WEBVTT - Fake News in the French Elections

0:00:00.200 --> 0:00:02.679
<v Speaker 1>Decrypted is brought to you by red Hat, whose broad

0:00:02.720 --> 0:00:06.080
<v Speaker 1>portfolio of open source technologies for the enterprise helps you

0:00:06.120 --> 0:00:08.320
<v Speaker 1>get from where you are to where you want to be.

0:00:09.119 --> 0:00:12.440
<v Speaker 1>Red Hat the open technology to help you realize your vision.

0:00:12.880 --> 0:00:15.640
<v Speaker 1>Learn more at red hat dot com slash open tech.

0:00:16.760 --> 0:00:19.599
<v Speaker 1>France is just days away from a crucial runoff in

0:00:19.640 --> 0:00:24.640
<v Speaker 1>its presidential campaign, very much like the last US presidential election.

0:00:24.920 --> 0:00:26.759
<v Speaker 1>It could go down in history. Is one of the

0:00:26.800 --> 0:00:32.200
<v Speaker 1>most consequential electoral decisions of this generation. So it looks

0:00:32.240 --> 0:00:35.320
<v Speaker 1>like in the past half we're seeing the numbers from

0:00:35.360 --> 0:00:38.440
<v Speaker 1>the actual votes, the actual const of the votes from

0:00:38.479 --> 0:00:43.239
<v Speaker 1>the Interior Ministry, showing that Immanuel mccon could actually be

0:00:43.400 --> 0:00:48.280
<v Speaker 1>leading twenty three or points three percent four Immanuel mccon.

0:00:49.920 --> 0:00:52.519
<v Speaker 1>The second and final round of voting is scheduled for

0:00:52.680 --> 0:00:57.040
<v Speaker 1>May seven. Candidates from France's mainstream political parties are all out,

0:00:57.600 --> 0:01:01.760
<v Speaker 1>leaving an independent Emmanuel Macon and Marie Lepin, leader of

0:01:01.800 --> 0:01:04.560
<v Speaker 1>the National Front, which has been known as a populist

0:01:04.680 --> 0:01:07.839
<v Speaker 1>right wing party. Our reporter in Paris, Marie ma Wad,

0:01:07.920 --> 0:01:11.200
<v Speaker 1>describes it like this, It's going from basically a boxing

0:01:11.200 --> 0:01:14.600
<v Speaker 1>match to a no holds barred ultimate fighting spectacle, if

0:01:14.600 --> 0:01:17.160
<v Speaker 1>you want to compare it that way. But as election

0:01:17.280 --> 0:01:20.800
<v Speaker 1>day approaches, we're getting a worrying case of deja vu.

0:01:21.160 --> 0:01:24.800
<v Speaker 1>Fake news stories are being read, liked, and shared online

0:01:24.800 --> 0:01:27.600
<v Speaker 1>in the critical final days as the voters make up

0:01:27.640 --> 0:01:30.680
<v Speaker 1>their mind about which candidates to back. Here's Maria again

0:01:31.280 --> 0:01:34.400
<v Speaker 1>one false report um which I think is my favorite

0:01:34.400 --> 0:01:38.160
<v Speaker 1>because it's the wackiest. It's about how the Lepans were

0:01:38.200 --> 0:01:42.000
<v Speaker 1>growing marijuana. According to this report, growing marijuana and their

0:01:42.040 --> 0:01:44.440
<v Speaker 1>basement to make a bit of cash to pay for

0:01:44.480 --> 0:01:48.440
<v Speaker 1>their campaign. After the US election, with a public outcry

0:01:48.520 --> 0:01:52.720
<v Speaker 1>over fake news, social media companies like Facebook, Twitter, and

0:01:52.800 --> 0:01:55.640
<v Speaker 1>Google said they were going to do something about it.

0:01:56.080 --> 0:02:08.320
<v Speaker 1>Have they? Hi, I'm Brad Stone and I'm Paga Cary.

0:02:08.840 --> 0:02:11.640
<v Speaker 1>And this week on Decrypted, we're taking another look at

0:02:11.680 --> 0:02:15.720
<v Speaker 1>the problem of fake news and whether changes at Facebook, Twitter,

0:02:15.760 --> 0:02:18.560
<v Speaker 1>and Google have been effective at curbing the reach of

0:02:18.639 --> 0:02:24.040
<v Speaker 1>misleading and sometimes outright fabricated stories. These companies are facing

0:02:24.080 --> 0:02:27.280
<v Speaker 1>their first major test now half a year after they

0:02:27.320 --> 0:02:30.000
<v Speaker 1>committed to tackling this issue. In the wake of the

0:02:30.080 --> 0:02:32.519
<v Speaker 1>U S elections, We'll take a look at how much

0:02:32.560 --> 0:02:35.240
<v Speaker 1>fake news is circulating in France ahead of the vote

0:02:35.480 --> 0:02:38.160
<v Speaker 1>and talk to expert fact checkers about the search for

0:02:38.240 --> 0:02:45.040
<v Speaker 1>a real solution. Stay with us. We've been covering the

0:02:45.080 --> 0:02:47.520
<v Speaker 1>issue of fake news for a few months now. It

0:02:47.600 --> 0:02:50.560
<v Speaker 1>first blew up as a major concern after the US election.

