Oumu Ly: Welcome to The Breakdown. My name is Oumou, I’m a fel­low at the Berkman Klein Center on the Assembly Disinformation Program. I am real­ly excit­ed to be joined today by Naima Green-Riley. Naima’s a PhD can­di­date at the Department of Government at Harvard University, with a par­tic­u­lar focus on pub­lic diplo­ma­cy and the glob­al infor­ma­tion space. She also was for­mer­ly a for­eign ser­vice offi­cer and a Pickering fel­low. Welcome Naima. Thanks so much for joining.

Naima Green-Riley: Well thank you so much for hav­ing me.

Ly: Thank you. So, our con­ver­sa­tion today cen­ters on for­eign inter­fer­ence in the upcom­ing elec­tion which is draw­ing real­ly real­ly close. At the time of this record­ing we’re about two weeks out from November 3rd. And a few of the big top­ics on my mind today, Naima, are you know, sort of one, the sort of big threat actors this time around. We know that 2016 was sort of a water­shed moment in terms of for­eign inter­fer­ence for American demo­c­ra­t­ic process­es. In terms of social media manip­u­la­tion in par­tic­u­lar, how do for­eign influ­ence efforts in 2020 look in con­trast to active mea­sures we saw in 2016? Maybe have the pri­ma­ry threat actors changed, opti­mized their meth­ods a lit­tle bit, or adopt­ed over­all new approach­es to influ­enc­ing pub­lic opinion.

Green: Well, you’re def­i­nite­ly right that 2016 marked the first time that the US start­ed to real­ly pay atten­tion to this type of online for­eign influ­ence activ­i­ty. And dur­ing that elec­tion year we saw a series of coor­di­nat­ed social media cam­paigns tar­get­ing var­i­ous groups of indi­vid­u­als in the United States and seek­ing to influ­ence their polit­i­cal thoughts and behavior. 

The cam­paigns were focused on sow­ing dis­cord in US pol­i­tics main­ly, by dri­ving a wedge between peo­ple on very polar­iz­ing top­ics. So they usu­al­ly involved either cre­at­ing, or ampli­fy­ing, con­tent on social media that would encour­age peo­ple to take more extreme view­points. So some exam­ples might be that vet­er­ans were often tar­get­ed. There was this one meme that was run by Russian trolls, basi­cal­ly, that showed a pic­ture of a US sol­dier, and then it had the text Hillary Clinton has a 69% dis­ap­proval rate amongst all vet­er­ans” on it. Clearly intend­ed to have impact on how those peo­ple were thinking.

Ly: Right.

Green: They might also give mis­lead­ing infor­ma­tion about the elec­tions. Like they might tell peo­ple that the elec­tion date was maybe sev­er­al days after the actu­al elec­tion date, and there­fore try and ruin peo­ple’s chances of using their right to vote. Some dis­in­for­ma­tion cam­paigns told peo­ple that they could tweet or text to their vote in so they did­n’t have to leave their homes. And also there was exploita­tion of real polit­i­cal sen­ti­ment in the US, often encour­ag­ing divi­sion. And par­tic­u­lar­ly divi­sions around race. And so there were YouTube chan­nels that would be called things like Don’t Shoot” or Black to Live” that shared con­tent about police vio­lence and Black Lives Matter. And some racial­ized cam­paign that were linked to those types of sites would then pro­mote ideas like the black com­mu­ni­ty can’t rely on the gov­ern­ment; it’s not worth vot­ing anyway. 

So that’s the type of stuff that we start­ed to see in 2016, and many of those efforts were either linked to the GRU, which is a part of the gen­er­al staff of the armed forces of Russia. Or the Internet Research Agency, the IRA, of Russia. And many char­ac­ter­ize the IRA as a troll farm, so an orga­ni­za­tion that par­tic­u­lar­ly focus­es on spread­ing false infor­ma­tion online. 

So since 2016, unfor­tu­nate­ly online influ­ence cam­paigns have only become more ram­pant and more com­pli­cat­ed. We’ve seen a more diverse range of peo­ple being tar­get­ed in the United States, so not just vet­er­ans and African Americans but also dif­fer­ent polit­i­cal groups from the far right to the far left. We’ve seen immi­grant com­mu­ni­ties be tar­get­ed, reli­gious groups. People who care about spe­cif­ic issues like gun rights or the Confederate flag. So basi­cal­ly the most con­tro­ver­sial top­ics are the top­ics that for­eign actors tend to drill deep on to try and influ­ence Americans. It’s just got­ten more and more complex. 

