WEBVTT - Truth to Power 

0:00:04.760 --> 0:00:12.960
<v Speaker 1>Sleepwalkers is a production of I Heart Radio and Unusual Productions. Hey,

0:00:14.440 --> 0:00:18.680
<v Speaker 1>how are you doing? Are you in a curious situation?

0:00:19.280 --> 0:00:24.120
<v Speaker 1>But not really? Where are you? Can you say where

0:00:24.160 --> 0:00:30.080
<v Speaker 1>you are? You sound like you can't talk. Yeah, okay,

0:00:31.000 --> 0:00:36.320
<v Speaker 1>I've been working on Sleepwalkers so much. Yeah, so I

0:00:36.400 --> 0:00:40.360
<v Speaker 1>know what you're thinking. Kara isn't normally that distracted? But

0:00:40.440 --> 0:00:44.200
<v Speaker 1>the truth is that wasn't her speaking. We're playing prerecorded

0:00:44.200 --> 0:00:47.239
<v Speaker 1>fakes of her voice to her cousin, created by AI.

0:00:47.520 --> 0:00:50.720
<v Speaker 1>Are you sleeping today? Wake you? I feel so tired.

0:00:51.600 --> 0:00:56.800
<v Speaker 1>I'm sorry. I wanted to talk to you about huh.

0:00:57.120 --> 0:01:00.400
<v Speaker 1>I stupidly left my wallet at home and I need

0:01:00.440 --> 0:01:02.880
<v Speaker 1>to order tickets to the screening before it sells out?

0:01:03.920 --> 0:01:09.720
<v Speaker 1>What's screening? I'm not? Are you you know? Could you

0:01:09.720 --> 0:01:13.360
<v Speaker 1>read me a card number real fast or text me

0:01:13.400 --> 0:01:17.640
<v Speaker 1>a pick up your card? I'll then you back. Are

0:01:17.680 --> 0:01:23.440
<v Speaker 1>you you talking to me? Your cousin Leslie? Right? Hello? Yeah?

0:01:23.680 --> 0:01:27.959
<v Speaker 1>I think we're crossing paths here. You're not answering me

0:01:28.160 --> 0:01:36.920
<v Speaker 1>in a weird You're answering me in a weird way.

0:01:37.560 --> 0:01:41.959
<v Speaker 1>So what was it like hearing Leslie respond to robot Carol? Well,

0:01:41.959 --> 0:01:44.880
<v Speaker 1>it reminded me that it's very easy to prank people

0:01:45.360 --> 0:01:47.920
<v Speaker 1>when they have no context for what you're doing. It

0:01:48.040 --> 0:01:50.680
<v Speaker 1>took her like a full minute to be like, Okay,

0:01:50.640 --> 0:01:53.040
<v Speaker 1>I's tired, not like that's not car you know, it's

0:01:53.040 --> 0:01:54.680
<v Speaker 1>always like cool my dad, And I'll say to him

0:01:54.680 --> 0:01:59.200
<v Speaker 1>after a minute, Dad, are you playing internet chess? Well,

0:01:59.200 --> 0:02:01.840
<v Speaker 1>there is what they all there's tech brain, which is

0:02:01.920 --> 0:02:05.480
<v Speaker 1>when someone's texting and talking to you, they're like and

0:02:05.560 --> 0:02:07.160
<v Speaker 1>that's sort of what it sounds like. She was like,

0:02:07.200 --> 0:02:11.200
<v Speaker 1>are you having another conversation? Has she forgiven you? She's

0:02:11.280 --> 0:02:15.880
<v Speaker 1>forgiven robo ka, I'm still not off the hook. Sorry.

0:02:17.240 --> 0:02:19.480
<v Speaker 1>Fake audio and fake video can be a lot of

0:02:19.480 --> 0:02:22.480
<v Speaker 1>fun for pranks, and there are some life changing the

0:02:22.600 --> 0:02:25.720
<v Speaker 1>positive uses for synthetic media that we hear about later.

0:02:26.360 --> 0:02:29.600
<v Speaker 1>But just how much trouble could deep fakes get us into?

0:02:30.040 --> 0:02:32.200
<v Speaker 1>And as they get easier to make, how can we

0:02:32.280 --> 0:02:34.359
<v Speaker 1>keep them out of the hands of the wrong people.

0:02:35.440 --> 0:02:52.519
<v Speaker 1>I'm as Velochen, Welcome to Sleepwalkers. The plan originally was

0:02:52.600 --> 0:02:55.880
<v Speaker 1>to get cousin Leslie's credit card detail that failed. Yeah,

0:02:56.160 --> 0:03:00.160
<v Speaker 1>Julian had the idea of having Kara Ai asked for

0:03:00.200 --> 0:03:02.960
<v Speaker 1>credit card information basically to prove how easy it is

0:03:03.360 --> 0:03:06.600
<v Speaker 1>to get somebody's credit card information. You can imagine if

0:03:07.080 --> 0:03:09.560
<v Speaker 1>it was a little bit better and you were talking

0:03:09.560 --> 0:03:12.239
<v Speaker 1>to someone and they were like, oh my god, my grandchild,

0:03:12.480 --> 0:03:14.320
<v Speaker 1>you know, needs money. Oh my god, my grandchild is

0:03:14.320 --> 0:03:17.080
<v Speaker 1>in trouble, that they would say, Okay, hold on a minute,

0:03:17.200 --> 0:03:18.840
<v Speaker 1>I'll get you the credit card number, you know what

0:03:18.880 --> 0:03:20.920
<v Speaker 1>I mean. Yeah, And I think that's what's so frightening

0:03:20.919 --> 0:03:23.880
<v Speaker 1>about this technology. We're going to dive later into how

0:03:23.919 --> 0:03:28.040
<v Speaker 1>you synthesized your voice, but it's the same technological underpinning

0:03:28.160 --> 0:03:30.720
<v Speaker 1>of the video that many people have seen of Jordan

0:03:30.800 --> 0:03:35.480
<v Speaker 1>Peel basically speaking in Barack Obama's mouth. We're entering an

0:03:35.520 --> 0:03:37.680
<v Speaker 1>era in which our enemies can make it look like

0:03:37.720 --> 0:03:41.080
<v Speaker 1>anyone is saying anything at any point in time, even

0:03:41.080 --> 0:03:44.240
<v Speaker 1>if they would never say those things. For instance, they

0:03:44.240 --> 0:03:47.280
<v Speaker 1>could have me say things like President Trump is a

0:03:47.320 --> 0:03:51.520
<v Speaker 1>total and complete dipshit. So that was a computer neural

0:03:51.560 --> 0:03:55.520
<v Speaker 1>network faking Barack Obama's facial features and mouth movement to

0:03:55.560 --> 0:03:57.920
<v Speaker 1>literally look like he was speaking the words that Jordan

0:03:57.960 --> 0:04:00.600
<v Speaker 1>Peele said, and that actually makes it even more persuasive

0:04:00.720 --> 0:04:03.040
<v Speaker 1>than the fake audio we just heard. Of your voice,

0:04:03.160 --> 0:04:06.000
<v Speaker 1>because when you see something, you tend to believe it.

0:04:06.040 --> 0:04:10.840
<v Speaker 1>That's why the phrases seeing is believing. Thanks. We're got

0:04:10.840 --> 0:04:13.200
<v Speaker 1>to come back to deep fakes, but before we get there,

0:04:13.240 --> 0:04:15.240
<v Speaker 1>we're going to take a look at some other online

0:04:15.240 --> 0:04:18.800
<v Speaker 1>trickery because the scariest part is that fakes actually don't

0:04:18.800 --> 0:04:21.719
<v Speaker 1>have to be as sophisticated as you'll call to cousin

0:04:21.800 --> 0:04:26.479
<v Speaker 1>Leslie Toreak Havoc. This is particularly true on Facebook. So

0:04:26.720 --> 0:04:29.400
<v Speaker 1>we went to their headquarters in Palo Alto to meet

0:04:29.480 --> 0:04:33.719
<v Speaker 1>Nathaniel Glica. He's the head of cybersecurity policy at Facebook,

0:04:33.760 --> 0:04:36.440
<v Speaker 1>and he told me about an incident last summer that

0:04:36.520 --> 0:04:40.400
<v Speaker 1>creates the true dilemma for him and his team. In July,

0:04:41.320 --> 0:04:44.200
<v Speaker 1>we conducted a takedown of a fairly small network of

0:04:44.320 --> 0:04:47.320
<v Speaker 1>pages that were operating in the US. Showed links back

0:04:47.360 --> 0:04:50.360
<v Speaker 1>to Russian actors, and what they were doing was, among

0:04:50.400 --> 0:04:54.000
<v Speaker 1>other things, creating events where they were inviting Americans to

0:04:54.040 --> 0:04:57.039
<v Speaker 1>come to protests, and in particular this was around the

0:04:57.279 --> 0:05:00.360
<v Speaker 1>Unite the Right two movement, which happened in It was

0:05:00.400 --> 0:05:04.080
<v Speaker 1>the anniversary of the bloody clashes in Charlottesville in seventeen,

0:05:04.520 --> 0:05:07.720
<v Speaker 1>and the far right wanted to gather again. This time

0:05:08.040 --> 0:05:11.000
<v Speaker 1>Russia was watching, and there was an event that popped

0:05:11.040 --> 0:05:13.760
<v Speaker 1>up which was the No Unite the Right to movement.