0:02:50.840 --> 0:02:54.040
<v Speaker 1>Regular listeners of this show may remember that we dedicated

0:02:54.040 --> 0:02:57.239
<v Speaker 1>an episode to the topic last year. For anyone who

0:02:57.240 --> 0:03:00.920
<v Speaker 1>needs a recap, basically, here's how it went. Mark Zuckerberg,

0:03:01.120 --> 0:03:05.920
<v Speaker 1>Facebook CEO, initially tried to brush off the issue. You know, personally,

0:03:06.919 --> 0:03:12.639
<v Speaker 1>I think the the idea that you know, fake news

0:03:12.919 --> 0:03:15.080
<v Speaker 1>on Facebook, of which you know it's a it's a

0:03:15.160 --> 0:03:20.240
<v Speaker 1>very small amount of of of the content UH influenced

0:03:20.560 --> 0:03:22.440
<v Speaker 1>in the election in any way, I think is a

0:03:22.480 --> 0:03:26.040
<v Speaker 1>pretty crazy idea. But very soon after Mark Zuckerberg made

0:03:26.040 --> 0:03:30.160
<v Speaker 1>those comments, Facebook started coming under enough pressure from enough

0:03:30.240 --> 0:03:33.720
<v Speaker 1>different quarters that pretty quickly the company had to backpedal.

0:03:34.160 --> 0:03:37.240
<v Speaker 1>Here's what Justin can, a tech investor and founder of

0:03:37.280 --> 0:03:40.080
<v Speaker 1>the streaming site Twitch, told us at the time. It's

0:03:40.080 --> 0:03:42.960
<v Speaker 1>a nonpartisan issue. I think people making decisions based on

0:03:43.040 --> 0:03:45.960
<v Speaker 1>fake news is not a good idea, right That's that

0:03:46.080 --> 0:03:49.200
<v Speaker 1>I think everyone can agree that that's actually going to

0:03:49.240 --> 0:03:53.119
<v Speaker 1>be bad for American society and discourse in general. Since then,

0:03:53.160 --> 0:03:56.280
<v Speaker 1>the main social networks have made a few changes. They

0:03:56.280 --> 0:03:59.840
<v Speaker 1>have partnered with local media on fact checking projects. Just

0:04:00.160 --> 0:04:03.680
<v Speaker 1>last week, Google made changes to its powerful search algorithm

0:04:03.720 --> 0:04:07.840
<v Speaker 1>intended to demote links to suspicious websites, and in France,

0:04:07.880 --> 0:04:11.680
<v Speaker 1>Facebook has deleted as many as thirty thousand accounts linked

0:04:11.760 --> 0:04:15.240
<v Speaker 1>to fake news. But there hasn't been a major political

0:04:15.280 --> 0:04:19.479
<v Speaker 1>event to test out these new policies until now. In France,

0:04:19.760 --> 0:04:22.760
<v Speaker 1>presidents get elected in two stages. The first round of

0:04:22.839 --> 0:04:26.360
<v Speaker 1>voting took place on April and narrowed the race down

0:04:26.400 --> 0:04:29.080
<v Speaker 1>from a field of eleven candidates to the final two.

0:04:29.720 --> 0:04:32.360
<v Speaker 1>In some ways, the dynamic playing out in France isn't

0:04:32.360 --> 0:04:35.440
<v Speaker 1>a million miles away from what US vote has experienced

0:04:35.560 --> 0:04:47.080
<v Speaker 1>last year. On one hand, there's the front runner, Emmanuel Macron.

0:05:00.200 --> 0:05:03.080
<v Speaker 1>Then there's Marine Lapan, who Macron is facing in the

0:05:03.120 --> 0:05:21.880
<v Speaker 1>final vote Las Islamists. We called Mary, who's our reporter

0:05:21.960 --> 0:05:24.080
<v Speaker 1>in Paris you heard from earlier, to give us a

0:05:24.120 --> 0:05:26.760
<v Speaker 1>little bit of background on these two. Now, these two

0:05:26.800 --> 0:05:32.039
<v Speaker 1>candidates disagree on almost everything. Lepin, on one hand, once

0:05:32.080 --> 0:05:35.880
<v Speaker 1>out of Europe, whereas Macon wants all in. Micon is

0:05:35.920 --> 0:05:38.760
<v Speaker 1>backing a bunch of free trade agreements, whereas lepan is

0:05:38.800 --> 0:05:42.400
<v Speaker 1>for protectionism. Marine Leapin has been a figure on Frances

0:05:42.440 --> 0:05:45.400
<v Speaker 1>far right for some time. A victory for Marine is

0:05:45.440 --> 0:05:48.120
<v Speaker 1>currently seen as a long shot, but after Brexit in

0:05:48.160 --> 0:05:52.320
<v Speaker 1>the election of Donald Trump, anything seems possible. Leapin is

0:05:52.360 --> 0:05:57.080
<v Speaker 1>definitely an anti euro and anti European candidate and that

0:05:57.440 --> 0:06:00.520
<v Speaker 1>helped her when the favors of Russia. But who wasn't

0:06:00.560 --> 0:06:03.840
<v Speaker 1>the only one in the French election who has been

0:06:03.839 --> 0:06:07.960
<v Speaker 1>calling for a sort of closer tie. But what has

0:06:08.040 --> 0:06:12.880
<v Speaker 1>tied mainly been closest to to Russia is really this

0:06:12.880 --> 0:06:16.479
<v Speaker 1>this sort of explicit um move that she made by

0:06:16.680 --> 0:06:19.680
<v Speaker 1>paying visit to put In. And it's specifically in this

0:06:19.880 --> 0:06:23.440
<v Speaker 1>context of populism that we're seeing fake news emerge in

0:06:23.480 --> 0:06:26.640
<v Speaker 1>France just the way it did in the US. Here's

0:06:26.680 --> 0:06:30.320
<v Speaker 1>Jonathan Ditsch, he's the CEO of Bacamo Social and he's

0:06:30.320 --> 0:06:33.840
<v Speaker 1>been tracking online discussions relating to the French election. In

0:06:34.040 --> 0:06:37.839
<v Speaker 1>just the past several months, we've seen h a quarter

0:06:37.960 --> 0:06:40.560
<v Speaker 1>of a million, I think a quarter of a million

0:06:41.839 --> 0:06:45.320
<v Speaker 1>references in French social media just with the keyword fake

0:06:45.360 --> 0:06:49.600
<v Speaker 1>news um, and that would be those references be mentioned

0:06:49.680 --> 0:06:52.919
<v Speaker 1>like in Facebook and Twitter or something like that. So

0:06:53.080 --> 0:06:55.279
<v Speaker 1>if you imagine, those are just the people talking about

0:06:55.320 --> 0:06:57.720
<v Speaker 1>it and then the reach that goes out from that.

