108 00:06:06,414 --> 00:06:10,294 Speaker 1: handbags in a store in New Orleans. Little Randall reidno, 109 00:06:10,774 --> 00:06:13,694 Speaker 1: but he was about to become the victim of an 110 00:06:13,734 --> 00:06:18,414 Speaker 1: Australian born tech genius whose AI program is now being 111 00:06:18,534 --> 00:06:24,294 Speaker 1: used by law enforcement. Juan Tantete was born in Melbourne 112 00:06:24,374 --> 00:06:27,414 Speaker 1: in nineteen eighty eight. His Tasmanian born mom is a nurse, 113 00:06:27,614 --> 00:06:29,894 Speaker 1: and his dad, who arrived in Australia on a scholarship 114 00:06:29,894 --> 00:06:33,494 Speaker 1: from Vietnam, is a literary translator. He grew up in 115 00:06:33,534 --> 00:06:35,774 Speaker 1: a household that shared a love of books, and reading. 116 00:06:36,014 --> 00:06:39,254 Speaker 1: They didn't have a TV, and according to reports, Quantantete 117 00:06:39,294 --> 00:06:42,574 Speaker 1: doesn't have one to this day. His parents noted that 118 00:06:42,694 --> 00:06:45,614 Speaker 1: little Huine was unique in that he would become interested 119 00:06:45,614 --> 00:06:48,974 Speaker 1: in something and would then focus on that thing four years, 120 00:06:49,494 --> 00:06:51,654 Speaker 1: recalling that when he was just two years old, he 121 00:06:51,694 --> 00:06:54,414 Speaker 1: fell in love with a particular piece of Vietnamese music, 122 00:06:54,694 --> 00:06:56,774 Speaker 1: which he would then listen to on a cassette tape 123 00:06:56,934 --> 00:07:00,334 Speaker 1: every single morning. He went to school in the Melbourne 124 00:07:00,334 --> 00:07:02,894 Speaker 1: inner city suburb of Footscray, where the young boy would 125 00:07:02,934 --> 00:07:06,534 Speaker 1: become obsessed with technology. From the age of six, he 126 00:07:06,614 --> 00:07:10,094 Speaker 1: was using his dad's computer at Monash University TYPEE commands 127 00:07:10,094 --> 00:07:11,974 Speaker 1: into the IBM to find out what it could do. 128 00:07:12,654 --> 00:07:15,094 Speaker 1: It was here he was first exposed to the Internet. 129 00:07:16,094 --> 00:07:18,774 Speaker 1: The Tantet family moved to Canberra when Juan was ten, 130 00:07:19,214 --> 00:07:21,934 Speaker 1: his dad getting a new job at the Australian National University, 131 00:07:22,334 --> 00:07:24,814 Speaker 1: and it would be in the nation's capital that Juan 132 00:07:24,854 --> 00:07:27,814 Speaker 1: would have the Internet at home for the very first time, 133 00:07:28,094 --> 00:07:33,734 Speaker 1: giving him the opportunity to experiment with programming. In two 134 00:07:33,734 --> 00:07:37,134 Speaker 1: thousand and seven, aged eighteen, he created his first company, 135 00:07:37,454 --> 00:07:40,334 Speaker 1: making apps, but when he failed to find any investors, 136 00:07:40,494 --> 00:07:43,054 Speaker 1: he dropped out of UNI, where he was studying computer science, 137 00:07:43,414 --> 00:07:46,014 Speaker 1: and despite pleas from his family and the teaching staff, 138 00:07:46,014 --> 00:07:48,254 Speaker 1: who desperately wanted to keep him in his course until 139 00:07:48,254 --> 00:07:52,014 Speaker 1: he graduated, Juan Tantet was determined to move to the US, 140 00:07:52,294 --> 00:07:56,414 Speaker 1: specifically Silicon Valley in San Francisco, and he never looked back. 141 00:07:57,254 --> 00:07:59,774 Speaker 1: After arriving in the US, he continued to run his 142 00:07:59,854 --> 00:08:03,894 Speaker 1: app business. He hired staff, pitched investors, and finally landed 143 00:08:03,934 --> 00:08:07,254 Speaker 1: some money from Indian entrepreneur naval Ravi Kant, who had 144 00:08:07,294 --> 00:08:10,414 Speaker 1: already put money into another couple of startup Twitter and 145 00:08:10,614 --> 00:08:14,294 Speaker 1: Uber Wine was also making some extra cash modeling on 146 00:08:14,334 --> 00:08:17,214 Speaker 1: the side. A rep from Ford Modeling Agency scouted him 147 00:08:17,254 --> 00:08:20,894 Speaker 1: when he was just nineteen. He created the company Happy 148 00:08:20,894 --> 00:08:24,174 Speaker 1: Appy in two thousand and nine and created Vidio, a 149 00:08:24,174 --> 00:08:28,014 Speaker 1: fishing app that spammed users contacts. He was investigated by 150 00:08:28,014 --> 00:08:31,894 Speaker 1: police after a worm spread. While everything seemed to be 151 00:08:31,934 --> 00:08:34,934 Speaker 1: moving really fast, the apps he was making weren't gaining 152 00:08:34,974 --> 00:08:37,574 Speaker 1: any traction, so he started to get into the open 153 00:08:37,614 --> 00:08:42,814 Speaker 1: source material available online about AI, teaching himself about its capabilities. 