0:05:13.880 --> 0:05:17.359
<v Speaker 1>This was a counter protest. There were authentic counter protests

0:05:17.360 --> 0:05:21.159
<v Speaker 1>being planned, but this one was being convened by a

0:05:21.240 --> 0:05:24.120
<v Speaker 1>group of inauthentic pages and accounts which were linked back

0:05:24.120 --> 0:05:27.120
<v Speaker 1>to Russia that were clearly attempting to sort of bring

0:05:27.160 --> 0:05:29.080
<v Speaker 1>Americans together in a space where they would go into

0:05:29.120 --> 0:05:32.720
<v Speaker 1>physical conflict. Immediately after creating the event, they then went

0:05:32.760 --> 0:05:37.400
<v Speaker 1>out and invited legitimate, unwitting activists to co host the

0:05:37.440 --> 0:05:40.680
<v Speaker 1>event with them. Let's pause for a moment. This is

0:05:40.760 --> 0:05:44.400
<v Speaker 1>Russia we're talking about, and they're creating a Facebook event

0:05:44.440 --> 0:05:47.920
<v Speaker 1>to appeal to liberal activists, designed to draw them into

0:05:47.960 --> 0:05:51.159
<v Speaker 1>physical conflict with the far right and create the kind

0:05:51.160 --> 0:05:54.440
<v Speaker 1>of scenes that tear at our social fabric. But the

0:05:54.440 --> 0:05:58.120
<v Speaker 1>people co hosting it are not Russian agitators, their u

0:05:58.240 --> 0:06:01.560
<v Speaker 1>S citizens acting in good faith. What we saw in

0:06:01.600 --> 0:06:04.440
<v Speaker 1>that case, and what we're increasingly seeing, is these actors

0:06:04.640 --> 0:06:08.520
<v Speaker 1>trying to lure their behavior with domestic actors to force

0:06:08.600 --> 0:06:10.880
<v Speaker 1>not just the platforms but all of us to ask,

0:06:10.920 --> 0:06:14.480
<v Speaker 1>how do you separate these Ultimately, Facebook had to make

0:06:14.480 --> 0:06:19.279
<v Speaker 1>a decision. We removed that event from Facebook because it

0:06:19.320 --> 0:06:22.520
<v Speaker 1>was created by inauthentic actors. If someone else had created it,

0:06:22.520 --> 0:06:25.159
<v Speaker 1>that event would have been fine. So we removed the event.

0:06:25.560 --> 0:06:27.919
<v Speaker 1>But then we reached out to the co hosts, the

0:06:27.960 --> 0:06:30.360
<v Speaker 1>authentic hosts, and we explained to them what had happened,

0:06:30.360 --> 0:06:32.400
<v Speaker 1>and we made clear if you want to host your

0:06:32.400 --> 0:06:34.599
<v Speaker 1>own event, you should do that. We just want to

0:06:34.600 --> 0:06:36.680
<v Speaker 1>make sure that we everyone understands what's happening and what

0:06:36.720 --> 0:06:38.799
<v Speaker 1>did they say, and what was their reaction to realize

0:06:38.800 --> 0:06:41.279
<v Speaker 1>that the free will had been manipulated in that way.

0:06:41.480 --> 0:06:43.920
<v Speaker 1>If you look at reactions, it's a range of sort

0:06:43.960 --> 0:06:46.920
<v Speaker 1>of disbelief, Right, I don't think this was what you're

0:06:46.920 --> 0:06:50.599
<v Speaker 1>saying it was, too, I can't believe this happened. To Okay,

0:06:50.640 --> 0:06:52.520
<v Speaker 1>that happened, but I strongly believe in this, and I'm

0:06:52.560 --> 0:06:54.520
<v Speaker 1>gonna go and I'm going to advocate for my issues

0:06:54.560 --> 0:07:00.600
<v Speaker 1>somewhere else. That spectrum of difficulty is exactly clear why

0:07:00.800 --> 0:07:05.160
<v Speaker 1>we see actors use these techniques, because there are no

0:07:05.279 --> 0:07:08.320
<v Speaker 1>easy answers here. My assumption going into this was that

0:07:08.440 --> 0:07:12.560
<v Speaker 1>detecting misinformation would be the biggest challenge for Facebook, but

0:07:12.640 --> 0:07:15.680
<v Speaker 1>that's the easy part. It's after you identify the fakes

0:07:15.680 --> 0:07:19.560
<v Speaker 1>that the really tough questions begin. We know that, particularly

0:07:19.560 --> 0:07:22.720
<v Speaker 1>the government actors in this space, part of their information

0:07:22.760 --> 0:07:26.480
<v Speaker 1>dominance strategy is to make themselves appear bigger and more

0:07:26.480 --> 0:07:29.240
<v Speaker 1>powerful than they are. They want to seem like they're everywhere,

0:07:29.560 --> 0:07:33.760
<v Speaker 1>and it's really easy to see foreign government manipulation under

0:07:33.800 --> 0:07:37.040
<v Speaker 1>every rock. I think it's really important not to play

0:07:37.120 --> 0:07:41.120
<v Speaker 1>into the hands of these actors and sort of overplay

0:07:41.160 --> 0:07:43.760
<v Speaker 1>their own influence. This is attention we struggle with. Whatever

0:07:43.840 --> 0:07:46.920
<v Speaker 1>we conduct a takedown for some of these operations, the

0:07:46.960 --> 0:07:49.480
<v Speaker 1>most attention it gets is when we take it down.

0:07:50.880 --> 0:07:54.040
<v Speaker 1>The entire situation puts Facebook in a catch twenty two.

0:07:54.560 --> 0:07:57.200
<v Speaker 1>If they leave the content up, they're helping to promote

0:07:57.200 --> 0:08:00.840
<v Speaker 1>a foreign government's nefarious agenda. If they take it down,

0:08:00.960 --> 0:08:03.760
<v Speaker 1>the foreign government gets all this attention for being more

0:08:03.840 --> 0:08:07.800
<v Speaker 1>powerful and cleverer than they actually are. These decisions are

0:08:07.800 --> 0:08:12.640
<v Speaker 1>incredibly hard. Think of Charlottesville, Think of pizza Gate. Think

0:08:12.640 --> 0:08:15.800
<v Speaker 1>of Lane Davis, who stabbed his own father after an

0:08:15.840 --> 0:08:20.600
<v Speaker 1>argument over the conspiracy theory about liberal pedophiles. Face can kill.

0:08:21.200 --> 0:08:24.240
<v Speaker 1>And Facebook has recognized this for a start, they hired

0:08:24.320 --> 0:08:27.120
<v Speaker 1>Nathaniel a former cyber crimes prosecutor in the U s

0:08:27.160 --> 0:08:30.680
<v Speaker 1>Department of Justice, and in March of this year, Mark

0:08:30.760 --> 0:08:35.320
<v Speaker 1>Zuckerberg announced a company wide pivot towards privacy and encrypted messaging,

0:08:35.720 --> 0:08:40.520
<v Speaker 1>including services like WhatsApp, which they own. But David Kirkpatrick,

0:08:40.760 --> 0:08:44.520
<v Speaker 1>founder of Tachonomy, notes that the pivot carries its own problems.

0:08:45.679 --> 0:08:48.400
<v Speaker 1>If you look at South Asia where there's a lot

0:08:48.559 --> 0:08:54.400
<v Speaker 1>of ethnic discord political violence, notably in India, Indonesia, Me

0:08:54.559 --> 0:08:59.080
<v Speaker 1>and mar Sri Lanka. One of the primary ways that

0:08:59.080 --> 0:09:03.480
<v Speaker 1>that spreads is in group messages in WhatsApp. People in

0:09:03.520 --> 0:09:06.640
<v Speaker 1>the US don't typically use what'sapp for group messages, but

0:09:06.720 --> 0:09:09.959
<v Speaker 1>in places like India and Indonesia they do. And these

0:09:09.960 --> 0:09:12.240
<v Speaker 1>groups on five or six people, your your parents, and

0:09:12.240 --> 0:09:14.880
<v Speaker 1>your brother and city. These are like you subscribe to

0:09:15.000 --> 0:09:19.560
<v Speaker 1>a political leader or a religious zealot. So this is

0:09:19.600 --> 0:09:22.640
<v Speaker 1>more like the dear leader being piped into your home right.

0:09:22.720 --> 0:09:26.560
<v Speaker 1>So the problem has been almost more severe in those

0:09:26.920 --> 0:09:31.400
<v Speaker 1>systems than on Facebook itself of fake news and ethnic

0:09:31.760 --> 0:09:38.120
<v Speaker 1>hatred being disseminated. Because WhatsApp is a encrypted service, so

0:09:38.280 --> 0:09:41.360
<v Speaker 1>the service itself can't even see what the messages are

0:09:41.400 --> 0:09:44.400
<v Speaker 1>that are being distributed. What's scary is it doesn't take

0:09:44.440 --> 0:09:47.840
<v Speaker 1>any technical sophistication or knowledge on the part of people

0:09:48.200 --> 0:09:53.240
<v Speaker 1>writing these messages spreading this misinformation. They're just using WhatsApp. Yeah,

0:09:53.280 --> 0:09:56.559
<v Speaker 1>and these are just messaging apps and social media platforms.