0:06:58.600 --> 0:07:01.360
<v Speaker 1>But that's just what the traditional media outlets are saying

0:07:01.480 --> 0:07:05.000
<v Speaker 1>about fake news. Here's how much actual content he's seeing.

0:07:05.520 --> 0:07:08.520
<v Speaker 1>When we measured this for the time period between the

0:07:08.560 --> 0:07:12.200
<v Speaker 1>beginning of November and a couple of days into April,

0:07:12.280 --> 0:07:17.520
<v Speaker 1>that reframe and alternative section encompassed about one quarter of

0:07:17.640 --> 0:07:25.880
<v Speaker 1>all of the links shared during that time. Just for

0:07:25.920 --> 0:07:30.600
<v Speaker 1>context there, reframe and alternative content is what Jonathan classifies

0:07:30.680 --> 0:07:35.000
<v Speaker 1>as potentially fake or deliberately misleading news that's opposed to

0:07:35.040 --> 0:07:38.400
<v Speaker 1>other kinds of content like news articles from traditional news

0:07:38.480 --> 0:07:42.760
<v Speaker 1>organizations or material coming directly from the campaign. The words

0:07:42.840 --> 0:07:46.920
<v Speaker 1>reframe an alternative are are euphemisms for something that's very dangerous,

0:07:46.920 --> 0:07:50.120
<v Speaker 1>and the fact that Jonathan dis just saying that perhaps

0:07:50.160 --> 0:07:53.600
<v Speaker 1>one quarter of all links shared during this time are

0:07:54.000 --> 0:07:57.280
<v Speaker 1>fake news is pretty scary. And on top of that,

0:07:57.640 --> 0:08:00.960
<v Speaker 1>Jonathan says a really significant portion of fake or junk

0:08:01.000 --> 0:08:04.600
<v Speaker 1>news stories relating to the French election have some kind

0:08:04.640 --> 0:08:10.480
<v Speaker 1>of link to Russian websites and news sources. We characterize

0:08:11.160 --> 0:08:14.360
<v Speaker 1>a site as having Russian influence if it links to

0:08:14.680 --> 0:08:18.040
<v Speaker 1>Russia Today or spot nick News or even maybe one

0:08:18.040 --> 0:08:21.120
<v Speaker 1>of the Russian newspapers. I mean, those are all indications

0:08:21.160 --> 0:08:24.760
<v Speaker 1>that the site may have Russian influence. We also see

0:08:25.360 --> 0:08:28.320
<v Speaker 1>in the sites that are in the reframing section it's

0:08:28.360 --> 0:08:33.600
<v Speaker 1>about them, that that have links to Russian driven sources.

0:08:34.360 --> 0:08:42.079
<v Speaker 1>In what we call the alternative section, it's about them.

0:08:42.120 --> 0:08:44.599
<v Speaker 1>It's hard not to want to know is this a

0:08:44.679 --> 0:08:47.959
<v Speaker 1>delivered strategy from the Russians, What if any effect is

0:08:48.000 --> 0:08:52.280
<v Speaker 1>it having on voters? And the truth is it's really

0:08:52.280 --> 0:08:55.240
<v Speaker 1>hard to say. It's hard to hard to hard to

0:08:55.280 --> 0:08:58.320
<v Speaker 1>really know that for sure. But we do know that

0:08:58.720 --> 0:09:04.440
<v Speaker 1>this content is being published in French around topics that

0:09:04.480 --> 0:09:10.200
<v Speaker 1>are relevant for the French election. And is that sufficient

0:09:10.240 --> 0:09:13.560
<v Speaker 1>to say it? I don't know, But it's almost impossible

0:09:13.640 --> 0:09:18.439
<v Speaker 1>to distinguish whether that's really overt action or it's just

0:09:19.440 --> 0:09:22.200
<v Speaker 1>a sympathy, if you will, for that point of view,

0:09:23.120 --> 0:09:26.439
<v Speaker 1>as in the US election, there's a lot of smoke here. Clearly,

0:09:26.600 --> 0:09:30.080
<v Speaker 1>Russian President Vladimir Putin has a stake in the French elections.

0:09:30.280 --> 0:09:33.560
<v Speaker 1>He'd like to see Western Europe further destabilized, which helps

0:09:33.600 --> 0:09:36.840
<v Speaker 1>him extend Russia's power and his influence on the world stage.

0:09:37.920 --> 0:09:40.560
<v Speaker 1>But even if it's not connected to Russia, it appears

0:09:40.600 --> 0:09:44.240
<v Speaker 1>that we fall for these fake news stories pretty easily,

0:09:44.720 --> 0:09:46.960
<v Speaker 1>and it doesn't take much for them to go viral.