154 00:08:43,814 --> 00:08:45,734 Speaker 1: He moved to New York in twenty sixteen, and a 155 00:08:45,814 --> 00:08:50,134 Speaker 1: year later would found a company called Clearview AI. What 156 00:08:50,254 --> 00:08:53,534 Speaker 1: hoantan Ted had realized was that he could create programs 157 00:08:53,614 --> 00:08:56,174 Speaker 1: like his old fishing apps and worms and use them 158 00:08:56,214 --> 00:09:00,134 Speaker 1: to scrape the entire Internet for our faces, creating a 159 00:09:00,254 --> 00:09:03,974 Speaker 1: database with billions of photos of all of us that 160 00:09:04,094 --> 00:09:07,454 Speaker 1: could then be searched. The New York Times broke the 161 00:09:07,454 --> 00:09:10,334 Speaker 1: story about uantan Ted's creation in twenty five twenty, but 162 00:09:10,414 --> 00:09:13,254 Speaker 1: it hit their front pages just weeks before the COVID 163 00:09:13,294 --> 00:09:16,614 Speaker 1: pandemic began, and the story all bit faded away. But 164 00:09:16,814 --> 00:09:20,254 Speaker 1: now governments around the world are starting to understand the 165 00:09:20,294 --> 00:09:23,734 Speaker 1: implications of this technology. A program that if you have 166 00:09:23,974 --> 00:09:26,774 Speaker 1: ever had your picture on the Internet, whether it's selfie 167 00:09:26,894 --> 00:09:29,094 Speaker 1: or you in the background in a crowd at a concert, 168 00:09:29,494 --> 00:09:34,654 Speaker 1: you absolutely already exist in So how does this database work? Well, 169 00:09:34,694 --> 00:09:37,454 Speaker 1: two years ago, Quantin Tet sat down with CNN Business 170 00:09:37,494 --> 00:09:38,174 Speaker 1: to explain it. 171 00:09:38,574 --> 00:09:43,094 Speaker 2: Clearview is basically a search engine for faces. So anyone 172 00:09:43,134 --> 00:09:46,254 Speaker 2: in law enforcement can upload a face to the system 173 00:09:46,414 --> 00:09:49,774 Speaker 2: and it finds any other publicly available material that matches 174 00:09:49,774 --> 00:09:53,334 Speaker 2: that particular face. So we spent about two years really 175 00:09:53,374 --> 00:09:57,654 Speaker 2: perfecting the technology and the accuracy and the raw facial 176 00:09:57,734 --> 00:10:00,654 Speaker 2: recognition technology. We have to search out of billions and 177 00:10:00,654 --> 00:10:04,054 Speaker 2: billions of photos and still provide an accurate match and 178 00:10:04,214 --> 00:10:05,974 Speaker 2: not have any false positives. 179 00:10:07,094 --> 00:10:09,134 Speaker 1: He also went into detail about how this tech has 180 00:10:09,174 --> 00:10:12,254 Speaker 1: already he made a difference while investigating child predators. 181 00:10:12,694 --> 00:10:16,374 Speaker 2: There was a person sending selfies to a twelve year 182 00:10:16,374 --> 00:10:19,094 Speaker 2: old girl online, but it was an undercover car and 183 00:10:19,174 --> 00:10:21,134 Speaker 2: what with clear view, they were able to find his 184 00:10:21,214 --> 00:10:25,174 Speaker 2: real identity. With previous systems, they could, you know, run 185 00:10:25,294 --> 00:10:27,494 Speaker 2: their face through their existing facial wreck, but they wouldn't 186 00:10:27,494 --> 00:10:31,174 Speaker 2: find anything, mainly because of some inaccuracies, but also it's 187 00:10:31,214 --> 00:10:33,854 Speaker 2: not a large data set. You only have people who've 188 00:10:33,854 --> 00:10:38,534 Speaker 2: been arrested before, and these child predators, they're totally different 189 00:10:38,534 --> 00:10:40,694 Speaker 2: people that come from all walks of life, and so 190 00:10:40,774 --> 00:10:42,854 Speaker 2: with our tool, they were able to match him to 191 00:10:43,174 --> 00:10:46,534 Speaker 2: an online profile. He was a tax accountant. This is 192 00:10:46,534 --> 00:10:49,454 Speaker 2: the very first case we ever helped with and six 193 00:10:49,574 --> 00:10:50,814 Speaker 2: days later they were arrested. 194 00:10:51,814 --> 00:10:54,534 Speaker 1: He explained to see an n that outside law enforcement, 195 00:10:54,574 --> 00:10:56,814 Speaker 1: there are some other industries that are employing his AI, 196 00:10:57,054 --> 00:10:58,094 Speaker 1: like the banking sector. 