Ly: I want to pick up on this point because so often par­tic­u­lar­ly racial issues form the basis of dis­in­for­ma­tion and influ­ence cam­paigns, because like you said they are the most divi­sive, con­tentious issues. I mean in what ways have you seen for­eign actors work to weaponize social issues in the United States just this year, maybe since the death of George Floyd?

Green: Well you know, it’s inter­est­ing because we focus a lot on dis­in­for­ma­tion as tar­get­ed towards the elec­tions, but a num­ber of dif­fer­ent types of behav­iors and activ­i­ties have been tar­get­ed through dis­in­for­ma­tion. So we’ve seen peo­ple try to manip­u­late things like cen­sus par­tic­i­pa­tion or cer­tain types of civic involve­ment. And the range of ways that actors are actu­al­ly using dif­fer­ent plat­forms is chang­ing too. So we’re see­ing text mes­sages and WhatsApp mes­sages being used to impact peo­ple in addi­tion to social media.

But after George Floyd was killed, as you might expect, because it’s a con­tro­ver­sial issue that affects Americans, absolute­ly there was sort of this onslaught of mis­in­for­ma­tion and dis­in­for­ma­tion that showed up online. So, there were claims that George Floyd did­n’t die. There were claims that were stok­ing con­spir­a­cy the­o­ries about protests that hap­pened after his death. 

And I have to say, not all dis- and mis­in­for­ma­tion is for­eign, so that’s why this is such a large prob­lem because there are many domes­tic actors that engage in dis­in­for­ma­tion cam­paigns as well. So, the nar­ra­tives that we’ve seen across the space come from so many dif­fer­ent peo­ple that some­times it can be hard to tar­get the the prob­lem to one par­tic­u­lar actor or one par­tic­u­lar motive. 

Ly: So in 2016, the Russian gov­ern­ment under­took real­ly sophis­ti­cat­ed meth­ods of influ­ence, cer­tain­ly for that par­tic­u­lar time and for that elec­tion, includ­ing mobi­liz­ing inau­then­tic nar­ra­tives via inau­then­tic users, lever­ag­ing wit­ting and unwit­ting Americans, and social media users. How would you con­trast the threat posed by Russia’s efforts with oth­er coun­tries known to be involved in ongo­ing influ­ence efforts?

Green: Well, I have to say that Russia con­tin­ues to be a coun­try of major con­cern. We saw just recent­ly this week the FBI announc­ing that Russia has been shown to have some infor­ma­tion about vot­er reg­is­tra­tion in the United States. Russian dis­in­for­ma­tion cam­paigns have def­i­nite­ly reemerged in the 2020 elec­tion cycle. But those cam­paigns only make up a small amount of the over­all activ­i­ties that Russia’s engag­ing in today, all with the goal of under­min­ing democ­ra­cy and erod­ing demo­c­ra­t­ic insti­tu­tions around the world. 

That being said, we’ve seen oth­er actors emerg­ing in this space. Within the first few months of the COVID-19 pan­dem­ic, Chinese agents were shown to be push­ing false nar­ra­tives with­in the US say­ing that President Trump was going to put the entire coun­try on lock­down. Iran has increas­ing­ly been involved in these types of cam­paigns as well. Recently they used mas­sive emails to affect US pub­lic opin­ion about the elections. 

And one more thing I want to men­tion is that this is real­ly a glob­al phe­nom­e­non. So you know, these actors, these state actors often out­source their activ­i­ty through sort of oper­a­tions in dif­fer­ent coun­tries. So for instance, there are sto­ries of a Russian troll farm that was set up in Ghana to push racial nar­ra­tives about the United States. And you know, there’ve also been troll farms that are set up by state actors in places like Nigeria, Albania, the Philippines. So what’s inter­est­ing here is that the indi­vid­u­als who’re actu­al­ly send­ing those mes­sages are either eco­nom­i­cal­ly motivated—they’re get­ting paid—or they might be ide­o­log­i­cal­ly moti­vat­ed. But they’re act­ing on behalf of these state actors. And that makes this not just a state-to-state issue but a real glob­al prob­lem that involves many peo­ple in dif­fer­ent parts of the world.