0:09:47.400 --> 0:09:50.600
<v Speaker 1>He's marry again. Most of these fake posts, and most

0:09:50.600 --> 0:09:54.199
<v Speaker 1>of these hoaxes or stories that look plausible or are

0:09:54.280 --> 0:09:58.080
<v Speaker 1>meant to look plausible, um, they're very often a deformed

0:09:58.160 --> 0:10:00.800
<v Speaker 1>version of a real fact. And the all there, of course,

0:10:00.880 --> 0:10:04.680
<v Speaker 1>is that people who are not ill intentioned will fall

0:10:04.760 --> 0:10:07.280
<v Speaker 1>for them and they'll start sharing them around because it

0:10:07.320 --> 0:10:11.559
<v Speaker 1>looks like a rumor that could maybe be true. But Brad,

0:10:11.600 --> 0:10:14.160
<v Speaker 1>if you think a quarter of all links sounds like

0:10:14.200 --> 0:10:16.400
<v Speaker 1>a large number, and I did while I was reporting

0:10:16.440 --> 0:10:19.320
<v Speaker 1>this story, just wait until you hear how much fake

0:10:19.400 --> 0:10:23.840
<v Speaker 1>news US voters were exposed to before the election. We'll

0:10:23.840 --> 0:10:32.600
<v Speaker 1>be right back after this word from our sponsor. You

0:10:32.679 --> 0:10:35.160
<v Speaker 1>know where you want to be. Red Hat has the

0:10:35.200 --> 0:10:38.520
<v Speaker 1>broad portfolio of open source technologies to get you there.

0:10:38.960 --> 0:10:42.840
<v Speaker 1>Meet your evolving business challenges head on with secure solutions

0:10:42.880 --> 0:10:48.320
<v Speaker 1>for the enterprise, including Linux platforms and containers, hybrid cloud infrastructure,

0:10:48.640 --> 0:10:54.200
<v Speaker 1>application integration and development, operations management, and beyond. Visit red

0:10:54.240 --> 0:10:57.480
<v Speaker 1>hat dot com slash open tech to learn more. Red

0:10:57.520 --> 0:11:00.920
<v Speaker 1>Hat the open technology to help you wise your vision.

0:11:09.240 --> 0:11:11.600
<v Speaker 1>Before the break, we heard there's actually a lot of

0:11:11.600 --> 0:11:14.880
<v Speaker 1>fake or misleading news being shared on French social media

0:11:15.160 --> 0:11:18.120
<v Speaker 1>just days before the country has to choose its next president.

0:11:18.800 --> 0:11:21.439
<v Speaker 1>But even though the numbers in France may seem high,

0:11:21.920 --> 0:11:24.600
<v Speaker 1>it turns out it's about half the amount of fake

0:11:24.640 --> 0:11:27.720
<v Speaker 1>news that US voters were exposed to before the November

0:11:27.720 --> 0:11:32.920
<v Speaker 1>twenty sixteen presidential election. My name is clem up in Disigo,

0:11:33.200 --> 0:11:37.199
<v Speaker 1>and I'm at the Oxford Internet Institute at the University

0:11:37.200 --> 0:11:41.880
<v Speaker 1>of Oxford. Clementine has been working on what she calls

0:11:42.040 --> 0:11:44.880
<v Speaker 1>junk news. That's a kind of catch all term for

0:11:45.000 --> 0:11:49.480
<v Speaker 1>news that is definitely misleading, if not entirely fake. The

0:11:49.600 --> 0:11:53.080
<v Speaker 1>free count with the US election, For instance, we studied

0:11:53.160 --> 0:11:56.960
<v Speaker 1>the case of Michigan UM. In Michigan, we found as

0:11:57.080 --> 0:12:00.199
<v Speaker 1>much fake news as professional news share them tweet tour

0:12:00.320 --> 0:12:03.320
<v Speaker 1>in the days leading up to the US election. So

0:12:03.520 --> 0:12:07.400
<v Speaker 1>compared to that, UM and in French voters were sharing

0:12:07.880 --> 0:12:10.560
<v Speaker 1>way bitter credit to news on Twitter than the U

0:12:10.640 --> 0:12:15.360
<v Speaker 1>s vooters. That is as much fake news as professional news.

0:12:15.880 --> 0:12:21.000
<v Speaker 1>So that's one junk news article for every credible story

0:12:21.559 --> 0:12:24.120
<v Speaker 1>oh Man. As a journalist, I find that to be

0:12:24.200 --> 0:12:27.439
<v Speaker 1>very distressing. Why was it so much worse than the US. Well,

0:12:27.480 --> 0:12:30.160
<v Speaker 1>for one thing, social media companies have been taking the

0:12:30.200 --> 0:12:33.800
<v Speaker 1>fake news issue more seriously, so it's possible that some

0:12:33.840 --> 0:12:37.560
<v Speaker 1>of their efforts are actually helping right. Facebook and Google

0:12:37.559 --> 0:12:39.920
<v Speaker 1>have introduced a whole host of new measures since the

0:12:40.000 --> 0:12:42.599
<v Speaker 1>US election to do things like limit the amount of

0:12:42.679 --> 0:12:46.200
<v Speaker 1>money fake news sites can make through advertising. Google took

0:12:46.240 --> 0:12:51.400
<v Speaker 1>a few different steps um. The first was to try

0:12:51.440 --> 0:12:54.680
<v Speaker 1>and remove as many fake news types as it could

0:12:54.840 --> 0:12:58.760
<v Speaker 1>from Google ad sence UM, which is the way that

0:12:58.800 --> 0:13:02.600
<v Speaker 1>a lot of these people made money right by loading

0:13:02.800 --> 0:13:07.319
<v Speaker 1>their their fake news sides with with ads. That's Alexios Manzarles.