0:09:56.600 --> 0:09:58.959
<v Speaker 1>But what they mean is that a single message can

0:09:59.000 --> 0:10:01.839
<v Speaker 1>spread like wild And of course the history of new

0:10:01.880 --> 0:10:05.240
<v Speaker 1>communication technology tends to go hand in hand with violence.

0:10:05.559 --> 0:10:09.000
<v Speaker 1>When the printing press I books came to Europe, they

0:10:09.040 --> 0:10:12.439
<v Speaker 1>at least religious wars, but they also made the world literate.

0:10:12.520 --> 0:10:15.200
<v Speaker 1>And we've mentioned this before. Technology is usually dual use,

0:10:15.559 --> 0:10:18.160
<v Speaker 1>which relates back to deep fikes. Mostly when you read

0:10:18.200 --> 0:10:21.080
<v Speaker 1>about deep fakes, probably thanks in part to the fact

0:10:21.080 --> 0:10:24.320
<v Speaker 1>they're called deep fakes, the coverage is not very positive.

0:10:25.000 --> 0:10:27.520
<v Speaker 1>There have been more and more stories though about positive

0:10:27.600 --> 0:10:29.920
<v Speaker 1>uses for deep fakes. So when we come back, I'm

0:10:29.920 --> 0:10:31.559
<v Speaker 1>gonna tell you more about how I faked my own

0:10:31.640 --> 0:10:33.440
<v Speaker 1>voice and also some of the things that I learned

0:10:33.480 --> 0:10:44.360
<v Speaker 1>in the process. We started this conversation a few weeks ago,

0:10:44.960 --> 0:10:48.080
<v Speaker 1>and then you asked those to create these artificial voice

0:10:48.280 --> 0:10:52.040
<v Speaker 1>based on your identity. That's Jose's to tell out, the

0:10:52.080 --> 0:10:54.839
<v Speaker 1>co founder of liar Bird. They're the company who made

0:10:55.040 --> 0:10:57.880
<v Speaker 1>robot Kara and helped me prank my cousin, and they've

0:10:57.880 --> 0:11:01.479
<v Speaker 1>published a version of their tools online at liarbird dot ai.

0:11:01.840 --> 0:11:04.640
<v Speaker 1>Here's how it works. I know it might sound a

0:11:04.640 --> 0:11:08.240
<v Speaker 1>bit like magic, but in reality, the way that our

0:11:08.240 --> 0:11:12.160
<v Speaker 1>algorithms work is basically they are just a pattern matching algorithms,

0:11:12.600 --> 0:11:17.880
<v Speaker 1>and so it's trying to figure out how to identify

0:11:18.040 --> 0:11:21.559
<v Speaker 1>the patents in your voice by comparing it against thousands

0:11:21.640 --> 0:11:23.840
<v Speaker 1>of other voices I should have tens of thousands of

0:11:23.840 --> 0:11:27.040
<v Speaker 1>other voices, and trying to figure out what is it

0:11:27.080 --> 0:11:30.520
<v Speaker 1>that makes your voice unique. Once Jose's algorithms identified what

0:11:30.600 --> 0:11:33.880
<v Speaker 1>was unique about my voice, obviously everything they had the

0:11:33.920 --> 0:11:36.640
<v Speaker 1>building blocks they needed to make a fake. Then we

0:11:36.679 --> 0:11:39.199
<v Speaker 1>sent Jose a set of sentences we wanted robot care

0:11:39.320 --> 0:11:42.040
<v Speaker 1>to say, and he used another set of algorithms to

0:11:42.120 --> 0:11:44.440
<v Speaker 1>turn the text into what we heard. The way they

0:11:44.440 --> 0:11:47.199
<v Speaker 1>do this is they use it's called a generative adversarial

0:11:47.360 --> 0:11:51.600
<v Speaker 1>network again, which is a system where one neural net

0:11:51.760 --> 0:11:54.559
<v Speaker 1>tries to trick another one a thousand times per second.

0:11:55.120 --> 0:11:58.040
<v Speaker 1>So each time the second network to tacks of fake,

0:11:58.440 --> 0:12:01.880
<v Speaker 1>the first one tries again It basically learns from its mistakes,

0:12:01.880 --> 0:12:04.680
<v Speaker 1>and once it tricks its adversary, it's ready to show

0:12:04.720 --> 0:12:08.360
<v Speaker 1>its results. In our case, liar Bird pits my fake

0:12:08.440 --> 0:12:11.199
<v Speaker 1>voice against my real voice until it sounds like this

0:12:11.679 --> 0:12:17.199
<v Speaker 1>sub dog Scara. As this technology becomes more widely available,

0:12:17.559 --> 0:12:20.920
<v Speaker 1>so does the potential for abuse. And while Liarbird develops

0:12:20.920 --> 0:12:24.439
<v Speaker 1>the technology, they don't take the ethics lightly. But Jose

0:12:24.600 --> 0:12:28.360
<v Speaker 1>has an entirely different fear. We believe that the biggest

0:12:28.480 --> 0:12:31.760
<v Speaker 1>risk of this kind of technology comes from the fact

0:12:31.840 --> 0:12:35.120
<v Speaker 1>that not a lot of people know about it. I

0:12:35.160 --> 0:12:38.640
<v Speaker 1>believe that society is not ready for what's going to

0:12:38.679 --> 0:12:42.440
<v Speaker 1>happen when this technology becomes widespread, and so I really

0:12:42.480 --> 0:12:46.959
<v Speaker 1>want to make my best effort in trying to showcase

0:12:47.040 --> 0:12:49.440
<v Speaker 1>it to the public so that they are at least

0:12:49.440 --> 0:12:53.120
<v Speaker 1>prepare for what's coming. When people know a scheme exists,

0:12:53.160 --> 0:12:55.559
<v Speaker 1>they're less likely to be tricked by it. But if

0:12:55.559 --> 0:12:58.000
<v Speaker 1>you don't know deep fakes are possible, you're much more

0:12:58.000 --> 0:13:00.800
<v Speaker 1>likely to fall for them. Leslie might been better equipped

0:13:00.840 --> 0:13:03.760
<v Speaker 1>to call my bluff had she known it was even possible.

0:13:04.160 --> 0:13:07.240
<v Speaker 1>But here's the thing, Well, there are inevitable misuses of

0:13:07.320 --> 0:13:10.400
<v Speaker 1>deep fix both behind US and on the horizon. There

0:13:10.400 --> 0:13:13.360
<v Speaker 1>are a number of extraordinary benefits of this technology, which

0:13:13.400 --> 0:13:15.760
<v Speaker 1>is why Jose is working on it. When people are

0:13:15.760 --> 0:13:18.040
<v Speaker 1>diagnosed with a LIST, it's because they start to lose

0:13:18.120 --> 0:13:21.480
<v Speaker 1>their movement skills seemed, let's say, their hands or their feet,

0:13:21.679 --> 0:13:24.240
<v Speaker 1>and so they go to the doctor and then the

0:13:24.280 --> 0:13:26.679
<v Speaker 1>doctor tells them like, you know what, this can be

0:13:27.320 --> 0:13:31.559
<v Speaker 1>als and this gets progressively worse. This was the case

0:13:31.600 --> 0:13:36.040
<v Speaker 1>for Pat Quinn, the co founder of the Ice Bucket Challenge,

0:13:36.760 --> 0:13:41.319
<v Speaker 1>creating a real fight within the a LS community. This

0:13:41.559 --> 0:13:46.760
<v Speaker 1>is a public battle now. Pat was diagnosed with a

0:13:46.920 --> 0:13:49.320
<v Speaker 1>l S and it ultimately took his ability to speak,

0:13:49.480 --> 0:13:52.560
<v Speaker 1>walk and use his hands. During this time. Since they're

0:13:52.559 --> 0:13:56.120
<v Speaker 1>diagnosed until they lose their voice, they have some time,

0:13:56.800 --> 0:13:59.240
<v Speaker 1>and so the idea is that during this time they

0:13:59.280 --> 0:14:03.199
<v Speaker 1>will be able to record themselves, ideally in a really

0:14:03.240 --> 0:14:06.559
<v Speaker 1>high quality setting. Then based on these recordings, we will

0:14:06.600 --> 0:14:08.920
<v Speaker 1>be able to create an artificial copy of their voice

0:14:09.559 --> 0:14:12.120
<v Speaker 1>which they will be able to continue using for the

0:14:12.200 --> 0:14:15.000
<v Speaker 1>rest of their life. Liar Bird has partnered with the

0:14:15.000 --> 0:14:19.000
<v Speaker 1>a l S Foundation to create Project Revoice. Just imagine

0:14:19.040 --> 0:14:22.840
<v Speaker 1>how it would feel for them, to, let's say, not

0:14:22.880 --> 0:14:25.760
<v Speaker 1>be able to tell their husband or their wife I

0:14:25.880 --> 0:14:28.560
<v Speaker 1>love you anymore, to tell this to their kids. And

0:14:28.640 --> 0:14:33.080
<v Speaker 1>so using this technology, they are able to keep this

0:14:33.280 --> 0:14:37.080
<v Speaker 1>really important part of their identities. Using the exact same

0:14:37.120 --> 0:14:40.200
<v Speaker 1>technology I used to create my deep fake liar bird

0:14:40.320 --> 0:14:42.640
<v Speaker 1>was able to give Path the ability to preserve his

0:14:42.720 --> 0:14:46.120
<v Speaker 1>voice for the rest of his life. It's a strange

0:14:46.160 --> 0:14:49.680
<v Speaker 1>feeling saying the first words at the second time. It's

0:14:49.800 --> 0:14:55.080
<v Speaker 1>like you don't realize how powerful, how personally I her

0:14:55.200 --> 0:14:58.640
<v Speaker 1>voice really is until it stay them from you. My

0:14:58.800 --> 0:15:02.440
<v Speaker 1>voice is how I back against a very disease. Take

0:15:02.480 --> 0:15:08.400
<v Speaker 1>it for me, say something, Listen to it. No voice.

