Lawfare Archive: Camille François on Covid-19 and the ABCs of Disinformation

Running isn't enough anymore. With PWC's managed services, your operations don't just run, they evolve continuously,

“powered by AI embedded directly into your workflows. So instead of maintaining yesterday's model,”

you're building tomorrow's advantage. PWC's managed services, we run your operations with tech and talent, so you can run faster, scale smarter, lead stronger. "House of the Dragon" "Dragon Camp" "Gang Dragon" "Tigerians" "Gang Tigerians" "In Triegen" "Farat" and "Eppische Schlachten" All that awaited you in the new Staffel!

Also, streamed the new Staffel "House of the Dragon" and all the series of "Game of Thrones" "Nur" on HBO Max! "I'm Sara Willrich" "In Triegen at Lawfare"

with an episode from the Lawfare Archive for June 27th, 2026.

On June 18th, as her final act is Director of National Intelligence. Tulsi Gabbard decostified a series of files related to the origins of COVID-19 and Dr. Anthony Fauci. Gabbard claims these documents prove that Fauci lied to Congress manipulated intelligence

“and funded research that contributed to the outbreak of COVID.”

Critics have pushed back against Gabbard, arguing the files do not prove any of her claims. For today's archive, I chose an episode from April 16th, 2020, in which Evelyn Duac and Quinto Jurassic spoke with Camille François of Graphica, a non-profit organization that works to identify and mitigate disinformation and misinformation online.

The trio discussed disinformation and misinformation during the coronavirus pandemic, including Iranian and Russian influence operations, François ABC framework for understanding disinformation and more. [Music] I'm Quinto Jurassic and this is the Lawfare podcast.

It's another episode of our Arbertos of Truth series on disinformation. This week, Evelyn Duac and I spoke with Camille François, the Chief Innovation Officer Graphica, where she works to identify and mitigate disinformation and misinformation online. We spoke to Camille yesterday April 15th.

Right after Graphica released a report on an Iranian influence operation, focused on COVID-19. Blaming the United States for supposedly creating the virus and praising China's response to the pandemic. Camille walked us through what Graphica found

and how this campaign compares to similar operations in the past. Like another recent campaign from Ghana, the Graphica helped uncover, which was linked to Russia and posted content aimed at black Americans. And we also discussed the ABC framework

that Camille has developed to understand disinformation campaigns. It's the Lawfare podcast. Camille François on COVID-19 in the ABCs of disinformation. Camille, so you're the Chief Innovation Officer at Graphica, and before that you were the Principal Researcher at Jigsaw,

which is a tech incubator at Google. You've also advised governments around the world on tech policy. So you've really seen all sides of this issue, and there's really no better person to talk us through what's going on right now. So thank you for joining us.

“Just to start off, what got you interested in this work originally?”

I think it's so much for having me on the podcast, and I'm really happy to be here. And he has to write sometimes I think that if there's a multi-stakeholder process somewhere, I've been in the shoes of every stakeholder around the table, so it's an interesting perspective.

I'm not sure how I ended up working on disinformation and information operations. I think it was kind of a natural evolution of the topic set.

I was working on, I've always worked on social technical problems

that emerged when you build into point acknowledging the world, and I was focused on emerging cyber security threats. And so quickly I ended up sort of in the deep rabbit hole of studying disinformation and information operations campaign. And faster than I could realize, it was my full-time job.

Yeah, it's certainly a growth industry at the moment. So plenty to keep us busy. So according to the website, you lead graphicors work to detect and mitigate disinformation, medium manipulation and harassment.

Yeah, about two years ago I think I left Google to create this team with this idea that we could create a team of people who could expose disinformation and study information operations and sort of information disorders with the platform but also more transparently with others, with investigative journalists and exposing the forensics that under,

you know, underpin the type of investigations that we do.

“It was quite a bet and I think so far so far we've managed to do what we wanted.”

And completely I had to build very interdisciplinary team of investigators, data scientists, analysts, scientists, researchers, and we work together on analyzing what happens in online movements and on exposing manipulative campaigns. Awesome. So in practice, what does that look like?