0:13:07.720 --> 0:13:10.520
<v Speaker 1>I'm the director of the International fact Checking Network at

0:13:10.559 --> 0:13:15.880
<v Speaker 1>pointer UM. We are a loose alliance of hundred or

0:13:15.960 --> 0:13:19.440
<v Speaker 1>so fact checking organizations around the world. Alexios has also

0:13:19.480 --> 0:13:25.040
<v Speaker 1>been working with Facebook. Facebook after much pr shaming, came

0:13:25.080 --> 0:13:29.400
<v Speaker 1>to a decision in December UM that both sort of

0:13:29.600 --> 0:13:33.679
<v Speaker 1>attacked the financial side of it UM by making sure

0:13:33.920 --> 0:13:36.280
<v Speaker 1>or trying to make sure that fake news sides couldn't

0:13:36.320 --> 0:13:40.199
<v Speaker 1>boost UM, they couldn't use their advertising assistance, but also

0:13:40.240 --> 0:13:44.080
<v Speaker 1>couldn't boost posts on Facebook. Facebook has been doing more

0:13:44.120 --> 0:13:47.640
<v Speaker 1>than just targeting ad revenue. They've developed tools to help

0:13:47.760 --> 0:13:49.960
<v Speaker 1>us a spot fake news when it appears in their

0:13:50.000 --> 0:13:53.200
<v Speaker 1>news feed. We know that people don't want to be

0:13:53.280 --> 0:13:57.640
<v Speaker 1>lied to or deceived on our platform, and that is

0:13:57.640 --> 0:14:01.640
<v Speaker 1>a role we take a d percent responsibility for. This

0:14:01.720 --> 0:14:06.040
<v Speaker 1>is Chris Cox, Facebook's chief product officer. He's addressing a

0:14:06.080 --> 0:14:09.800
<v Speaker 1>crowd of people at Facebook's developer conference called F eight.

0:14:10.320 --> 0:14:13.240
<v Speaker 1>This was launched in France. It's called Perspectives. So if

0:14:13.280 --> 0:14:15.280
<v Speaker 1>you find a story and feed, or you read an

0:14:15.320 --> 0:14:18.640
<v Speaker 1>article about a topic, when you return, you're presented with

0:14:18.679 --> 0:14:20.800
<v Speaker 1>the points of view of all of the French candidates

0:14:20.840 --> 0:14:23.760
<v Speaker 1>on that topic. This isn't just the candidate that you

0:14:24.240 --> 0:14:27.320
<v Speaker 1>are you've connected with. This is all the candidates and

0:14:27.360 --> 0:14:29.640
<v Speaker 1>it's a pattern that we think is super interesting because

0:14:29.640 --> 0:14:33.040
<v Speaker 1>it's presenting a complete story. Because this feature has been

0:14:33.080 --> 0:14:35.880
<v Speaker 1>working well in France, Facebook is looking at making it

0:14:35.960 --> 0:14:39.240
<v Speaker 1>more widely available. Google has also made changes to its

0:14:39.240 --> 0:14:42.520
<v Speaker 1>search results. In particular, there's a new fact check tag

0:14:42.560 --> 0:14:45.960
<v Speaker 1>that publishers can add to news stories, and it's supposed

0:14:46.000 --> 0:14:49.080
<v Speaker 1>to be visible in the list of search results. Here's Marie.

0:14:49.640 --> 0:14:53.400
<v Speaker 1>Facebook and Google. They've teamed up with local newspapers in

0:14:53.440 --> 0:14:57.720
<v Speaker 1>a project called cross Check, which spots hoaxes and flexed

0:14:57.800 --> 0:15:00.720
<v Speaker 1>them to readers online with the help of local journalists.

0:15:01.040 --> 0:15:03.480
<v Speaker 1>So far, it seems that initiatives like cross check have

0:15:03.560 --> 0:15:06.440
<v Speaker 1>been well received, but I do have some concerns about

0:15:06.440 --> 0:15:09.120
<v Speaker 1>whether these kinds of projects can really address the fake

0:15:09.160 --> 0:15:12.600
<v Speaker 1>news issue at scale. A story could go viral in

0:15:12.720 --> 0:15:15.600
<v Speaker 1>minutes or hours, but fact checking is a more time

0:15:15.640 --> 0:15:19.560
<v Speaker 1>consuming process. By the time a story gets debunked, thousands,

0:15:19.600 --> 0:15:22.320
<v Speaker 1>maybe tens of thousands of people might have already read

0:15:22.360 --> 0:15:25.520
<v Speaker 1>it and believed it. Cross Check spotted all the fakes

0:15:25.600 --> 0:15:28.880
<v Speaker 1>that I previously talked about, and I guess that attests

0:15:28.920 --> 0:15:33.160
<v Speaker 1>to some success of that initiative. Of course, the challenge

0:15:33.240 --> 0:15:36.520
<v Speaker 1>for social media is getting rid of a fake story

0:15:36.680 --> 0:15:39.880
<v Speaker 1>quickly enough that it doesn't leave a mark in voters

0:15:39.960 --> 0:15:44.840
<v Speaker 1>minds after making the rounds. And that's not all, Alexio says.

0:15:44.880 --> 0:15:49.120
<v Speaker 1>There's some evidence that articles debunking fake news don't actually

0:15:49.160 --> 0:15:51.400
<v Speaker 1>get back to that same group of people who read

0:15:51.480 --> 0:15:54.720
<v Speaker 1>the false story in the first place. I think we've

0:15:54.720 --> 0:15:59.680
<v Speaker 1>seen um some indicating that hoax has spread faster and

0:15:59.800 --> 0:16:02.680
<v Speaker 1>the neer than their corrections. M but I haven't seen

0:16:02.720 --> 0:16:10.480
<v Speaker 1>anything truly reliable on who gets fake news and who doesn't.