0:15:10.680 --> 0:15:13.840
<v Speaker 1>Since revoicing pat Lierbard has received a number of emails

0:15:13.840 --> 0:15:16.640
<v Speaker 1>from als patients asking if it's possible for them to

0:15:16.680 --> 0:15:19.720
<v Speaker 1>do the same thing, preserve this part of themselves which

0:15:19.760 --> 0:15:22.440
<v Speaker 1>they know they're going to lose, and Jose has heard

0:15:22.440 --> 0:15:25.480
<v Speaker 1>from people who have lost family in other ways. For instance,

0:15:25.520 --> 0:15:28.680
<v Speaker 1>we have received that quite a lot, actually very emotional

0:15:28.760 --> 0:15:33.200
<v Speaker 1>emails about people telling some variation of this. My wife

0:15:33.240 --> 0:15:38.080
<v Speaker 1>died three months ago, and I have two children, age

0:15:38.200 --> 0:15:41.720
<v Speaker 1>four and six, and I would really really love to

0:15:41.840 --> 0:15:44.200
<v Speaker 1>be able to tell them a good night story in

0:15:44.240 --> 0:15:46.480
<v Speaker 1>the voice of their mother, or to tell them that,

0:15:46.520 --> 0:15:48.680
<v Speaker 1>in the mother's voice, I love you, I am proud

0:15:48.720 --> 0:15:53.560
<v Speaker 1>of you, be happy. The tools on lierbrard AI are

0:15:53.600 --> 0:15:57.080
<v Speaker 1>intentionally less advanced and meant to just spread awareness, but

0:15:57.240 --> 0:16:01.440
<v Speaker 1>Liarbrard's more bespoke tools open amazing possibilities for changing how

0:16:01.480 --> 0:16:03.960
<v Speaker 1>we deal with loss and grief. I would like to

0:16:03.960 --> 0:16:07.080
<v Speaker 1>ask you just one question, which is like, how would

0:16:07.120 --> 0:16:10.400
<v Speaker 1>you feel, let's say, about recording the voice of your

0:16:10.440 --> 0:16:13.200
<v Speaker 1>parents and keeping them What do you think would you

0:16:13.200 --> 0:16:14.720
<v Speaker 1>like to the lease or or how do you feel

0:16:14.720 --> 0:16:17.480
<v Speaker 1>about that? It was interesting when jose asked me because

0:16:17.480 --> 0:16:19.880
<v Speaker 1>I had actually thought about it ever since I learned

0:16:19.880 --> 0:16:25.840
<v Speaker 1>about Liar Bird when I was fifteen. So fourteen years ago,

0:16:26.680 --> 0:16:31.160
<v Speaker 1>my dad died in a fatal car accident, and nobody

0:16:31.240 --> 0:16:34.000
<v Speaker 1>prepares for accidents, you know. One minute my dad walked

0:16:34.000 --> 0:16:37.080
<v Speaker 1>out the door, and forty five minutes later the police

0:16:37.080 --> 0:16:39.280
<v Speaker 1>showed up at the same door to tell us what happened,

0:16:40.160 --> 0:16:43.960
<v Speaker 1>and so I never got to see or speak to

0:16:43.960 --> 0:16:49.280
<v Speaker 1>my dad ever again. Sometimes my therapist will ask me

0:16:49.360 --> 0:16:51.600
<v Speaker 1>if I think about what I would talk about with

0:16:51.640 --> 0:16:54.160
<v Speaker 1>my dad if he was still alive, and I always

0:16:54.160 --> 0:16:56.280
<v Speaker 1>say that, you know, I don't. I don't think about

0:16:56.320 --> 0:17:00.160
<v Speaker 1>that too much because it's sad to think about at

0:17:00.200 --> 0:17:02.640
<v Speaker 1>that because he's not actually around, and because I know

0:17:02.720 --> 0:17:05.639
<v Speaker 1>I can't talk about him. But it's also hard to

0:17:05.680 --> 0:17:08.560
<v Speaker 1>conceive of. You know, I can't recall off the top

0:17:08.600 --> 0:17:12.280
<v Speaker 1>of my head what he sounds like, and sometimes I'll

0:17:12.280 --> 0:17:14.359
<v Speaker 1>hear his voice when we watch home movies and it

0:17:14.359 --> 0:17:17.800
<v Speaker 1>always spooks me out. So the idea of having his

0:17:17.920 --> 0:17:21.520
<v Speaker 1>disembodied voice asked me things like how do you like

0:17:21.600 --> 0:17:24.720
<v Speaker 1>working on this podcast? Or what's the most amazing thing

0:17:24.760 --> 0:17:28.119
<v Speaker 1>you've learned, or even saying things like I'm so proud

0:17:28.160 --> 0:17:31.040
<v Speaker 1>of you. Do you know that? I'm not sure how

0:17:31.119 --> 0:17:35.600
<v Speaker 1>I'd react to his voice like that. Regardless, the thought

0:17:35.600 --> 0:17:38.280
<v Speaker 1>of it is something in the realm of possibility is

0:17:38.400 --> 0:17:43.760
<v Speaker 1>equal parts chilling and exciting. I actually think, given the chance,

0:17:43.960 --> 0:17:51.400
<v Speaker 1>I might do it. This is not the science fiction

0:17:51.760 --> 0:17:55.640
<v Speaker 1>thing or something that will exist years from now. It's

0:17:55.640 --> 0:17:58.600
<v Speaker 1>something that exists already to the people can even go on.

0:17:58.720 --> 0:18:02.359
<v Speaker 1>And as my cousin Leslie learned, these deep fakes are

0:18:02.359 --> 0:18:06.560
<v Speaker 1>already good enough to use on unassuming family members suburb

0:18:07.800 --> 0:18:12.920
<v Speaker 1>This isn't Kara, this is artificial terra. Oh my God,

0:18:13.280 --> 0:18:26.120
<v Speaker 1>my voice right now, it's AI. This is awful. When

0:18:26.119 --> 0:18:29.280
<v Speaker 1>we started reporting on deep fakes, I never anticipated how

0:18:29.359 --> 0:18:32.199
<v Speaker 1>moving the technology could be. I was more focused on

0:18:32.200 --> 0:18:35.560
<v Speaker 1>the dangers, and they are worth considering too. One person

0:18:35.640 --> 0:18:38.000
<v Speaker 1>who is out in front bringing awareness to the potential

0:18:38.119 --> 0:18:41.639
<v Speaker 1>harms of fake media is Danielle Citron. She's a legal

0:18:41.640 --> 0:18:44.159
<v Speaker 1>professor at the University of Maryland and the author of

0:18:44.200 --> 0:18:48.360
<v Speaker 1>Hate Crimes in Cyberspace. Machine learning technology and neural networks

0:18:48.400 --> 0:18:52.720
<v Speaker 1>can learn from your photo and voice that's taken from

0:18:52.760 --> 0:18:57.560
<v Speaker 1>recordings of your voice, can sufficiently learn enough about your

0:18:57.600 --> 0:19:00.760
<v Speaker 1>face and the way it moves and you or voice

0:19:00.960 --> 0:19:05.120
<v Speaker 1>so that it can create really incredibly difficult to debunk

0:19:06.080 --> 0:19:09.000
<v Speaker 1>videos of you doing and saying things you never did.

0:19:10.240 --> 0:19:12.960
<v Speaker 1>Now we all know how dangerous the simple recent word

0:19:13.040 --> 0:19:17.320
<v Speaker 1>can be. Danielle got interested in how fake video could

0:19:17.320 --> 0:19:21.400
<v Speaker 1>increase the forces of hate exponentially. There was a whole

0:19:21.440 --> 0:19:25.200
<v Speaker 1>Reddit thread devoted to deep fake sex videos of celebrities,

0:19:25.240 --> 0:19:28.919
<v Speaker 1>female celebrities like Emma Watson and Hathaway and others. If

0:19:28.960 --> 0:19:31.440
<v Speaker 1>you went through the thread, which I did, you can

0:19:31.480 --> 0:19:35.280
<v Speaker 1>see the conversation moving beyond Emma Watson to my bitch

0:19:35.320 --> 0:19:38.080
<v Speaker 1>girlfriend or that woman I hated in high school, and

0:19:38.080 --> 0:19:40.439
<v Speaker 1>it was it was all the conversation about women, you

0:19:40.480 --> 0:19:42.959
<v Speaker 1>know what I thought was like the evil Cyberus stocking

0:19:43.080 --> 0:19:46.399
<v Speaker 1>was all based on Crewe doctored photos of someone naked,

0:19:46.520 --> 0:19:48.400
<v Speaker 1>but if you worked at it, you could figure it out.