197 00:10:58,614 --> 00:11:03,334 Speaker 2: All our customers are people who are trained investigators, so 198 00:11:03,414 --> 00:11:07,134 Speaker 2: for example, someone who does anti frauditive bank, so there 199 00:11:07,214 --> 00:11:10,054 Speaker 2: might be you know, someone who's had their identity stolen 200 00:11:10,214 --> 00:11:12,494 Speaker 2: all the money taken out of their account because someone 201 00:11:12,574 --> 00:11:14,694 Speaker 2: in person needed their identity, and they're able to use 202 00:11:14,694 --> 00:11:17,654 Speaker 2: the tool to find you know, a lead to someone 203 00:11:17,654 --> 00:11:20,054 Speaker 2: who was the real person taking the money out of 204 00:11:20,094 --> 00:11:21,054 Speaker 2: the account. 205 00:11:21,254 --> 00:11:25,014 Speaker 1: Clearview AI have also provided their technology to Ukraine's Ministry 206 00:11:25,054 --> 00:11:29,894 Speaker 1: of Defense to identify Russian combatants and verify casualties. Now 207 00:11:29,934 --> 00:11:33,094 Speaker 1: this AI technology works whether the photo posted of you 208 00:11:33,174 --> 00:11:36,134 Speaker 1: online has a filter on your face, whether you're wearing 209 00:11:36,134 --> 00:11:38,494 Speaker 1: a mask in the photo, or if you're half turned 210 00:11:38,494 --> 00:11:41,854 Speaker 1: away from the camera. The company developed dee blur and 211 00:11:42,014 --> 00:11:46,014 Speaker 1: mask removal tools using machine learning to enhance unclear images 212 00:11:46,094 --> 00:11:50,414 Speaker 1: and reveal covered faces. Quantintet says it is very very 213 00:11:50,494 --> 00:11:54,294 Speaker 1: rare to get a false positive identification, but rare doesn't 214 00:11:54,334 --> 00:11:58,694 Speaker 1: mean never. Remember poor Randall Reid, the man accused of 215 00:11:58,734 --> 00:12:02,254 Speaker 1: theft in a state he'd never visited. The CCTV camera 216 00:12:02,294 --> 00:12:05,134 Speaker 1: at the store had vision of the thief. He did 217 00:12:05,174 --> 00:12:08,014 Speaker 1: look a bit like Read, but a little heavier, more muscular. 218 00:12:08,774 --> 00:12:12,454 Speaker 1: Read's lawyer if the detective who investigated this case had 219 00:12:12,534 --> 00:12:17,294 Speaker 1: used AI technology to identify the suspect. Turns out he did, 220 00:12:17,694 --> 00:12:19,974 Speaker 1: but he wrote on the warrant to arrest mister Reed 221 00:12:20,134 --> 00:12:23,854 Speaker 1: that had been advised by a credible source. The Jefferson 222 00:12:23,894 --> 00:12:27,214 Speaker 1: Parish Sheriff's office that created the warrant signed a contract 223 00:12:27,214 --> 00:12:32,014 Speaker 1: with Clearview AI in twenty nineteen. Huantantet said they aren't 224 00:12:32,014 --> 00:12:35,334 Speaker 1: to be held responsible because a facial recognition match shouldn't 225 00:12:35,374 --> 00:12:39,094 Speaker 1: be the entire investigation, rather the beginning, and that other 226 00:12:39,174 --> 00:12:42,494 Speaker 1: factors to determine the right suspect need to be investigated 227 00:12:42,534 --> 00:12:46,014 Speaker 1: before an arrest is made. Juan, though, says one false 228 00:12:46,094 --> 00:12:49,414 Speaker 1: arrest is one too many and he has tremendous empathy 229 00:12:49,454 --> 00:12:53,454 Speaker 1: for the wrongly accused. While access to Clearview AI has 230 00:12:53,494 --> 00:12:57,734 Speaker 1: been limited to investigators, coffeecap versions with smaller databases have 231 00:12:57,854 --> 00:13:00,814 Speaker 1: popped up online and they're being used by everyday people 232 00:13:01,094 --> 00:13:03,334 Speaker 1: to find others who may not want to be found. 233 00:13:04,174 --> 00:13:07,494 Speaker 1: This has far reaching implications, for example, for women who 234 00:13:07,534 --> 00:13:10,494 Speaker 1: are trying to keep off the radars of domestic violence perpetrators, 235 00:13:10,814 --> 00:13:14,534 Speaker 1: those who've changed identities to escape violent criminals, or people 236 00:13:14,534 --> 00:13:16,894 Speaker 1: who've used other names on things like only Fans to 237 00:13:16,894 --> 00:13:20,174 Speaker 1: protect their public image and families from their private activities. 