0:16:10.960 --> 0:16:13.560
<v Speaker 1>There's been some work from Balta Coot, Chalky at the I,

0:16:13.680 --> 0:16:15.960
<v Speaker 1>M T and Luka that claims that these are really

0:16:16.000 --> 0:16:20.120
<v Speaker 1>two separate audiences. Um, I don't you know. I I

0:16:20.120 --> 0:16:22.120
<v Speaker 1>could see that being the case, but I think we

0:16:22.160 --> 0:16:26.800
<v Speaker 1>need more and more research to prove that. Another issue

0:16:26.880 --> 0:16:30.480
<v Speaker 1>with the spread of fake news is bought. These are

0:16:30.520 --> 0:16:34.080
<v Speaker 1>automated accounts that can spue out thousands of tweets or

0:16:34.160 --> 0:16:38.520
<v Speaker 1>posts in a single day. Facebook recently deleted thirty thou

0:16:38.800 --> 0:16:42.080
<v Speaker 1>of these accounts, and Twitter has been active here too.

0:16:42.760 --> 0:16:47.800
<v Speaker 1>Twitter is uh is quite responsive to these, to these

0:16:47.880 --> 0:16:52.880
<v Speaker 1>questions of of chank news circulating over the platform. UM

0:16:53.280 --> 0:16:56.800
<v Speaker 1>so yeah, it's also I mean, the question is also

0:16:57.400 --> 0:17:03.760
<v Speaker 1>whether you can automate the detection of faith news. Um

0:17:03.960 --> 0:17:06.880
<v Speaker 1>so this is also something that the project is going on,

0:17:07.080 --> 0:17:13.440
<v Speaker 1>or you can automately automatically sign votes and automated accounts

0:17:13.560 --> 0:17:17.119
<v Speaker 1>in order to to take a decision maybe to close

0:17:17.200 --> 0:17:21.000
<v Speaker 1>the accounts. That was Clementine. Again. At the same time,

0:17:21.200 --> 0:17:23.480
<v Speaker 1>we do have to be careful. It's too soon to

0:17:23.560 --> 0:17:26.320
<v Speaker 1>really know how much of an impact these measures are having.

0:17:26.720 --> 0:17:30.119
<v Speaker 1>As Alexio said, it hasn't yet been studied carefully, so

0:17:30.160 --> 0:17:34.480
<v Speaker 1>we can't rule out this one simpler explanation. The term

0:17:34.520 --> 0:17:38.679
<v Speaker 1>faith news didn't exist in the French social media dialogue

0:17:39.160 --> 0:17:44.480
<v Speaker 1>before roughly the end of the US elections, and now

0:17:44.600 --> 0:17:50.520
<v Speaker 1>it has come out of nowhere to represent a massive,

0:17:51.160 --> 0:17:56.679
<v Speaker 1>massive part of the discussion. That's Jonathan again. He's the

0:17:56.840 --> 0:18:00.720
<v Speaker 1>consultant in Paris who studies online conversations. Right. Part of

0:18:00.760 --> 0:18:02.919
<v Speaker 1>the answer might be that we're just more aware of

0:18:02.960 --> 0:18:05.040
<v Speaker 1>the issue now, so perhaps we're less likely to be

0:18:05.160 --> 0:18:07.920
<v Speaker 1>taken in by a fake news story. And the answer

0:18:08.040 --> 0:18:10.560
<v Speaker 1>probably also has something to do with how much the

0:18:10.560 --> 0:18:14.760
<v Speaker 1>public trusts their media institutions. Here in the US. Surveys

0:18:14.800 --> 0:18:17.879
<v Speaker 1>have suggested that number is depressingly low. But there is

0:18:17.920 --> 0:18:21.480
<v Speaker 1>still this tricky issue of foreign meddling in elections. Because

0:18:21.480 --> 0:18:23.440
<v Speaker 1>it can be so hard to figure out who first

0:18:23.480 --> 0:18:26.280
<v Speaker 1>published a story or where it came from, it's easy

0:18:26.359 --> 0:18:29.920
<v Speaker 1>for websites associated with foreign governments to circulate their version

0:18:29.960 --> 0:18:33.880
<v Speaker 1>of the story. Here's Jonathan again. We've also seen stories

0:18:33.960 --> 0:18:38.440
<v Speaker 1>which seemed to emanate from foreign sources that are designed

0:18:38.480 --> 0:18:42.920
<v Speaker 1>to to try to create a more sympathetic point of view. Um.

0:18:42.960 --> 0:18:47.520
<v Speaker 1>You know, to take some of the stories about about

0:18:47.600 --> 0:18:53.960
<v Speaker 1>Macolm's alleged hidden accounts overseas banking. This is something that's

0:18:53.960 --> 0:18:58.119
<v Speaker 1>been rubbished um by traditional media, and yet it continues

0:18:58.160 --> 0:19:03.680
<v Speaker 1>to exist uh in the reframing in alternative media sources.

0:19:06.240 --> 0:19:09.399
<v Speaker 1>There are allegations of more substantial Russian medaling in the

0:19:09.400 --> 0:19:12.960
<v Speaker 1>French elections too. For example, the reports that Russian max

0:19:13.080 --> 0:19:17.040
<v Speaker 1>cyber groups managed to breach the Mark Kron campaign. That's

0:19:17.080 --> 0:19:19.640
<v Speaker 1>not unlike what happened last year when Russian hack has

0:19:19.680 --> 0:19:24.200
<v Speaker 1>infiltrated the Democratic National Committee. Here's our reporter in Paris, Marie.