0:19:48.920 --> 0:19:52.439
<v Speaker 1>Now we can put people into pornography in ways that

0:19:52.480 --> 0:19:56.520
<v Speaker 1>are devastate their careers. So, Kara, I do think it

0:19:56.560 --> 0:19:59.280
<v Speaker 1>says something that this new technology is being used to

0:19:59.280 --> 0:20:02.320
<v Speaker 1>target women. And a lot of these conversations are happening

0:20:02.320 --> 0:20:04.600
<v Speaker 1>on the same forums on Reddit where the in cell

0:20:04.680 --> 0:20:07.359
<v Speaker 1>movement was born, right, So I think this is especially

0:20:07.440 --> 0:20:10.400
<v Speaker 1>important when we talk about famous women and their likeness.

0:20:11.440 --> 0:20:13.760
<v Speaker 1>A lot of men on the Internet want to see

0:20:13.760 --> 0:20:17.359
<v Speaker 1>their favorite actresses in positions that they wouldn't be able

0:20:17.400 --> 0:20:22.240
<v Speaker 1>to see those actresses in, and so with this technology,

0:20:22.320 --> 0:20:25.080
<v Speaker 1>it's quite easy to put someone's face on somebody else's

0:20:25.119 --> 0:20:29.040
<v Speaker 1>body without the consent of the actual actress and actually sag.

0:20:29.080 --> 0:20:32.440
<v Speaker 1>The Screen Actors Guild held a panel a few weeks

0:20:32.440 --> 0:20:34.840
<v Speaker 1>ago to bring this up that like, yes, we we're

0:20:34.880 --> 0:20:37.480
<v Speaker 1>talking about this in terms of democracy and our political

0:20:37.520 --> 0:20:39.679
<v Speaker 1>system and the upcoming election, but we also have to

0:20:39.680 --> 0:20:42.560
<v Speaker 1>talk about this in terms of the livelihood of women

0:20:42.720 --> 0:20:45.919
<v Speaker 1>who make money on their likeness and whose likeness is

0:20:45.960 --> 0:20:49.840
<v Speaker 1>now being misappropriated. Yeah, because it can destroy their careers

0:20:49.960 --> 0:20:53.200
<v Speaker 1>and silence them. There's actually a case in India where

0:20:53.280 --> 0:20:57.159
<v Speaker 1>people attempt to use deep fate pornography to intimidate and

0:20:57.359 --> 0:21:00.920
<v Speaker 1>silence a journalist called run ayub and I spoke about

0:21:00.960 --> 0:21:05.600
<v Speaker 1>that case with Danielle the Indian journalist who had been

0:21:05.680 --> 0:21:09.960
<v Speaker 1>very critical of Hindu politics, nationalist politics, and a deep

0:21:10.000 --> 0:21:13.639
<v Speaker 1>fake sex video sort of was spread basically to discredit

0:21:13.680 --> 0:21:19.040
<v Speaker 1>her um and spread through texting networks and went viral,

0:21:19.320 --> 0:21:23.000
<v Speaker 1>and she basically was devastated and went offline, stopped writing

0:21:23.560 --> 0:21:26.320
<v Speaker 1>for like three weeks. She's a journalist, this is what

0:21:26.440 --> 0:21:29.320
<v Speaker 1>she does for a living, right, So imagine that kind

0:21:29.320 --> 0:21:33.720
<v Speaker 1>of granular individual harm um and compare it with harm

0:21:33.920 --> 0:21:36.199
<v Speaker 1>to CEOs the night before an I p O. A

0:21:36.240 --> 0:21:39.159
<v Speaker 1>deep fake is release that shows this person taking a

0:21:39.200 --> 0:21:41.320
<v Speaker 1>bribe or doing drugs or whatever. I'm making it up,

0:21:41.640 --> 0:21:45.000
<v Speaker 1>but that tanks the I p O. Right. This kind

0:21:45.040 --> 0:21:48.240
<v Speaker 1>of video manipulation used to be confined to places like Disney,

0:21:48.520 --> 0:21:51.320
<v Speaker 1>and the output was blockbuster movies that are fictional but

0:21:51.400 --> 0:21:55.240
<v Speaker 1>not fake. Now AI is being consumerized, and the tools

0:21:55.240 --> 0:21:59.000
<v Speaker 1>to create convincing video are spreading, and that means creating

0:21:59.000 --> 0:22:02.359
<v Speaker 1>the kind of chaos Danielle describes is also more and

0:22:02.440 --> 0:22:06.920
<v Speaker 1>more accessible. That threatens all of us. One person working

0:22:07.000 --> 0:22:10.040
<v Speaker 1>on the issue is Hani for Reed of Dartmouth University,

0:22:10.119 --> 0:22:14.000
<v Speaker 1>who has been called the father of digital forensics. I'm

0:22:14.040 --> 0:22:18.000
<v Speaker 1>concerned that once we know you can create fake content,

0:22:18.640 --> 0:22:22.119
<v Speaker 1>there is nothing stopping anybody from saying that any video

0:22:22.240 --> 0:22:26.040
<v Speaker 1>is fake. Everybody has plausible the liability. So rewind two

0:22:26.119 --> 0:22:29.600
<v Speaker 1>years ago when the Access Hollywood tape came out of

0:22:29.640 --> 0:22:33.040
<v Speaker 1>President Trump saying what he does to women. The response

0:22:33.040 --> 0:22:35.639
<v Speaker 1>from the campaign was not this is fake. It was

0:22:35.720 --> 0:22:39.160
<v Speaker 1>we apologized, this was locker room talk. They found ways

0:22:39.200 --> 0:22:43.840
<v Speaker 1>of trying to excuse it. If that was today, guaranteed

0:22:44.040 --> 0:22:45.640
<v Speaker 1>he would have said it was fake. And in fact,

0:22:45.680 --> 0:22:47.960
<v Speaker 1>a year ago, after having apologized for the for the

0:22:48.000 --> 0:22:51.160
<v Speaker 1>audio recording, he said it was fake. And so now

0:22:51.320 --> 0:22:55.400
<v Speaker 1>politicians have plausible deniability, and at a time when our

0:22:55.520 --> 0:22:58.600
<v Speaker 1>US president is demonizing the press and telling everybody that

0:22:58.640 --> 0:23:02.440
<v Speaker 1>you can't believe anything, that credible deniability holds some weight.

0:23:02.560 --> 0:23:05.800
<v Speaker 1>And so I'm extremely concerned. Now, how do we distinguish

0:23:05.920 --> 0:23:08.040
<v Speaker 1>what's what, and that I think for a democracy is

0:23:08.040 --> 0:23:11.960
<v Speaker 1>going to be incredibly challenging. So when nothing is believable,

0:23:12.880 --> 0:23:15.680
<v Speaker 1>the mischief doer can say it's a lie. Do you

0:23:15.680 --> 0:23:17.359
<v Speaker 1>know what I'm saying? Like the person who commits the

0:23:17.400 --> 0:23:21.720
<v Speaker 1>crime or does something and says something incriminatory can say,

0:23:21.840 --> 0:23:24.520
<v Speaker 1>that's a fake. So the more you educate you both

0:23:24.560 --> 0:23:29.280
<v Speaker 1>but deep fix the evil doers can leverage that and say, well,

0:23:29.320 --> 0:23:34.520
<v Speaker 1>you can't believe anything, right. Danielle calls this the liar's dividend.

0:23:35.160 --> 0:23:38.080
<v Speaker 1>In a world where nothing can be trusted, everything can

0:23:38.080 --> 0:23:42.080
<v Speaker 1>be denied, and even documented bad deeds can be explained away.

0:23:42.119 --> 0:23:44.400
<v Speaker 1>This kind of thing is accelerated by deep fakes, though,

0:23:44.440 --> 0:23:46.919
<v Speaker 1>which is why I think there are some attempts to

0:23:47.040 --> 0:23:50.040
<v Speaker 1>correct it with law, with law like the Anti Deep

0:23:50.080 --> 0:23:54.880
<v Speaker 1>Fakes Law, very similar the Malicious Deep Fake Prohibition Act

0:23:54.880 --> 0:23:58.280
<v Speaker 1>of eighteen, which was introduced by this Republican center from

0:23:58.320 --> 0:24:01.879
<v Speaker 1>Nebraska named Ben Sas and it basically aims to outlaw

0:24:02.080 --> 0:24:06.080
<v Speaker 1>fraud in connection to audio visual records. But I don't

0:24:06.080 --> 0:24:08.520
<v Speaker 1>know if this law will path. In any case, not

0:24:08.600 --> 0:24:11.199
<v Speaker 1>all deep fakes are malicious, and so we have to

0:24:11.200 --> 0:24:15.040
<v Speaker 1>be careful with laws which are too broad. As we

0:24:15.160 --> 0:24:17.880
<v Speaker 1>heard in your Liabird piece, there are so amazingly positive

0:24:17.880 --> 0:24:22.280
<v Speaker 1>applications of deep fake technology. Here's Honi for read talking

0:24:22.320 --> 0:24:25.320
<v Speaker 1>about deep fakes and the movie business. Can you imagine

0:24:25.320 --> 0:24:29.280
<v Speaker 1>a world where the actor can simply license their appearance

0:24:29.359 --> 0:24:30.920
<v Speaker 1>and they never have to show up on the set.

0:24:31.000 --> 0:24:32.560
<v Speaker 1>You say, look, here's a bunch of images of me.

0:24:32.800 --> 0:24:35.840
<v Speaker 1>Synthesize me doing whatever you want. I'm basically an animated

0:24:35.880 --> 0:24:38.399
<v Speaker 1>character for you, and then anybody can be in the movies.