238 00:13:21,094 --> 00:13:24,814 Speaker 1: The system's ability to track and identify anyone whose photo 239 00:13:24,814 --> 00:13:28,174 Speaker 1: appears online, has led critics to describe it as creating 240 00:13:28,214 --> 00:13:31,694 Speaker 1: a perpetual police lineup of every citizen on the planet. 241 00:13:32,574 --> 00:13:36,854 Speaker 1: The company's practices have also sparked legal challenges worldwide. In 242 00:13:36,894 --> 00:13:41,414 Speaker 1: twenty twenty two, the UK privacy watchdog find CLEARVIEWAI fourteen 243 00:13:41,454 --> 00:13:45,454 Speaker 1: point five million dollars for violating its privacy laws. However, 244 00:13:45,534 --> 00:13:49,174 Speaker 1: the decision was later overruled because UK authorities don't have 245 00:13:49,254 --> 00:13:53,334 Speaker 1: authority to issue fines to a foreign company. France, Italy, 246 00:13:53,414 --> 00:13:57,094 Speaker 1: Greece and other countries in the EU also issued CLEARVIEWAI 247 00:13:57,254 --> 00:14:00,014 Speaker 1: with fines of thirty three million dollars and more. They 248 00:14:00,054 --> 00:14:02,934 Speaker 1: imposed further penalties when the company did not comply with 249 00:14:03,014 --> 00:14:06,774 Speaker 1: legal orders. In the US, the company faced a class action, 250 00:14:06,814 --> 00:14:09,534 Speaker 1: which was settled back in June, allowing them to continue 251 00:14:09,694 --> 00:14:12,414 Speaker 1: you selling the tool to US law enforcement agencies, but 252 00:14:12,574 --> 00:14:16,694 Speaker 1: not to the private sector. In Australia, the Privacy Regulator 253 00:14:16,774 --> 00:14:20,534 Speaker 1: ruled in twenty twenty one that CLEARVIEWAI violated our privacy 254 00:14:20,614 --> 00:14:24,094 Speaker 1: laws by collecting images of Australians without their consent. It 255 00:14:24,334 --> 00:14:27,334 Speaker 1: ordered the company to stop harvesting images and delete the 256 00:14:27,334 --> 00:14:30,814 Speaker 1: ones they'd already collected. In a ninety day period. There 257 00:14:30,894 --> 00:14:34,254 Speaker 1: is no evidence that CLEARVIEWAI and ever complied with this demand, 258 00:14:34,414 --> 00:14:36,734 Speaker 1: and as far as we're aware, they are still collecting 259 00:14:36,774 --> 00:14:41,214 Speaker 1: Australian faces. At one stage you could email CLEARVIEWAI and 260 00:14:41,254 --> 00:14:43,734 Speaker 1: ask to be removed from the database, and some people 261 00:14:43,814 --> 00:14:46,694 Speaker 1: say they have successfully done so, but it seems now 262 00:14:46,774 --> 00:14:50,654 Speaker 1: even those requests are not responded to. Then in August 263 00:14:50,654 --> 00:14:53,814 Speaker 1: this year, Australia's privacy regulator revealed that they would be 264 00:14:53,814 --> 00:14:58,134 Speaker 1: taking no further action against CLEARVIEWAI. Doctor Niki Swainey is 265 00:14:58,134 --> 00:15:01,414 Speaker 1: the founder and director of AI heer Way Nikki, are 266 00:15:01,454 --> 00:15:04,974 Speaker 1: Australian privacy laws really unable to stop our faces from 267 00:15:04,974 --> 00:15:05,934 Speaker 1: being harvested like this? 268 00:15:06,654 --> 00:15:10,654 Speaker 3: I think the issue with Clearview is looking at scraping 269 00:15:10,734 --> 00:15:15,134 Speaker 3: billions of images of Australians and people's faces for a 270 00:15:15,174 --> 00:15:19,734 Speaker 3: life insurance business, right, And the issue with a tool 271 00:15:19,854 --> 00:15:23,014 Speaker 3: like that is the regulation space, as we know with 272 00:15:23,174 --> 00:15:25,614 Speaker 3: AI is really trying to play catch up with how 273 00:15:25,654 --> 00:15:27,974 Speaker 3: fast this tech is moving. So it's not really a 274 00:15:28,014 --> 00:15:29,694 Speaker 3: surprise that we've dropped it because now we've started to 275 00:15:29,734 --> 00:15:32,894 Speaker 3: think deeply about what AI might mean for us, and 276 00:15:33,054 --> 00:15:34,854 Speaker 3: the fact that we don't have a whole lot of 277 00:15:35,254 --> 00:15:39,414 Speaker 3: transparency and equity and fairness associated with this in terms 278 00:15:39,454 --> 00:15:41,334 Speaker 3: of what it means for if we are in there, 279 00:15:41,694 --> 00:15:45,014 Speaker 3: I mean, whether it's Clearview or it's another image generation tool. 280 00:15:45,374 --> 00:15:47,894 Speaker 3: Chances are that if you have photos of yourself online, 281 00:15:47,934 --> 00:15:52,214 Speaker 3: which we all do these days, they are almost definitely 282 00:15:52,294 --> 00:15:55,214 Speaker 3: fed into some form of AI tool as it stands 283 00:15:55,254 --> 00:15:57,614 Speaker 3: right now, because we just don't have the regulation and 284 00:15:57,654 --> 00:15:59,894 Speaker 3: the guardrails to make sure that we can have a 285 00:16:00,134 --> 00:16:02,374 Speaker 3: informed choice about whether or not that happens. 