0:19:25.200 --> 0:19:27.600
<v Speaker 1>Some experts have said, you know, it's hard to pin

0:19:27.680 --> 0:19:31.320
<v Speaker 1>down who's behind these cyber attacks, but they look like

0:19:31.400 --> 0:19:36.919
<v Speaker 1>they've been perpetrated by the same group that tried to

0:19:37.080 --> 0:19:41.240
<v Speaker 1>infiltrate the Democratic Party in the US, and a group

0:19:41.280 --> 0:19:44.679
<v Speaker 1>called the pond Storm, which some people have said is

0:19:44.760 --> 0:19:55.000
<v Speaker 1>linked to Russian intelligence services. Getting a handle on fake

0:19:55.040 --> 0:19:58.240
<v Speaker 1>news isn't just important for France. There's a general election

0:19:58.320 --> 0:20:00.960
<v Speaker 1>coming up in June in the UK and then Germany

0:20:01.040 --> 0:20:05.000
<v Speaker 1>federal elections of scheduled for September plus, there is a

0:20:05.040 --> 0:20:08.320
<v Speaker 1>ton of fake news that has nothing to do with politics.

0:20:08.800 --> 0:20:11.119
<v Speaker 1>Watch out for the health advice you're reading about on

0:20:11.119 --> 0:20:14.480
<v Speaker 1>the web. There is definitely fake news and other topics,

0:20:14.480 --> 0:20:19.159
<v Speaker 1>and I would think that health specialists have a particularly

0:20:19.160 --> 0:20:24.800
<v Speaker 1>strong grip on how widespread fake news about um cures

0:20:24.920 --> 0:20:28.600
<v Speaker 1>or or alleged cures to all kinds of ailments are

0:20:28.640 --> 0:20:31.119
<v Speaker 1>on the internet. I remember seeing some of these on Zeka,

0:20:31.720 --> 0:20:35.840
<v Speaker 1>for instance. But bringing it back to politics, there are

0:20:35.840 --> 0:20:39.040
<v Speaker 1>some encouraging signs that the public is becoming more aware

0:20:39.080 --> 0:20:42.720
<v Speaker 1>of what they're reading and sharing online. I asked Jonathan

0:20:42.760 --> 0:20:45.240
<v Speaker 1>whether he saw a spike in the number of fake

0:20:45.359 --> 0:20:48.240
<v Speaker 1>news articles being shared before the first round of voting

0:20:48.280 --> 0:20:53.360
<v Speaker 1>in France, which took place on April, and he told me, actually,

0:20:53.359 --> 0:20:55.840
<v Speaker 1>we saw the opposite. What we saw is that there

0:20:55.920 --> 0:21:02.440
<v Speaker 1>is a greater sharing of traditional and campaign UH links

0:21:02.440 --> 0:21:06.600
<v Speaker 1>from the traditional media and campaign sources over the month

0:21:06.640 --> 0:21:09.720
<v Speaker 1>of April UH. And that's actually been a fairly steady

0:21:09.840 --> 0:21:12.919
<v Speaker 1>increase over the past couple of months. And so what

0:21:13.000 --> 0:21:15.600
<v Speaker 1>that means now is, whereas it might have been about

0:21:15.840 --> 0:21:18.439
<v Speaker 1>one in every four links was from these reframing or

0:21:18.560 --> 0:21:22.320
<v Speaker 1>alternative sites, now it's just under one in every five.

0:21:23.400 --> 0:21:26.680
<v Speaker 1>Jonathan says he's seeing more articles being shared on average

0:21:26.680 --> 0:21:30.720
<v Speaker 1>per account, and he's seeing more accounts sharing articles, so

0:21:30.840 --> 0:21:34.280
<v Speaker 1>traditional media stories are rising as a portion of the total,

0:21:34.960 --> 0:21:37.760
<v Speaker 1>and platforms like Facebook and Google will be refining their

0:21:37.760 --> 0:21:40.840
<v Speaker 1>tools all the time. For example, spamming used to be

0:21:40.920 --> 0:21:43.480
<v Speaker 1>common on both sites. It's loss of a problem today,

0:21:43.640 --> 0:21:45.760
<v Speaker 1>or at least we're all better at ignoring it, And

0:21:45.800 --> 0:21:48.439
<v Speaker 1>it seems like both companies are tackling fake news the

0:21:48.480 --> 0:21:52.560
<v Speaker 1>same way. Although interestingly, Alexios was very clear that while

0:21:52.760 --> 0:21:55.440
<v Speaker 1>a lot of the content is being distributed on Facebook

0:21:55.480 --> 0:21:58.600
<v Speaker 1>and Google, um so therefore they have an important stake

0:21:58.640 --> 0:22:01.480
<v Speaker 1>in the outcome, these companies shouldn't have to take all

0:22:01.520 --> 0:22:06.240
<v Speaker 1>the responsibility. But this is really a job for everyone.

0:22:06.440 --> 0:22:09.360
<v Speaker 1>It's a job for high school teachers who need to

0:22:09.359 --> 0:22:13.320
<v Speaker 1>teach the digitally native not to be digitally naive. It's

0:22:13.359 --> 0:22:15.280
<v Speaker 1>a job for fact checkers who need to make sure

0:22:15.359 --> 0:22:19.200
<v Speaker 1>they're reaching beyond their bubbles. It's a job for media

0:22:19.280 --> 0:22:22.960
<v Speaker 1>organizations to double up, double down, sorry on their own

0:22:22.960 --> 0:22:26.000
<v Speaker 1>fact checking processes before they published things, and to be

0:22:26.040 --> 0:22:28.800
<v Speaker 1>more honest about corrections. So I do think this is

0:22:28.840 --> 0:22:32.480
<v Speaker 1>not there's no silver bullet, and Jonathan had similar views.