0:24:38.440 --> 0:24:40.840
<v Speaker 1>You can imagine customized movies. Imagine I go to the

0:24:40.880 --> 0:24:42.760
<v Speaker 1>movie and say, look, I'd like to see this movie,

0:24:43.119 --> 0:24:45.480
<v Speaker 1>but with George Clooney and not Kevin Spacey in it.

0:24:45.800 --> 0:24:48.720
<v Speaker 1>Please synthesize that for me. Can we do that today

0:24:48.800 --> 0:24:52.200
<v Speaker 1>or tomorrow now? But in theory, that is essentially where

0:24:52.200 --> 0:24:54.000
<v Speaker 1>we're going. So if you if you haven't seen, some

0:24:54.040 --> 0:24:56.040
<v Speaker 1>of these people are creating all these deep fake videos

0:24:56.040 --> 0:24:58.879
<v Speaker 1>of Nick Cage and inserted into all these different movies.

0:24:59.400 --> 0:25:01.680
<v Speaker 1>That's essentially and that's not the full length movie, they're

0:25:01.680 --> 0:25:04.080
<v Speaker 1>doing it into clips, but that's essentially the trend where

0:25:04.080 --> 0:25:05.840
<v Speaker 1>you can just put your favorite actor or actress into

0:25:05.880 --> 0:25:08.800
<v Speaker 1>whatever movie you want and just watch it. It's personalized movies.

0:25:09.800 --> 0:25:12.119
<v Speaker 1>I'm not gonna lie. I find it super weird that

0:25:12.200 --> 0:25:15.240
<v Speaker 1>Nicholas Cage has become the posted boy for having his

0:25:15.359 --> 0:25:18.520
<v Speaker 1>face deep faction to various movies. I wonder if you

0:25:18.560 --> 0:25:22.240
<v Speaker 1>actually asked Internet nerds, why neck Cage, what do you

0:25:22.240 --> 0:25:24.639
<v Speaker 1>think would like? I have no idea. Well, he's kind

0:25:24.680 --> 0:25:26.760
<v Speaker 1>of already a meme, right, he was, and he was

0:25:26.800 --> 0:25:29.480
<v Speaker 1>in face off where his face was switched with another

0:25:29.520 --> 0:25:31.520
<v Speaker 1>person's face, So he's always sort of in the poster

0:25:31.640 --> 0:25:34.840
<v Speaker 1>child for face swapping, you know. I think actually one

0:25:34.840 --> 0:25:37.360
<v Speaker 1>thing that I thought about is this idea of representation.

0:25:37.400 --> 0:25:40.040
<v Speaker 1>You know, if there's a movie or movies or series

0:25:40.119 --> 0:25:44.240
<v Speaker 1>like James Bond where the lead character has been historically white,

0:25:44.840 --> 0:25:48.840
<v Speaker 1>and you want to show your African American son James Bond,

0:25:49.280 --> 0:25:53.240
<v Speaker 1>it would be kind of cool to make James Bond black, right,

0:25:53.359 --> 0:25:55.880
<v Speaker 1>because then your child could be watching a movie where

0:25:55.960 --> 0:25:59.120
<v Speaker 1>James Bond looks like your child. Absolutely, And I think

0:25:59.160 --> 0:26:01.159
<v Speaker 1>one of the big pro BOMs in the movie business

0:26:01.160 --> 0:26:04.640
<v Speaker 1>and the media business in general is representation. So more

0:26:04.680 --> 0:26:08.040
<v Speaker 1>people do have access to this technology now, but it

0:26:08.200 --> 0:26:11.720
<v Speaker 1>used to be that only a Hollywood special Effects company

0:26:11.840 --> 0:26:15.600
<v Speaker 1>would have access to this technology. When you remove the gatekeepers,

0:26:15.920 --> 0:26:19.480
<v Speaker 1>you get these incredible explosions of culture, but you also

0:26:19.560 --> 0:26:22.760
<v Speaker 1>get real threats to the social fabric. And so in

0:26:22.760 --> 0:26:25.320
<v Speaker 1>the case of deep fakes, and they're all very well

0:26:25.400 --> 0:26:28.119
<v Speaker 1>when they're labeled as fake or when we know they're fake,

0:26:28.640 --> 0:26:32.600
<v Speaker 1>but when they're posing as real, that's when we start

0:26:32.640 --> 0:26:35.879
<v Speaker 1>to be really under threat, I think as a society.

0:26:35.920 --> 0:26:38.520
<v Speaker 1>But there are people working on this as ever. Cat

0:26:38.560 --> 0:26:41.199
<v Speaker 1>and Mouse. When we come back, we'll talk about some

0:26:41.280 --> 0:26:50.159
<v Speaker 1>of the ways they're fighting back. When it comes to

0:26:50.200 --> 0:26:54.280
<v Speaker 1>deep fakes. Pandora's box is open, and as Jose argues,

0:26:54.320 --> 0:26:58.040
<v Speaker 1>there's no turning back the clocks. The technology exists. So,

0:26:58.200 --> 0:27:01.040
<v Speaker 1>knowing deep fakes and fake news have become more sophisticated,

0:27:01.400 --> 0:27:04.240
<v Speaker 1>I wanted to find out how actual news organizations are

0:27:04.240 --> 0:27:07.200
<v Speaker 1>thinking about the problem. So I spoke with John mccathwaite,

0:27:07.440 --> 0:27:10.800
<v Speaker 1>editor in chief of Bloomberg News, and he actually started

0:27:10.960 --> 0:27:14.679
<v Speaker 1>by pointing out that fate news isn't new news. I

0:27:14.720 --> 0:27:16.840
<v Speaker 1>think that one crucial thing when you look at fake

0:27:16.920 --> 0:27:18.720
<v Speaker 1>news is tom it's always been there. You know. The

0:27:18.760 --> 0:27:21.800
<v Speaker 1>first bit of fate news was the trojan horse fake

0:27:21.880 --> 0:27:24.320
<v Speaker 1>news and propaganda have for ever been some of the

0:27:24.440 --> 0:27:28.560
<v Speaker 1>more exotic weapons in global conflict. John points to another

0:27:28.600 --> 0:27:32.280
<v Speaker 1>example involving the famous British by and author of James

0:27:32.359 --> 0:27:36.399
<v Speaker 1>bond Ian Fleming. Supposedly one of his great schemes was

0:27:36.480 --> 0:27:42.520
<v Speaker 1>to drop lots of jumbo sized condoms over Germany um

0:27:42.520 --> 0:27:46.439
<v Speaker 1>and label sort of British small on the outside in

0:27:46.520 --> 0:27:49.560
<v Speaker 1>the end in the name, with the aim, no doubt wrongly,

0:27:49.600 --> 0:27:53.639
<v Speaker 1>of destabilizing German manhood. My point is that there are

0:27:53.680 --> 0:27:56.600
<v Speaker 1>many many ways in which you can do this. But

0:27:56.840 --> 0:28:00.159
<v Speaker 1>the most interesting thing to me about fate news is

0:28:00.160 --> 0:28:04.480
<v Speaker 1>that really in modern history it's tied very heavily to technology.

0:28:04.560 --> 0:28:06.840
<v Speaker 1>What tends to happen is a new technology comes along

0:28:07.400 --> 0:28:10.720
<v Speaker 1>which suddenly sets media free. If we look to history

0:28:10.840 --> 0:28:13.920
<v Speaker 1>we can understand this moment better. We mentioned the early

0:28:14.000 --> 0:28:17.600
<v Speaker 1>printing press before and how it enabled explosions of ideology

0:28:17.680 --> 0:28:21.119
<v Speaker 1>and led to religious conflicts. Well, when the printing press

0:28:21.200 --> 0:28:24.520
<v Speaker 1>was industrialized in the nineteenth century, there was another fake

0:28:24.640 --> 0:28:28.160
<v Speaker 1>news boom. Go back to I think it's no one.

0:28:28.240 --> 0:28:30.399
<v Speaker 1>You have the invention of the steam press in London,

0:28:30.800 --> 0:28:33.720
<v Speaker 1>and what that does It enables people to multiple by

0:28:33.760 --> 0:28:37.080
<v Speaker 1>ten the amount of paper that you can print suddenly

0:28:37.080 --> 0:28:40.640
<v Speaker 1>all the way across Europe, and then in America, free

0:28:40.680 --> 0:28:44.120
<v Speaker 1>ship newspaper starts bringing up. Because you can distribute far more,

0:28:44.200 --> 0:28:47.280
<v Speaker 1>you can reach far more people, far more quickly. And

0:28:47.320 --> 0:28:49.560
<v Speaker 1>the most notorious of this was The New York Sun

0:28:49.640 --> 0:28:53.040
<v Speaker 1>at one time, I think, the world's biggest selling paper,

0:28:53.240 --> 0:28:56.360
<v Speaker 1>run by Michael Benjamin Day, and he would run some

0:28:56.480 --> 0:28:59.160
<v Speaker 1>stories like the moon was populated by people who are

0:28:59.160 --> 0:29:03.280
<v Speaker 1>half human half bat. But what happened, and I think

0:29:03.280 --> 0:29:07.000
<v Speaker 1>this will happen again, is that consumers said, we don't

0:29:07.000 --> 0:29:09.680
<v Speaker 1>want to read that, we need facts. And so if

0:29:09.680 --> 0:29:12.600
<v Speaker 1>you look back at many of the big newspapers of

0:29:12.640 --> 0:29:14.840
<v Speaker 1>our time, the New York Times, the Economists were I

0:29:14.920 --> 0:29:17.440
<v Speaker 1>used to work, many of these things came from that

0:29:17.480 --> 0:29:21.160
<v Speaker 1>particular period because people paid more to get things they trusted. Well,

0:29:21.200 --> 0:29:24.520
<v Speaker 1>that is definitely happening again. In other words, most of

0:29:24.560 --> 0:29:28.280
<v Speaker 1>the high quality press today, the New York Times, the Economist,

0:29:28.440 --> 0:29:33.120
<v Speaker 1>which John also edited, came from consumer demand for trustworthy information.