286 00:16:02,854 --> 00:16:05,814 Speaker 1: Is part of the problem. The fact that this is global, 287 00:16:06,014 --> 00:16:10,934 Speaker 1: so Clearview is based in the US, but we're trying 288 00:16:11,014 --> 00:16:14,574 Speaker 1: to have them commit to laws here in Australia which 289 00:16:14,654 --> 00:16:17,694 Speaker 1: they legally don't have to abide by, and we can 290 00:16:17,814 --> 00:16:21,054 Speaker 1: find them. Other countries have fined CLEARVIEWAI over this, but 291 00:16:21,774 --> 00:16:23,734 Speaker 1: there's been cases where they've kind of said, well, we 292 00:16:23,814 --> 00:16:27,574 Speaker 1: can't say whether that face is someone who lives in 293 00:16:27,614 --> 00:16:30,254 Speaker 1: Italy or not, like we can't prove that they do 294 00:16:30,414 --> 00:16:32,134 Speaker 1: or prove that they don't. Is part of the problem, 295 00:16:32,214 --> 00:16:34,014 Speaker 1: the fact that it is so global and hard to 296 00:16:34,054 --> 00:16:34,974 Speaker 1: really pin it down. 297 00:16:35,214 --> 00:16:37,334 Speaker 3: Yeah, exactly, and I think we're seeing that play out 298 00:16:37,334 --> 00:16:40,694 Speaker 3: in other spaces, you know, when we look at limitations 299 00:16:40,694 --> 00:16:42,814 Speaker 3: that we're trying to put onto social media, we're getting 300 00:16:42,814 --> 00:16:44,734 Speaker 3: to the point where we realize that these companies are 301 00:16:44,814 --> 00:16:48,414 Speaker 3: much bigger than any one country's legislation. So we do 302 00:16:48,494 --> 00:16:51,534 Speaker 3: have really big issues with wanting to put guardrails on 303 00:16:51,614 --> 00:16:56,374 Speaker 3: things that exist outside of our area of control. And 304 00:16:56,574 --> 00:16:59,574 Speaker 3: it is really hard to understand how these tools are 305 00:16:59,574 --> 00:17:03,894 Speaker 3: put together and to impose our sense of responsibility on 306 00:17:03,934 --> 00:17:05,894 Speaker 3: other companies that just don't have to live up to 307 00:17:05,894 --> 00:17:09,814 Speaker 3: those standards. And even if every country was to put 308 00:17:09,934 --> 00:17:13,734 Speaker 3: in equivalent of EU compliance laws when it comes to AI, 309 00:17:13,854 --> 00:17:17,054 Speaker 3: which are among the kind of strictest, even if Australia 310 00:17:17,134 --> 00:17:22,134 Speaker 3: makes our voluntary AI guardrails mandatory, I think there's still 311 00:17:22,174 --> 00:17:24,254 Speaker 3: always going to be pockets for this stuff to exist 312 00:17:24,374 --> 00:17:26,614 Speaker 3: because it is just so hard to pin down when 313 00:17:26,654 --> 00:17:30,414 Speaker 3: you have these global corporations and these systems that exist 314 00:17:30,454 --> 00:17:33,254 Speaker 3: on cloud, and there's a really kind of blurry line 315 00:17:33,294 --> 00:17:36,854 Speaker 3: between when does something that you publicly put out there 316 00:17:37,414 --> 00:17:39,534 Speaker 3: become the property of somebody else and how do you 317 00:17:39,534 --> 00:17:42,454 Speaker 3: define what's misuse once you've given over the right to 318 00:17:43,014 --> 00:17:45,294 Speaker 3: have ownership over that material. 319 00:17:45,414 --> 00:17:48,974 Speaker 1: Clearvieway, I've said that they really only wanted this to 320 00:17:49,014 --> 00:17:52,094 Speaker 1: be used in a law enforcement capacity or other ways 321 00:17:52,094 --> 00:17:54,774 Speaker 1: in which it would benefit humanity, But we've already seen 322 00:17:54,774 --> 00:17:57,734 Speaker 1: a bunch of copycat versions that are now available online 323 00:17:57,734 --> 00:18:01,414 Speaker 1: to the public. What implication does this have for women 324 00:18:01,454 --> 00:18:03,894 Speaker 1: in particular who are trying to protect their privacy and 325 00:18:03,974 --> 00:18:07,054 Speaker 1: might be doing so for some pretty serious safety reasons. 326 00:18:07,374 --> 00:18:09,854 Speaker 1: But for women in particular, this feels very dangers. 