0:22:32.800 --> 0:22:35.200
<v Speaker 1>He put it to me like this, The real reason

0:22:35.280 --> 0:22:38.720
<v Speaker 1>why people are susceptible to this is because they don't

0:22:38.800 --> 0:22:41.280
<v Speaker 1>feel like their stories being told and they don't feel

0:22:41.320 --> 0:22:45.640
<v Speaker 1>like they're hearing something that's relevant to them. P I. I

0:22:45.440 --> 0:22:48.800
<v Speaker 1>I do agree with Alexios and Jonathan. It is everybody's

0:22:48.840 --> 0:22:52.000
<v Speaker 1>responsibility to get educated and maybe be a little bit

0:22:52.040 --> 0:22:55.000
<v Speaker 1>skeptical about the stories they read. But I gotta say

0:22:55.040 --> 0:22:58.640
<v Speaker 1>I feel like Google and Facebook have an extra responsibility here.

0:22:59.040 --> 0:23:03.159
<v Speaker 1>Their success has destabilized the traditional media. We've seen it

0:23:03.160 --> 0:23:06.199
<v Speaker 1>all across the world, and you know the fact that

0:23:06.280 --> 0:23:09.919
<v Speaker 1>it is open to vacuum for these less uh you know,

0:23:10.200 --> 0:23:12.479
<v Speaker 1>trusted news sources to come in and then in some

0:23:12.520 --> 0:23:16.000
<v Speaker 1>cases mislead people. It's tough to handle, particularly being a

0:23:16.040 --> 0:23:18.159
<v Speaker 1>member of the press. I think these companies, you know,

0:23:18.200 --> 0:23:20.880
<v Speaker 1>they've taken a passive attitude. They've gone and said, well,

0:23:20.920 --> 0:23:23.960
<v Speaker 1>we're going to address bad things when we see them,

0:23:23.960 --> 0:23:26.399
<v Speaker 1>we'll get them off the network. And that's frankly not

0:23:26.480 --> 0:23:29.439
<v Speaker 1>good enough. The velocity of news and sharing moves so

0:23:29.560 --> 0:23:31.880
<v Speaker 1>quickly now you almost have to prevent it from ever

0:23:31.960 --> 0:23:34.159
<v Speaker 1>taking hold in the first place, and we've seen the

0:23:34.200 --> 0:23:37.320
<v Speaker 1>destructive outcome that fake news can have. I think I

0:23:37.359 --> 0:23:39.879
<v Speaker 1>agree with you. We wouldn't tolerate this kind of content

0:23:39.960 --> 0:23:43.240
<v Speaker 1>being distributed on any other platform. So um, I think

0:23:43.280 --> 0:23:45.960
<v Speaker 1>if we're applying standards across the board, the web needs

0:23:46.000 --> 0:23:48.640
<v Speaker 1>to catch up to whatever the outcome in France, one

0:23:48.680 --> 0:23:51.320
<v Speaker 1>hopes that people are making their decisions based on real

0:23:51.440 --> 0:24:00.359
<v Speaker 1>news instead of date news. And with that, that's it

0:24:00.440 --> 0:24:03.600
<v Speaker 1>for this week's Decrypted. Thanks for listening. We'd love to

0:24:03.640 --> 0:24:06.080
<v Speaker 1>hear what you thought of this episode. You can record

0:24:06.080 --> 0:24:08.920
<v Speaker 1>a voice message and send it to us at Decrypted

0:24:09.240 --> 0:24:13.040
<v Speaker 1>at Bloomberg dot net or I'm on Twitter at Pia

0:24:13.119 --> 0:24:16.720
<v Speaker 1>Gattari and I'm at brad Stone. If you haven't already,

0:24:16.880 --> 0:24:20.160
<v Speaker 1>please subscribe to our show on Apple podcast or wherever

0:24:20.200 --> 0:24:23.119
<v Speaker 1>you normally listen. While you're there, please leave us a

0:24:23.240 --> 0:24:26.680
<v Speaker 1>rating and review. It really helps more listeners find the show.

0:24:28.000 --> 0:24:32.520
<v Speaker 1>This episode was produced by Akuta, Liz Smith, and Magnus Hendrickson.

0:24:33.000 --> 0:24:36.360
<v Speaker 1>A big thanks to Marie ma Wad for contributing from Paris,

0:24:36.440 --> 0:24:39.239
<v Speaker 1>and to our new reporter Nico Grant, who helped with

0:24:39.320 --> 0:24:42.280
<v Speaker 1>a big chunk of the reporting and research for today's show,

0:24:42.800 --> 0:24:46.159
<v Speaker 1>Alec McCabe is head of Bloomberg Podcasts. We'll see you

0:24:46.200 --> 0:24:52.160
<v Speaker 1>next week. Decrypted is brought to you by red Hat,

0:24:52.280 --> 0:24:55.800
<v Speaker 1>whose broad portfolio of open source technologies for the enterprise

0:24:55.880 --> 0:24:58.000
<v Speaker 1>helps you get from where you are to where you

0:24:58.040 --> 0:25:01.440
<v Speaker 1>want to be. Red Hat the open technology to help

0:25:01.480 --> 0:25:04.680
<v Speaker 1>you realize your vision. Learn more at red hat dot com,

0:25:04.800 --> 0:25:05.760
<v Speaker 1>slash opent tech