Like where does an investigation start when you walk into the office? What's the kind of first thing that you do?

Yeah, that's an interesting question and I've always had a part-time

answering this one because it really depends on the days and it depends on the investigation. So if I think of the one we published this morning, it looks at a thread actor who's been very well identified by loads of different researchers by the platform and that we're familiar with. And so, you know, the investigation of the published this morning, again,

looking at Iran, looks at an actor that we know well and gives an update on what is the latest with that actor. And that's a very different sort of workflow than starting from scratch with a lead and trying to figure out, is there a campaign, what's going on, how can we attribute this thread actor and what's the impact? So every day is a bit different depending on what we look at.

And it can't really say that there is like a clear, this is how the day goes. Every day of the week, which I'm sorry for. No, that makes a lot of sense. Yeah, absolutely. So talk to us a little bit about this report that you guys just released on this influence operation relation to the coronavirus. What did you find?

Yeah, so we've been looking at how the states, you know, different governments are responding to the coronavirus, geopolitical tensions online.

“And I think the first thing that's important to say is there are sort of two different types of response”

and we treat them very differently.

The first one is state sponsored messaging, right, official state sponsored channels,

official press organs that are accoladed with the governments and so you can think of Russia today, sputnik on the Russia side, press TV on the Iranian side. You can think of, you know, Chinese state sponsored media. And that is one part of what we've been looking at, which has been fascinating. The other part is when these governments use covert messaging, right,

when they do influence operations, when they run accounts that are not saying, hey, this account is manned or sponsored by the government. And the report we published this morning looks at an information operation originating from Iran that talks about the coronavirus. And here is specifically blames the US response and praises the Chinese response.

And so as we think about state sponsored disinformation in the age of coronavirus,

we always look at those both sides of this corner, right, the overt messaging

and then the covert campaigns.

“So when you say originated from Iran, do you in this report?”

I mean, the attribution issue is interesting. Could you walk us through what steps you go through for attribution and what kind of threshold you have before you feel comfortable attributing sort of covert or over campaigns to particular actors? Yeah, that's a great question actually. So I'm actually working on a paper that looks at how we're attributing information operations

and how much we can learn from the cybersecurity practice of attributing to directors and which part of it are a little bit different. In the specific case of the report we published this morning, we're very lucky because we're talking about an actor that's been very studied by the community. In 2018, fire I published a first report explaining that these pages were actually men in Iran

and were inauthentic and so pretending to be something they were not. And then after that, there were series of takedowns by Twitter, by Google, by Facebook and multiple investigations across the industry that sort of confirmed the body of evidence that here we were looking at a threat actor which was extraordinary, persistent and voluminous and indeed, based in Iran.

The set is coordinated and inauthentic and the set is based in Iran.

The next step is going to be, okay, which specific entity is behind this operation, which we haven't really gotten clarity on but I'm hoping that again, in a very collaborative manner, we can get to the bottom of this one.

“It's really important in reality to think about this different entities,”

and you can see that emerging clearly when people talk about Russian information operations, which tend to be put in one bucket. But in reality, what those different entities are doing are extraordinarily different from one another. And so, campaigns run by the Internet research agency troll farm look extraordinarily different than campaigns run by Russian military intelligence like the GRU.

And we still have actors on the Russia side that are large, competent and persistent,

that we're not really sure which specific entity is actually behind them.

So an example that comes to mind here is the secondary infection campaign. When we work on attribution, as I said, it's a quite iterative process, and we try to have it be a collaborative process when we can. And we focus on making sure that all our forensics are public, and that we can explain why we make the determination that we do.

Sometimes we don't have enough confidence to attribute a campaign. And in those places, we say that. So an example that comes to mind is last Thanksgiving in 2019, we saw a campaign a few days ahead of the UK general election that was circulating the trade leaks, the USUK diplomatic discussions.

And here, the campaign had the hallmarks and TTPs of an actor that we know as secondary infection. But we didn't have the sort of ultimate proof to say, yes, this is secondary infection. And so what we wrote was, we are seeing this campaign. We detailed the forensics of everything we're observing. And we wrote, there's only two options.