0:29:33.760 --> 0:29:36.280
<v Speaker 1>And that same consumer demand may help us out of

0:29:36.280 --> 0:29:40.480
<v Speaker 1>today's predicament. But there is one key difference. Now we

0:29:40.560 --> 0:29:43.360
<v Speaker 1>have deep fakes. It's worth a lot of money to

0:29:43.400 --> 0:29:45.000
<v Speaker 1>a lot of people to try to fool us. So

0:29:45.080 --> 0:29:47.320
<v Speaker 1>you look at things like Twitter handles that aren't quite

0:29:47.320 --> 0:29:50.800
<v Speaker 1>the same, some mixture between humans and computers. You used

0:29:50.800 --> 0:29:54.760
<v Speaker 1>to deal with those. What is harder at the moment

0:29:54.880 --> 0:29:57.600
<v Speaker 1>is video. So to give you an example, I think

0:29:57.600 --> 0:29:59.160
<v Speaker 1>a year or so ago, there was some tack in

0:29:59.240 --> 0:30:01.840
<v Speaker 1>a subway in New Org. We could verify really quite

0:30:01.920 --> 0:30:05.680
<v Speaker 1>quickly that the subway attack had happened, but almost immediately

0:30:05.720 --> 0:30:08.280
<v Speaker 1>there was a picture on Twitter of one of the

0:30:08.320 --> 0:30:11.960
<v Speaker 1>alleged assailants lying in a pool of blood. Now, trying

0:30:12.000 --> 0:30:15.440
<v Speaker 1>to verify that that was true was much harder, and

0:30:15.480 --> 0:30:17.520
<v Speaker 1>it came down to things like working out whether that

0:30:17.600 --> 0:30:20.480
<v Speaker 1>was the correct subway floor. You can look at pixels,

0:30:20.560 --> 0:30:22.960
<v Speaker 1>you can look at all those different things, But yes,

0:30:23.200 --> 0:30:26.480
<v Speaker 1>verifying video is often harder than verifying facts. Do you

0:30:26.520 --> 0:30:29.440
<v Speaker 1>have any tools or technologies that you're licensing or spending

0:30:29.440 --> 0:30:31.440
<v Speaker 1>money on to do it. We spend a lot of

0:30:31.440 --> 0:30:34.040
<v Speaker 1>money on technology across all these fronts. With more and

0:30:34.080 --> 0:30:37.880
<v Speaker 1>more news coming directly from social media, large news organizations

0:30:37.920 --> 0:30:41.320
<v Speaker 1>like Bloomberg News need to be able to verify which

0:30:41.320 --> 0:30:44.760
<v Speaker 1>photos and videos are real and whether they actually relate

0:30:44.800 --> 0:30:47.680
<v Speaker 1>to the events they're investigating, which is why Harni for

0:30:47.800 --> 0:30:51.440
<v Speaker 1>Reid is in such high demand. Suddenly, the need to

0:30:51.520 --> 0:30:56.520
<v Speaker 1>authenticate content has really global implications. Everything from our court

0:30:56.640 --> 0:31:00.760
<v Speaker 1>to our national security, to our democratical actions to citizens

0:31:00.760 --> 0:31:03.480
<v Speaker 1>safety is starting to rely on our ability to tell

0:31:03.560 --> 0:31:05.640
<v Speaker 1>the real from the fake. And so I think this

0:31:05.760 --> 0:31:08.760
<v Speaker 1>field of forensics, this field of authentication, has never been

0:31:08.800 --> 0:31:11.959
<v Speaker 1>more important, and that's what Harney spends his days working

0:31:11.960 --> 0:31:15.920
<v Speaker 1>on at Dartmouth. He develops techniques to analyze and authenticate

0:31:16.040 --> 0:31:19.960
<v Speaker 1>digital media. Ahead of the elections, He's working on what

0:31:20.080 --> 0:31:23.920
<v Speaker 1>he calls a soft biometric tool to detect fake videos

0:31:24.000 --> 0:31:29.200
<v Speaker 1>of specific politicians such as Bernie Saunders, Elizabeth Warren, and

0:31:29.280 --> 0:31:32.240
<v Speaker 1>Donald Trump. UM I would say the game is going

0:31:32.320 --> 0:31:36.000
<v Speaker 1>to be that we never eliminate the ability to create

0:31:36.040 --> 0:31:38.960
<v Speaker 1>fake content, but what we do is we raise the bar.

0:31:39.520 --> 0:31:41.320
<v Speaker 1>We take it out of the hands of the amateurs,

0:31:41.320 --> 0:31:43.440
<v Speaker 1>we take it out of the hands of the average

0:31:43.480 --> 0:31:46.600
<v Speaker 1>person downloading some code, and we make it more difficult,

0:31:46.640 --> 0:31:48.920
<v Speaker 1>more time consuming, and more risky. And this is the

0:31:48.960 --> 0:31:52.040
<v Speaker 1>same thing that we do with counterfeit currency. You can

0:31:52.080 --> 0:31:55.080
<v Speaker 1>still create counterfeit currency today, but it's really hard, still

0:31:55.120 --> 0:31:57.880
<v Speaker 1>a risk, but it's a more manageable risk. On the

0:31:57.920 --> 0:32:00.960
<v Speaker 1>subject of money, there are digital orencies which are much

0:32:00.960 --> 0:32:04.120
<v Speaker 1>more difficult to counterfeit than coins and banknotes. You've heard

0:32:04.120 --> 0:32:07.640
<v Speaker 1>of Bitcoin and ethereum, which you're enabled by blockchain, a

0:32:07.760 --> 0:32:12.080
<v Speaker 1>so called distributed ledger. Information about transactions is shared between

0:32:12.120 --> 0:32:15.800
<v Speaker 1>all the users of the currency, rather than authenticated and

0:32:15.880 --> 0:32:19.680
<v Speaker 1>guarded by a bank. Sharing this kind of information across

0:32:19.680 --> 0:32:23.160
<v Speaker 1>a crowd of people with multiple backup copies has a

0:32:23.320 --> 0:32:26.480
<v Speaker 1>range of uses. One thing Harney is looking at is

0:32:26.600 --> 0:32:31.600
<v Speaker 1>using blockchain to authenticate images and videos. At source, we're

0:32:31.640 --> 0:32:35.320
<v Speaker 1>gonna start seeing, um the use of a different type

0:32:35.320 --> 0:32:38.000
<v Speaker 1>of camera. So there are now companies out there that

0:32:38.120 --> 0:32:42.040
<v Speaker 1>create what are called secure imaging pipelines, and so when

0:32:42.080 --> 0:32:46.400
<v Speaker 1>you record an image or video, they extract a unique

0:32:46.440 --> 0:32:50.120
<v Speaker 1>signature from that content, they cryptographically signed it, and they

0:32:50.160 --> 0:32:53.320
<v Speaker 1>put that on the blockchain. So that's basically a distributed

0:32:53.400 --> 0:32:56.400
<v Speaker 1>ledger that's very very hard, if not impossible, to manipulate.

0:32:56.720 --> 0:32:59.719
<v Speaker 1>Perhaps staying ahead of the perpetrators and making fakes more

0:32:59.760 --> 0:33:02.400
<v Speaker 1>differ are cooled is the best we can do. But

0:33:02.520 --> 0:33:06.080
<v Speaker 1>what about our usage? How much responsibility do we have

0:33:06.240 --> 0:33:09.840
<v Speaker 1>to navigate the web thoughtfully? And how much responsibility should

0:33:09.840 --> 0:33:13.080
<v Speaker 1>be on the platforms. We have Facebook, Twitter, we have

0:33:13.200 --> 0:33:17.760
<v Speaker 1>yelped because they're not responsible for user generated content. What's

0:33:17.800 --> 0:33:22.040
<v Speaker 1>interesting is that, like Nathaniel at Facebook, Danielle also sees

0:33:22.160 --> 0:33:26.320
<v Speaker 1>risks in over zealous moderation. If you put too much

0:33:26.560 --> 0:33:31.480
<v Speaker 1>responsibility on the platform, you will likely incentivize over censorship.

0:33:31.840 --> 0:33:33.880
<v Speaker 1>So all the great things that we think about a

0:33:33.920 --> 0:33:37.080
<v Speaker 1>lot of these platforms, and especially the social media, the

0:33:37.160 --> 0:33:40.080
<v Speaker 1>Parkland survivors or Black Lives Matters, right, we don't want

0:33:40.080 --> 0:33:43.800
<v Speaker 1>to lose the facility and new enablements for organizing and speech.