Ly: So turn­ing to the plat­forms for a sec­ond, what are your thoughts on some of the inter­ven­tions plat­forms have announced so far? Maybe like lim­it­ing retweets and shares via pri­vate mes­sage, label­ing posts and accounts asso­ci­at­ed with state-run media orga­ni­za­tions. You know, the list of inter­ven­tions sort of goes on.

Green: Yeah. All of the things that you men­tioned are a good start, I would say. At the end of the day I think it’s got­ta be a major focus on how can we inform social media users of the poten­tial threats in the infor­ma­tion envi­ron­ment, and how can we best equip them to real­ly under­stand what they’re con­sum­ing. So I think that part of the answer is for these tech com­pa­nies to of their own accord con­tin­ue to cre­ate poli­cies that will address this issue. But, we also need bet­ter leg­is­la­tion, and that leg­is­la­tion has to focus on pri­va­cy rights, has the focus on online adver­tis­ing, polit­i­cal adver­tis­ing, tech sec­tor reg­u­la­tion. And then we need poli­cies that will enforce this type of thing mov­ing for­ward. So it can’t all be upon the tech com­pa­nies with­out that guid­ance, because I don’t know that they nec­es­sar­i­ly have the total will to do all that’s nec­es­sary to real­ly get at this problem.

Social media com­pa­nies have already start­ed to label con­tent. They’re also search­ing for inau­then­tic behav­ior, espe­cial­ly coor­di­nat­ed inau­then­tic behav­ior online. But I think that there is par­tic­u­lar work to be done in terms of the way that we think about con­tent label­ing. So, when plat­forms are label­ing con­tent, they are usu­al­ly label­ing con­tent from some sort of state-run media. And if it’s state-run media, much of the state-run media that they’re look­ing at is not com­plete­ly a covert oper­a­tion; it’s not of a sit­u­a­tion where like this media source just does­n’t want any­one to know that it’s asso­ci­at­ed with the state. 

But, it might be pret­ty dif­fi­cult for the audi­ence to actu­al­ly deter­mine that that out­let is from a state-run site. So an exam­ple would be RT, for­mer­ly known as Russia Roday. There’s a rea­son I think that it went from Russia Today to RT. If you go to the RT web site, you will see a big ban­ner that says ques­tion more; RT” and then there’s lots of infor­ma­tion about how RT works all over the world in order to help peo­ple to uncov­er the truth. And then if you scroll all­l­ll the way to the bot­tom of the web site, you’ll see RT has the sup­port of Moscow or the Russian gov­ern­ment, some­thing to that effect.

Ly: Yeah.

Green: So, it’s dif­fi­cult for peo­ple to actu­al­ly know where this con­tent is com­ing from. And this sum­mer, Facebook made good on a pol­i­cy that they had said that they were going to enact for some time where they now label cer­tain types of con­tent. And basi­cal­ly they say that they’ll label any con­tent it seems like it’s whol­ly are ful­ly under edi­to­r­i­al con­trol that’s influ­enced by the state, by some state gov­ern­ment. And so lots of Chinese and Russian sites or out­lets are includ­ed in this pol­i­cy so far. And accord­ing to Facebook they’re going to increase the num­ber of out­lets that get this label. And basi­cal­ly what you see is like, on the post you see Chinese state-controlled media; Russian state-controlled media,” some­thing to that effect. 

That’s help­ful because now a per­son does­n’t have to click, and then go to the web site, and scroll to the bot­tom of the page to find out that this out­let comes from Russia. 

Ly: Surprise!

Green: But, at the same time, I still think we need to do more in terms of help­ing Americans to under­stand why it’s an iss—why state actors are try­ing to reach them, lit­tle old me who lives in some small city or some small town in the mid­dle of America. And how nar­ra­tives can be manip­u­lat­ed. And so only if that’s done, in con­nec­tion with label­ing more of these types of out­lets on social media do I think you get more impact. 

YouTube does some­thing else. In 2018 they start­ed to label their con­tent. But the way they were label their con­tent is they basi­cal­ly label any­thing that is government-sponsored. So, if some out­let is fund­ed in whole or in part by a gov­ern­ment, there’s a ban­ner that comes up at the bot­tom of the video that tells peo­ple that. And so you’ll see RT labeled as Russian con­tent, but you also see BBC labeled as British con­tent so it does­n’t have to do with the edi­to­r­i­al con­trol of the outlet.