Either this is indeed the operator that we know as secondary infection, because it uses the same tactics again. And it uses the same hallmarks of that campaign. Or it is someone who's trying very hard to mimic the secondary infection operators. At that stage, we couldn't say more.

And when we published this, others in the industry picked up our report, picked up the forensics that we had shared, we're able to continue investigating, and eventually said, okay, we picked it up from here, and we actually are able to confirm that this actor is the Russian actress secondary infection.

So there's a lot there, and I'd love to dig into it more.

“But as a first matter, can you just explain for listeners, you mentioned TTP?”

What does that stand for? So it's interesting because in information operations, we borrow often from the world of cyber. And TTP here stands for tactics, techniques, and procedures. It's sort of like, how campaigns are run, right? So in the case of disinformation, it could be the types of behaviors

that we see those actors adopt in the course of running their campaigns. And so here you can see again, like, how this emerging field of people will work on detecting and enforcing against information operations, borrow from the practice of info set and sort of borrow to the vocabulary from info set. Sometimes very successfully, sometimes there's still a lot of like adaptation

that needs to be done. Absolutely. So I would love to dig into the secondary infection work that you mentioned more. Both just give our listeners some background on what you guys did with that, what the campaign was, and then how this campaign that you've discovered from Iran

“or on the coronavirus compares, like how are their tactics different?”

Yeah, it's interesting because when you're sort of like down on the rabbit pool of this work, all these campaigns are so, are so different to you, right? They kind of become like different friends. You're pretty cool. How different they are from one another.

Secondary infection is a campaign that was initially exposed in 2018. By Facebook, Facebook saw a handful of posts on their platforms to get down, shared it. This was exhumined.

And again, in this first examination, we're able to see okay,

these are the whole marks of how this threat actor functions, right? This is what it does when it does a campaign online. And by sharing those publicly and by sharing the forensics, it's enabled others to see when that operation was coming back,

We have a bit more coming on that front, so I can share more in the near future,

but that's an interesting operation because it's a very mysterious one, as in we've seen facets of what it does, including recently, when it talked about COVID-19 too, but we still don't really know who is the, as I said, the actual unit, right, the actual actor that's behind this secondary infection campaign.

Although here, we have a strong attribution from the Facebook threat intelligence team that confirmed that this operation was originating in Russia. The Iranian operation that we wrote about this morning is significant for a lot of different reasons. What I particularly like with this specific campaign is that it reminds us that many different governments are using information operation,

kind of as a new staple of how geopolitical affairs are unfolding in the 21st century.

We tend to think about, again, we tend to think about Russia a lot when we think about information operation, and we tend to think that all of this started in 2016, but if you look at Iranian information operations on social media, they started way back, right?

We have traces from back 2014 of the same operator doing fake profiles, doing strategic messaging, doing coordinated and authentic behavior. Similarly, here we have an actor who's been at it for many years, and who's been very persistent and created a lot of volume. This specific operation is also a good case study in what collaborated research

and enforcement looks like. Because as I said, many researchers and platforms dig into it, and they were multiple waves of take down on all these platforms. So we do see it coming back, and this is what we wrote about this morning.

“But honestly, we also see their operations being less and less effective,”

because of these waves of take down and exposures across the years and months. Yeah, so that's all super fascinating. One of the things I wanted to ask about is you mentioned it's part of geopolitics now, it's just kind of standard. This campaign that you reported about this morning,

can you talk a little bit about who you think the audience was? Because I've been one of the things that's super interesting is a lot of it is in English. So it doesn't seem to necessarily be directed at domestic audiences. Is that correct? Am I understanding that correctly? Yeah, yeah, and it's a great question too, because, you know, so as I said,

it is a new state, like information operations are a new staple of geopolitics. And again, I think when we talk about I/O people think about Russian people think about elections, but in reality, we now have a lot of different governments running these types of operations.

“And this totally happens outside of election cycles, right?”