0:33:44.280 --> 0:33:46.520
<v Speaker 1>So if you put too much liability on the platforms,

0:33:46.520 --> 0:33:50.440
<v Speaker 1>they're going to overreact and anything anyone complains about and

0:33:50.480 --> 0:33:53.840
<v Speaker 1>have very aggressive filters. So we might very well miss

0:33:53.880 --> 0:33:57.040
<v Speaker 1>Black Lives Matter, we might not have Parkland and never

0:33:57.200 --> 0:34:00.880
<v Speaker 1>see it because you're gonna have overly are of censorship.

0:34:01.120 --> 0:34:05.240
<v Speaker 1>Here's Nathaniel again. Whenever people come together in a new medium,

0:34:05.400 --> 0:34:07.320
<v Speaker 1>you're going to have people that try to manipulate and

0:34:07.320 --> 0:34:09.720
<v Speaker 1>try to take advantage. I think one of the things

0:34:09.760 --> 0:34:12.440
<v Speaker 1>that's really fundamentally true that we have done when we

0:34:12.480 --> 0:34:15.799
<v Speaker 1>think about the Internet generally social media as well, is

0:34:15.840 --> 0:34:18.799
<v Speaker 1>we've removed some of the traditional gatekeeping mechanisms that have

0:34:18.840 --> 0:34:22.000
<v Speaker 1>existed in the past, and that has meant that far

0:34:22.080 --> 0:34:25.040
<v Speaker 1>more people could engage, much more quickly and much more

0:34:25.120 --> 0:34:27.120
<v Speaker 1>vocally than ever before, and that has led to some

0:34:27.200 --> 0:34:29.560
<v Speaker 1>incredible things. If you think about the me too movement,

0:34:29.800 --> 0:34:32.120
<v Speaker 1>which really part of what drives it and enables it

0:34:32.160 --> 0:34:34.520
<v Speaker 1>is the ability to route around some of those gatekeepers, right,

0:34:35.120 --> 0:34:37.359
<v Speaker 1>But at the same time, you're also going to see

0:34:37.400 --> 0:34:39.520
<v Speaker 1>malicious actors try to misuse that. I think that is

0:34:39.520 --> 0:34:42.720
<v Speaker 1>a fundamental truth for any form of media. The question

0:34:42.840 --> 0:34:48.200
<v Speaker 1>is how do you enable authentic engagement while making the

0:34:48.239 --> 0:34:53.160
<v Speaker 1>types of manipulation that we see more difficult. If Facebook

0:34:53.200 --> 0:34:58.360
<v Speaker 1>and other platforms are too destructive of society, ultimately everyone loses,

0:34:58.840 --> 0:35:02.440
<v Speaker 1>even the technology company needs and their shareholders. So how

0:35:02.440 --> 0:35:06.359
<v Speaker 1>do we move from understanding that to finding solutions. Here's

0:35:06.440 --> 0:35:12.439
<v Speaker 1>David Kirkpatrick again. If we are going to retain democracy,

0:35:12.960 --> 0:35:19.320
<v Speaker 1>we need technical systems, digital systems, technologies that more effectively

0:35:19.719 --> 0:35:25.160
<v Speaker 1>and persuasively compelling lee distribute knowledge so that we have

0:35:25.280 --> 0:35:30.040
<v Speaker 1>citizens that are capable of functioning in a democratic landscape

0:35:30.120 --> 0:35:35.440
<v Speaker 1>that is more complex, more rapidly changing, and ultimately more global.

0:35:35.719 --> 0:35:38.680
<v Speaker 1>And as far as Hanni ferred is concerned, this has

0:35:38.719 --> 0:35:42.040
<v Speaker 1>become everyone's problem, so we all have a part to

0:35:42.120 --> 0:35:44.560
<v Speaker 1>play in solving it. I think two things are gonna

0:35:44.560 --> 0:35:46.840
<v Speaker 1>have to change. So one is the technology to authenticate

0:35:46.840 --> 0:35:48.439
<v Speaker 1>it is going to have to get better. So whether

0:35:48.480 --> 0:35:50.680
<v Speaker 1>that's authenticating at the source or the types of things

0:35:50.680 --> 0:35:54.200
<v Speaker 1>that I do with authenticating content and operating that at scale,

0:35:54.280 --> 0:35:56.000
<v Speaker 1>that's going to have to get better. But I think

0:35:56.000 --> 0:35:58.120
<v Speaker 1>what's also going to have to change is how we

0:35:58.239 --> 0:36:01.360
<v Speaker 1>as consumers of digital content think about what we see.

0:36:01.680 --> 0:36:04.719
<v Speaker 1>We are going to have to become more critical, more reasoned.

0:36:05.000 --> 0:36:07.239
<v Speaker 1>We have to get out of our echo chambers. We

0:36:07.320 --> 0:36:09.960
<v Speaker 1>have to stop allowing social media to manipulate us in

0:36:09.960 --> 0:36:12.040
<v Speaker 1>the way that they do. So I think the solution

0:36:12.120 --> 0:36:14.520
<v Speaker 1>is at least too prong and potentially three with some

0:36:14.680 --> 0:36:17.960
<v Speaker 1>legislative relief on the line to really force the companies

0:36:18.000 --> 0:36:19.879
<v Speaker 1>to do better than they have been over the last

0:36:19.920 --> 0:36:23.000
<v Speaker 1>few years. So does the good outweigh the bad? I

0:36:23.040 --> 0:36:25.840
<v Speaker 1>don't know. We have to have a hard conversation. People

0:36:25.880 --> 0:36:29.680
<v Speaker 1>in who work in infectious disease and physicists who develop weaponry,

0:36:30.000 --> 0:36:32.640
<v Speaker 1>they think about this all the time. We as technologies

0:36:32.719 --> 0:36:34.799
<v Speaker 1>have not quite thought about this as much in the

0:36:34.800 --> 0:36:37.359
<v Speaker 1>past because our field is so young. But I think

0:36:37.400 --> 0:36:39.759
<v Speaker 1>now you know it's time to wake up and start

0:36:39.840 --> 0:36:42.680
<v Speaker 1>asking those hard questions and having those conversations before it's

0:36:42.680 --> 0:36:47.920
<v Speaker 1>too late. Once again, we're being aged to wake up

0:36:47.960 --> 0:36:50.840
<v Speaker 1>from our sleep book, and we do have some months

0:36:50.840 --> 0:36:53.560
<v Speaker 1>is at least when it comes to deep fakes, we

0:36:53.600 --> 0:36:56.560
<v Speaker 1>can make it akin to counterfeiting money. The people who

0:36:56.560 --> 0:36:59.320
<v Speaker 1>do it will get prosecuted, and program is like Honey

0:36:59.400 --> 0:37:02.759
<v Speaker 1>will work on detection technology, but we still have to

0:37:02.800 --> 0:37:05.400
<v Speaker 1>hold the bills up to the light before we decide

0:37:05.440 --> 0:37:08.920
<v Speaker 1>whether to accept them. That's our job, that is, if

0:37:08.920 --> 0:37:12.040
<v Speaker 1>we're not too busy watching Nicholas Cage starring as Thelma

0:37:12.120 --> 0:37:16.719
<v Speaker 1>and Louise in Thelma and Louise. Even more complicated than

0:37:16.760 --> 0:37:20.600
<v Speaker 1>deep fakes is the concentration of power at companies like Facebook.

0:37:21.239 --> 0:37:23.560
<v Speaker 1>In the next episode, we visit a secret lab at

0:37:23.600 --> 0:37:27.800
<v Speaker 1>Google to understand what happens when technology companies start taking

0:37:27.800 --> 0:37:30.120
<v Speaker 1>on the role of the state, and we speak with

0:37:30.200 --> 0:37:33.400
<v Speaker 1>Lena Khan, who has proposed new regulation to balance the

0:37:33.440 --> 0:37:38.279
<v Speaker 1>power of big technology companies like Amazon. I'm asloshin See

0:37:38.280 --> 0:37:53.399
<v Speaker 1>you next time. Sleepwalkers is a production of I Heart

0:37:53.480 --> 0:37:58.120
<v Speaker 1>Radio and Unusual productions. For the latest AI news, live interviews,

0:37:58.160 --> 0:38:01.120
<v Speaker 1>and behind the scenes footage, find us on Instagram, at

0:38:01.120 --> 0:38:06.080
<v Speaker 1>Sleepwalker's podcast or at Sleepwalker's podcast dot com. Sleepwalkers is

0:38:06.120 --> 0:38:09.440
<v Speaker 1>hosted by me Ozveloshin and co hosted by me Kara Price.

0:38:09.600 --> 0:38:12.560
<v Speaker 1>Were produced by Julian Weller with help from Jacobo Penzo

0:38:12.680 --> 0:38:16.120
<v Speaker 1>and Taylor Chacogne. Mixing by Tristan McNeil and Julian Weller.

0:38:16.400 --> 0:38:20.120
<v Speaker 1>Recording assistance this episode from tofarrelf Our Story editor is

0:38:20.160 --> 0:38:24.640
<v Speaker 1>Matthew Riddle. Sleepwalkers is executive produced by me Ozveloshin and

0:38:24.719 --> 0:38:28.120
<v Speaker 1>Mangesh Hattiga. For more podcasts from My Heart Radio, visit

0:38:28.160 --> 0:38:30.920
<v Speaker 1>the I heart Radio app, Apple Podcasts, or wherever you

0:38:31.000 --> 0:38:32.240
<v Speaker 1>listen to your favorite shows.