327 00:18:10,414 --> 00:18:13,734 Speaker 3: Yes, exactly, It's something that I'm particularly concerned about. I mean, 328 00:18:13,774 --> 00:18:17,894 Speaker 3: when we look at using billions of photos without consent 329 00:18:18,414 --> 00:18:20,454 Speaker 3: for essentially a profit driven tool, I don't think it 330 00:18:20,454 --> 00:18:23,414 Speaker 3: really matters what the intent of that profit driven tool is. 331 00:18:23,454 --> 00:18:25,454 Speaker 3: But it's a profit driven tool. We're really blurring that 332 00:18:25,534 --> 00:18:30,134 Speaker 3: line between innovation and exploitation. And it's not really a 333 00:18:30,134 --> 00:18:33,054 Speaker 3: surprised that the fastest growing area of AI is in 334 00:18:33,694 --> 00:18:39,214 Speaker 3: fake girlfriends and pornographic AI, and of that sort of 335 00:18:39,294 --> 00:18:44,414 Speaker 3: genre of AI, one hundred percent of that content victimizes women. 336 00:18:44,454 --> 00:18:46,774 Speaker 3: And that's a huge concern. And you know, even in 337 00:18:46,814 --> 00:18:48,934 Speaker 3: the kind of post election era that we're all living 338 00:18:48,934 --> 00:18:51,614 Speaker 3: in right now in the US, we've seen this staggering 339 00:18:51,694 --> 00:18:57,374 Speaker 3: increase in violent, misogynistic rhetoric targeting women online and AI 340 00:18:57,454 --> 00:19:01,134 Speaker 3: really amplifies that. So for me, that's a massive, massive danger. 341 00:19:01,254 --> 00:19:03,814 Speaker 3: And I guess the harsh truth is, you know, less 342 00:19:03,854 --> 00:19:06,694 Speaker 3: than thirty percent of machine learning workforces women, so we 343 00:19:06,774 --> 00:19:08,974 Speaker 3: don't have a lot of women voices in the spaces. 344 00:19:09,574 --> 00:19:11,254 Speaker 3: And the harsh truth of that is there's a good 345 00:19:11,254 --> 00:19:15,174 Speaker 3: proportion of those men that consciously or unconsciously are probably 346 00:19:15,734 --> 00:19:18,174 Speaker 3: spurring on this sort of rhetoric. So I think it 347 00:19:18,174 --> 00:19:21,454 Speaker 3: does become really unsafe for women, and we have to 348 00:19:21,694 --> 00:19:25,054 Speaker 3: have to be talking about equity and safety and regulation 349 00:19:26,014 --> 00:19:28,014 Speaker 3: while you know, not throwing the baby out the bathwater. 350 00:19:28,054 --> 00:19:30,534 Speaker 3: AI can do fantastic things too, and it definitely has 351 00:19:30,574 --> 00:19:33,494 Speaker 3: that potential, but we really need to be having these conversations, 352 00:19:33,494 --> 00:19:35,654 Speaker 3: and I don't think we're being as loud enough about 353 00:19:35,654 --> 00:19:38,734 Speaker 3: it at all. I guess I also want to just 354 00:19:39,014 --> 00:19:42,014 Speaker 3: put out there that there is a lot of fear 355 00:19:42,334 --> 00:19:44,574 Speaker 3: and there is a lot of fear rhetoric around AI, 356 00:19:44,614 --> 00:19:48,614 Speaker 3: and I don't think that that's always necessary. These people 357 00:19:48,654 --> 00:19:51,414 Speaker 3: are really big players. They at the moment are driving 358 00:19:51,734 --> 00:19:54,734 Speaker 3: technology that is changing the very foundation of how we 359 00:19:54,814 --> 00:19:58,094 Speaker 3: operate as human beings, and especially in something like open AI, 360 00:19:58,134 --> 00:20:01,454 Speaker 3: which really has the market share of generative AI that 361 00:20:01,614 --> 00:20:06,054 Speaker 3: is a huge amount of power. But beyond their decisions 362 00:20:06,094 --> 00:20:08,374 Speaker 3: around how these things are put together, AI is one 363 00:20:08,374 --> 00:20:10,774 Speaker 3: of these things that we actually users have quite a 364 00:20:10,774 --> 00:20:13,334 Speaker 3: lot of input into. So every time you're interacting with 365 00:20:13,414 --> 00:20:17,014 Speaker 3: generative AI, we are putting our language, the things that 366 00:20:17,054 --> 00:20:20,534 Speaker 3: we care about, what we're validating our ideas into these machines, 367 00:20:20,574 --> 00:20:23,174 Speaker 3: and by default they're using that information to train the 368 00:20:23,174 --> 00:20:26,254 Speaker 3: models of AI. So I think the other particular thing 369 00:20:26,334 --> 00:20:29,014 Speaker 3: is and that people should be aware of that as consumers, 370 00:20:29,014 --> 00:20:31,494 Speaker 3: you actually have quite a lot of control over the 371 00:20:31,614 --> 00:20:35,174 Speaker 3: end product, much more so than pretty much anything else 372 00:20:35,254 --> 00:20:38,254 Speaker 3: that is commercialized at this point. So I do think 373 00:20:38,294 --> 00:20:39,974 Speaker 3: we need to be concerned. I think we need to 374 00:20:39,974 --> 00:20:42,974 Speaker 3: be having more conversations about it. I think regulation really 375 00:20:43,014 --> 00:20:45,494 Speaker 3: needs to catch up, and that's a hard thing because 376 00:20:45,494 --> 00:20:48,614 Speaker 3: AI is moving faster than anything that we've had to 377 00:20:48,654 --> 00:20:51,294 Speaker 3: tackle before. We need more women in the space and 378 00:20:51,334 --> 00:20:53,654 Speaker 3: just more diversity and representation so that we can have 379 00:20:53,694 --> 00:20:57,734 Speaker 3: those conversations about ethical and responsible AI use. But as consumers, 380 00:20:57,774 --> 00:20:59,774 Speaker 3: you should feel like you do have power and you 381 00:20:59,814 --> 00:21:02,214 Speaker 3: do have a say, because every time you interact with it, 382 00:21:02,214 --> 00:21:03,814 Speaker 3: you're helping to shape what this looks like. 383 00:21:04,814 --> 00:21:06,294 Speaker 1: Is it a case of those of us who are 384 00:21:06,294 --> 00:21:08,854 Speaker 1: not involved in this industry but still want to have 385 00:21:08,894 --> 00:21:13,494 Speaker 1: some influence here in that we contribute to what this 386 00:21:13,614 --> 00:21:15,694 Speaker 1: AI is learning from. So, for example, I work for 387 00:21:15,694 --> 00:21:19,414 Speaker 1: a women's media company, so our conversations are definitely part 388 00:21:19,454 --> 00:21:22,294 Speaker 1: of the Internet and part of what would train AI 389 00:21:22,374 --> 00:21:24,854 Speaker 1: in general. For example, I saw a video of a 390 00:21:24,894 --> 00:21:27,854 Speaker 1: guy who asked chat gpt what the world would look 391 00:21:27,934 --> 00:21:29,974 Speaker 1: like if men got their period. Oh I saw that 392 00:21:30,014 --> 00:21:32,774 Speaker 1: once the other day too. Yeah, so chat gpt really 393 00:21:32,774 --> 00:21:34,934 Speaker 1: fed back, like it really pointed out all the ways 394 00:21:34,974 --> 00:21:37,014 Speaker 1: in which misogyny plays a role in how women are 395 00:21:37,054 --> 00:21:40,814 Speaker 1: undermined in healthcare and the workplace, et cetera. So can 396 00:21:40,854 --> 00:21:43,134 Speaker 1: we actually train the misogyny out of AI? 397 00:21:43,734 --> 00:21:45,854 Speaker 3: Yeah, I mean there's real possibility we can, right. So 398 00:21:46,614 --> 00:21:49,574 Speaker 3: the thing about artificial intelligence and generative AI, it's not 399 00:21:49,734 --> 00:21:55,174 Speaker 3: this magic black box of mysterious things. It's data and 400 00:21:55,174 --> 00:21:58,054 Speaker 3: its algorithms. And you know, I spent ten years as 401 00:21:58,094 --> 00:22:01,334 Speaker 3: a data scientist and my entire job was looking at 402 00:22:01,374 --> 00:22:03,694 Speaker 3: lots and lots of information and trying to find patterns 403 00:22:03,694 --> 00:22:08,014 Speaker 3: and stories within that information. And so AI is working 404 00:22:08,014 --> 00:22:11,774 Speaker 3: on the same premise when we asking, you know, classic 405 00:22:11,894 --> 00:22:13,614 Speaker 3: like a picture of a doctor or a picture of 406 00:22:13,614 --> 00:22:15,614 Speaker 3: a CEO. It's looking at all the data that it 407 00:22:15,654 --> 00:22:17,894 Speaker 3: has about that and then it's just presenting to us 408 00:22:17,934 --> 00:22:19,974 Speaker 3: what it thinks that will please us. 409 00:22:20,174 --> 00:22:20,334 Speaker 1: Right. 410 00:22:20,414 --> 00:22:23,254 Speaker 3: I always describe generator AI like an overfed laborador. It's 411 00:22:23,254 --> 00:22:25,174 Speaker 3: just super keen to get a pat on the head 412 00:22:25,654 --> 00:22:27,454 Speaker 3: and to be told that it's doing a good job. 413 00:22:27,774 --> 00:22:30,574 Speaker 3: So it brings us back to the common denominator now 414 00:22:30,654 --> 00:22:32,934 Speaker 3: where we have control over that is. The more that 415 00:22:32,974 --> 00:22:35,774 Speaker 3: we can correct these models, the more that we can say, no, 416 00:22:35,854 --> 00:22:37,734 Speaker 3: that's not what I want, this is what I'm after, 417 00:22:37,934 --> 00:22:39,814 Speaker 3: or please look at this from a different point