It now focuses on any wedge issue, any sensitive geopolitical topic. Each operator is going to have its own audiences. Traditionally, this Iranian network has been very broad and very global in its audiences. And so if you look at this sort of media of sorts that they stood up, the IUVM, the International Union of Virtual Media,

you see it operating in multiple languages and in multiple formats, right? So they're doing videos, so you can contact them on WhatsApp. They have websites that have blogs. And so you can see here that their attempt is to reach a broad global audiences, very clearly. And that's, for instance, very different than some of the other campaigns,

which we see really double down on specific target audiences that they're trying to manipulate, for instance, in some of the Russian campaigns. Yeah, so talk to us a little bit more about the specifics of how this work. You said they're putting out videos. You also write about how they're putting out a lot of cartoons, news articles.

“Like, what particular products are they producing and how are they distributing them?”

Yeah, so that specific operators loves memes. Don't we all know, right?

And how they're distributing them is kind of an always evolving fight,

because it's a cat and mouse game of when are they going to get caught? And when are their new pages going to be shut down? So they have a few videos, so they're distributing that on online video channels. They have a few memes, so they're trying to sort of like crawl back on Instagram, but they get put it out of Instagram pretty regularly.

And so, you know, they just do their best to have social media channels that disseminate

And again, in this cat and mouse dynamic, where more often they do get shut down on social media.

And so their own websites is becoming sort of an important platform for them. That's interesting, too, because this dynamic of a cat and mouse game between an operator and researchers and platforms for tracking it is also something that we're quite familiar with

“and that we've seen other actors do in respond to, right?”

So I've spent a lot of time investigating the internet research agency across the years. And my sort of, it's bizarre thing to say, but I think my favorite year in the history of the internet research agency

is 2017 because then you start having a slow realization that this entity exists

and a clear realization of what it's actually doing on social media. And you have the platforms each starting to respond to these campaigns and to take them down. And then relationship to that in reaction to that, you have the internet research agency troll farm also responding to the platform starting to take down the pages. So sometimes, you know, when a Facebook page is coming down, they go on Twitter to complain about it. And then when the Twitter comes down, they go on Google ads to buy ads to redirect to their new website.

And you kind of sort of see this cat and mouse live game playing live as the operator

and the platforms and the researchers are responding to one another.

I love that you have a favorite year of IRA operations. I'm looking forward to releasing the greatest hits. So that's something that what you just mentioned, something that you guys spend so much time looking at. Just the right amount.

“Just to close on that IRA bit, it's interesting because I think when people think about the internet research agency troll farm,”

they tend to think of 2016. But in reality, here you have an actor that's been active since 2014. And the target is started to target American audiences in 2014. And that continued to be active up until traces that we now seen in 2020. And so if you think about the IRA, it's not a 2016 story.

It really is a story of how an actor has learned and evolved in targeting American audiences. And you know, general global audiences across the years learning new techniques and trying out new ways to achieve its objectives. So I'd love to talk about that then, the cat and mouse thing, because something that's so striking in the report that you've just released, is how few followers and sort of how little engagement these accounts got. And so, I mean, I assume that given that you're ready to declare a permanent victory in the cat and mouse game,

you've got it all sorted out and these campaigns will be permanently disabled from now on.

“Or what do you know of us about? What's the next evolution? How is this developing?”

Yeah, so in many regards, the report we published this morning is kind of an easy one, right? It talks about an actor that we know and we know how to track them and it's kind of like, hey, here's the latest. And as you said, it's a pretty good story of the community in general, being on top of a threat actor on which everyone has sort of aligned and agreed on. There are a lot of more complicated trends and threats that I see that I think definitely keeps us on our toes. One of the last rational operations we wrote about operation double the seat is named double the seat, because the operation is not only deceiving the people who are engaging with its content on social media.

It's also deceiving the people who are producing the content that's being put on social media. And here we have a person who's tied to the Russian Internet Research Agency who's actually paying people in Ghana who have no idea that they're actually part of this type of operation. And you know trends like this new new forms, new evolutions like this really really keeps the keep the challenges in the space and forces to be ahead of the next creative campaigns that we're going to see. And this was one in which we managed to get to the bottom of it because the CNN investigative team did extra ordinary work, but actually going on the ground in Ghana and sort of getting to the bottom of who are the people behind these posts.