One final thing on this, because I think this is real­ly impor­tant. So I have heard sto­ries of peo­ple who let’s say for what­ev­er rea­son have stum­bled upon some sort of con­tent from a for­eign actor.

Ly: Yeah.

Green: And so, this con­tent might come up because some­body shared some­thing and they watched the video, right. So they watch a video, let’s say they watch an RT video. Maybe they weren’t try­ing to find the RT video and maybe they also aren’t the type of per­son who would watch a lot of con­tent from RT. But they watched that one video. They con­tin­ue to scroll on their news feed. And then they get a sug­ges­tion. You might enjoy this.”

Now, the next thing that they get comes from Sputnik. It comes from RT again. So now they’re get­ting fed infor­ma­tion about the US polit­i­cal sys­tem that is being por­trayed by a for­eign actor, and they weren’t even look­ing for it. I think that that’s anoth­er thing that we’ve got to tack­le, is the algo­rithms that are used in order to uphold tech com­pa­nies’ busi­ness mod­els. Because in some cas­es, those algo­rithms will be harm­ful to peo­ple because they’ll actu­al­ly feed them infor­ma­tion from for­eign actors that might have mali­cious intent. 

Ly: Naima, this week the FBI con­firmed that Iran was respon­si­ble for an influ­ence effort giv­ing the appear­ance of elec­tion inter­fer­ence. And in this par­tic­u­lar episode, US vot­ers in Florida and I think a num­ber of oth­er states received threat­en­ing emails from a domain appear­ing to belong to a white suprema­cist group. Can you talk a lit­tle bit about what in par­tic­u­lar the FBI revealed and what its sig­nif­i­cance is for the election?

Green: Right. So, there was a press con­fer­ence on October 21st in which the FBI announced that they had uncov­ered an email cam­paign that was orches­trat­ed by Iran. The emails pur­port­ed them­selves to come from the Proud Boys, which as you men­tioned is a far-right group with ties to white suprema­cy. And it was also a group that had recent­ly been ref­er­enced in US pol­i­tics in the first pres­i­den­tial debate. But actu­al­ly now we know that these emails came from Iran. And some of the indi­vid­u­als who received the con­tents of the email post­ed them online. So they were addressed to the email users by name, and they said we are in pos­ses­sion of all of your infor­ma­tion; email, address, tele­phone, every­thing.” And then they said they knew that the indi­vid­u­als was reg­is­tered as a Democrat because they had gained access to the US vot­ing infra­struc­ture. And they said You will vote for Trump on elec­tion day or we will come after you.”

So first of all, they includ­ed a huge amount of intim­i­da­tion. Second of all, they were pur­port­ing them­selves to be this group that they were not. And third of all they absolute­ly were attempt­ing to con­tribute to dis­cord in the run-up to the elec­tion. It’s dan­ger­ous activ­i­ty. It is alarm­ing activ­i­ty. It’s some­thing that I think will have mul­ti­ple impacts for a time to come. Because even though the FBI was able to iden­ti­fy that this hap­pened, that goal of shak­ing vot­er con­fi­dence of course may have been a lit­tle bit suc­cess­ful in that instance. And so, one of the things that is good about this is that the FBI was able to iden­ti­fy this very quick­ly; to make an announce­ment to the US pub­lic that it had hap­pened; to be clear about what happened. 

Unfortunately, what they announced was not just that the GMail users were receiv­ing this email and there was false infor­ma­tion in it. They also said that they had infor­ma­tion that both Russia and Iran have actu­al­ly obtained reg­is­tra­tion infor­ma­tion from the United States. And that’s con­cern­ing as well. There appears to be good coor­di­na­tion between the pri­vate sec­tor and the gov­ern­ment on this issue. Google announced the num­ber of GMail users that are esti­mat­ed to have been tar­get­ed through the Iranian cam­paign. Unfortunately the num­ber is about 25,000 email users, which is no small amount. And so this is just anoth­er instance of how not social media but the Internet realm—email—can be used as a way to tar­get American pub­lic opinion. 

Ly: Thank you so much for join­ing, Naima. I real­ly enjoyed our con­ver­sa­tion and I know our view­ers will too.

Green: Excellent! Well, I real­ly enjoyed this so thanks for hav­ing me.

Ly: Thank you.

Further Reference

Medium post for this episode, with intro­duc­tion and edit­ed text