of view, 418 00:22:40,214 --> 00:22:42,734 Speaker 3: or no, that's not what your average forty five year 419 00:22:42,774 --> 00:22:45,054 Speaker 3: old woman looks like, you know, try again put in 420 00:22:45,094 --> 00:22:47,454 Speaker 3: these sorts of elements. That's not what I'm wanting in 421 00:22:47,454 --> 00:22:49,094 Speaker 3: this writing. I do want to look at it from 422 00:22:49,094 --> 00:22:51,654 Speaker 3: a feminist point of view. The more that we're shaping 423 00:22:52,134 --> 00:22:55,334 Speaker 3: how these things operate and what results they give us, 424 00:22:55,774 --> 00:22:58,494 Speaker 3: I think that that's a really important and should feel 425 00:22:58,494 --> 00:23:01,014 Speaker 3: like a really empowering choice, because at the moment, men 426 00:23:01,054 --> 00:23:03,494 Speaker 3: are still much more likely to be using generative AI 427 00:23:03,534 --> 00:23:06,414 Speaker 3: compared to women, and for me, that's the risk, right, 428 00:23:06,574 --> 00:23:09,894 Speaker 3: The risk is that we keep perpetuating these sterea types 429 00:23:09,934 --> 00:23:13,054 Speaker 3: because the only people interacting with it aren't picking up 430 00:23:13,094 --> 00:23:15,174 Speaker 3: on the fact that, oh, that doesn't look quite how 431 00:23:15,214 --> 00:23:17,694 Speaker 3: I would want it, or that's not really in alignment 432 00:23:17,734 --> 00:23:20,294 Speaker 3: with my lived experience. But the more diversity we put 433 00:23:20,334 --> 00:23:23,534 Speaker 3: into it, yes, we absolutely can help to shape generative 434 00:23:23,574 --> 00:23:26,974 Speaker 3: AI that actually shows us who we are as a 435 00:23:27,014 --> 00:23:30,094 Speaker 3: current society and not just thirty five year old white 436 00:23:30,134 --> 00:23:31,574 Speaker 3: American dude version of everything. 437 00:23:32,174 --> 00:23:33,854 Speaker 1: So I guess we have to get a bit over 438 00:23:33,894 --> 00:23:36,334 Speaker 1: our fear of AI and all of the scary things 439 00:23:36,374 --> 00:23:38,334 Speaker 1: that we're being told it will do, like take away 440 00:23:38,334 --> 00:23:42,574 Speaker 1: our jobs or turn everything into a homogenized, robotic version 441 00:23:42,614 --> 00:23:45,094 Speaker 1: of life for humanity. We have to kind of get 442 00:23:45,134 --> 00:23:47,414 Speaker 1: past that and to start using it and training it. 443 00:23:47,654 --> 00:23:49,774 Speaker 3: Yeah, and I mean, you know again, like that is 444 00:23:49,774 --> 00:23:52,534 Speaker 3: a real fear. I think that that fear is more 445 00:23:52,614 --> 00:23:55,494 Speaker 3: likely to come true if we don't engage. 446 00:23:56,534 --> 00:24:01,294 Speaker 1: One. Timetep maintains that clearview AI only searches publicly available information, 447 00:24:01,414 --> 00:24:04,534 Speaker 1: similar to Google, and refuses to sell the technology to 448 00:24:04,614 --> 00:24:08,894 Speaker 1: authoritarian states like China or Russia, saying they're simply misunderstood. 449 00:24:09,574 --> 00:24:12,534 Speaker 1: Levee's public opinion about what they do will change in time. 450 00:24:13,254 --> 00:24:16,094 Speaker 1: The company is now valued at approximately one hundred and 451 00:24:16,094 --> 00:24:19,614 Speaker 1: thirty million dollars yeu wess. Despite those regulatory challenges and 452 00:24:19,694 --> 00:24:23,894 Speaker 1: bands in multiple countries, their trajectory suggests that facial recognition 453 00:24:23,934 --> 00:24:27,534 Speaker 1: technology will likely become more prevalent, not less, in the 454 00:24:27,574 --> 00:24:30,094 Speaker 1: coming years. The challenge for all of us will be 455 00:24:30,174 --> 00:24:33,174 Speaker 1: finding the right balance between leveraging these powerful tools for 456 00:24:33,254 --> 00:24:37,454 Speaker 1: legitimate law enforcement purposes while protecting our individual privacy rights 457 00:24:37,614 --> 00:24:39,894 Speaker 1: and making sure we all don't end up getting caught 458 00:24:39,974 --> 00:24:43,694 Speaker 1: up in false negative identification too. Thanks for taking the 459 00:24:43,694 --> 00:24:45,854 Speaker 1: time to feed your mind with us today. The quickie 460 00:24:45,894 --> 00:24:48,614 Speaker 1: is produced by me Claire Murphy and our executive producer 461 00:24:48,654 --> 00:24:51,454 Speaker 1: Taylor Strato, with audio production by Tiagan Sadler.