Do they understand what they're doing? How do we uncover the fullness of this operation? And this took many institutions. This is another sort of great case study of how effective we are when we work together because the initial lead came from two professors at Clemson University who had been following the IRA for a very long time. The CNN investigative team did a lot of great work on the ground, both Facebook and Twitter and ourselves were also involved in investigating. So this sort of story of sometimes it does take multiple institutions working from different perspectives to to get to the bottom of of these new and increasingly creative information operations.

So instead of maintaining yesterday's model you're building tomorrow's advantage PWC's managed services we run your operations with tech and talent so you can run faster scale smarter lead stronger.

The world-wide Syrian phänomeness zurück comes from the island in drone guide weiter. Up sofort, there's the 3rd Staffel House of the Dragon by HBO Max. Here's the whole world from Westeros on one board, Game of Thrones on Night of the Seven Kingdoms and of course House of the Dragon. Drachen camp from against Drachen, Targaryens, Gang Targaryens in Triegen, Farad and E. Pische Schlachten. All that was awaited by the new Staffel. Also streamed yet the new Staffel House of the Dragon and all the Syrian from Game of Thrones, Nur on HBO Max.

“Online with Sign, with PPP is the most simple. When you're on the Apple Pay button tips you can't miss any of the details directly.”

You want something else? Just simply tap on your Instagram to update your business. Then you can also use your business to update your business and in the second to read.

With PPP, you'll get online shopping, and even a few. Okay, so we definitely want to ask you about the Operation Double to Seat, but before we do that, I want to pick up on one of my pet curiosities, which is this collaboration and this community that you're talking about of researchers that includes journalists, platforms, maybe sometimes the government and things like that.

“I'm wondering if you could sort of just walk us through what is the nature of the collaborations and how you work together, what's your relationship with platforms and what kind of information sharing there is?”

Yeah, that's a great question. And again, it really depends on the specific cases.

I can tell you what we're trying to do in this space, and I can't speak for everybody else, right? So we want to encourage a space that's based on collaboration and transparency. And so when we write a report when we expose a campaign, we go out of our way to include as much detail as possible. All the forensics we can because we also write for the rest of the community thinking that the open source investigation movement may pick up where we left off. Some of our investigator colleagues either in media or the platforms may be able to answer some of the questions we were not able to answer.

And so we sort of try to raise the bar in rigor and transparency when we expose something because we think that others are going to be able to pick that up. So you'll never see us, for instance, say, hey, we found a super secret, you know, Russian campaign and just trust us.

“It's there and it's really bad, right? So it's also a question of, like, how do you create this shared standards of rigor and transparency that enable these collaborations to work? Now, you know, it's a very young space at the end of the day, right?”

You think back to 2016 where we didn't really have clear definitions across the industry of information operations on social media, we didn't really have teams that were focused on this. There wasn't really specific terms of services that apply to these types of campaigns. In the end of the day, you still have a community that had to come together pretty quickly, but it rapidly, and so a lot of this is still structuring itself. It of course stands on large body of historical work and we benefit from a lot of different subfields, but that specific community that's focused on

the detecting and forcing against information operations of social media, at the end of the day, it's still pretty young, and so I think a lot of how these collaborations work and don't work and sort of put themselves into place is still kind of, yeah, it's still forming. Before we ask you more about Operation Double Decide, I wanted to follow up with just one last question on this Iranian operation, which is, you mentioned that this information information operations have kind of become just one more tool in the toolkit for state actors, which suggests that, you know, an operation around the coronavirus is kind of the same as an operation around anything else.

But on the other hand, there are some researchers who have also suggested that health disinformation is different, in some way, right, that there's a perhaps it's easier for platforms to aggressively moderate.

Yeah, that's a fantastic question because it involves a lot of terms that nobody agree on. So, political disinformation, health misinformation, information operations, that sort of like, I think we're going to talk about it a bit, but that's sort of the point of writing more, more structured frameworks to categorize and sort of tea of part of those different sub-categories of what can be a pretty tangled space around disinformation.

What I mean when I say information operations are now one more tool in the toolbox of how governments use online means in the pursuit of their geopolitical goals.

“I think from a government perspective, you see them using hacking, fishing, public diplomacy. Now, we increasingly see them using information operations. It's more from a government using digital means to pursue their geopolitical strategy.”

That's the framework that I mean when I say, that's just a new tool into toolbox. From the platforms perspective, I think there's a lot that's very different between information operations and health misinformation. And again, we get into the depths of definitions in the space. One of the earlier definitions that I think most researchers agree on, with the difference between disinformation and misinformation, with disinformation being wrong information propagated with the intent to deceive and misinformation being wrong information propagated by people who don't have an intent to deceive, right?

“So if I share a bit of wrong information on coronavirus, I'm not particularly trying to trick you perhaps as simply believe this information and want to share because I care about you and I think you might use it, right?”

What's interesting is when you have, as we wrote in the report this morning, information operations that create disinformation around the origins of the virus, very quickly, you will have people sharing it, not because they're intending to deceive others or not because they're trying to pursue Iranian geopolitical interests, but just because they're consuming COVID-19 use in a very networking environment and want to share and discuss this with others, right?

So from a platform's perspective, it's very different, from a research perspective, it's very different because you can have one type of information more thing into something else.

Yeah, you know, I saw a recent survey information that suggested that a significant percent of Americans think that COVID was engineered as a bio weapon. So that's just a demonstration that I'm sure most of those Americans wouldn't argue that they're parroting Iranian propaganda, but it clearly is an idea that's out there and that people are picking up on and using.

“And I think that there's something that's super interesting in the dynamic that you just described, which is we see this constant back and forth between six sponsored information operations and sort of homegrown conspiracy theories, right?”

When we see a foreign actor trying to play up a wedge or trying to divide societies more often than not, they will take a route on an existing division or an existing conspiracy or on an existing sort of fracture in our own information environment. Those things are not sort of independent from one another, you know, the shape and nature of our information environment with the conspiratorial crowds with the very divisive language with the polarization is the landscape on which you see those, you know, foreign interference and foreign disinformation campaigns take ground.

So let's talk about another operation that did exactly that and played on existing fractures in society, we have been giving teases for it for the last 10 minutes, tell us about operation double the state, which is this totally bananas operation that we're part of uncovering what did you find there. Yeah, that that was pretty bananas as as you say, in operation double the seed, an individual affiliated with the Russian internet research agency has paid a series of other individuals in Ghana to run pages that are targeting American audiences.

And here it's interesting because in many ways you see how similar to previous IRA operations the content was and in many ways, of course, it's also a departure from what we've known and and shows sort of new evolutions.

And here, I think that we may want to pause and explain a little bit what those pages were doing and what it looks like and how is it even possible to be part of an information operations without knowing about it.

“And the trick is in reality, a lot of these operations are not doing crazy content all the time they're just trying to build audiences and a lot of the content they produce is content that's designed to engage the audience that they're targeting.”

So when you look at the post that that these guys were putting out, it's a lot of for instance, post focus on black empowerment, black beauty and sort of African American issues in the US, but they're not particularly divisive and they're they're empowering and uplifting memes and so it's actually, you know, you can if you keep that in mind in terms of a lot of the operation was producing this type of content. It kind of helps understand how it is possible for someone to believe that they're just doing, you know, a normal NGO campaign that just celebrates black history months and not realize that they're part of an operation.

That specific phase of a campaign is what we call the audience building phase. We've seen this in many information operation. The first sort of set of posts in the first month is as often just putting out this uplifting and engaging kind of feel good content. Because you want to amass followers and sort of create an audience around those accounts and then once you have an audience that's engaged and sharing with you, you can sort of start, you know, coding quote weaponizing I don't really like this term, this is one that comes to mind right now, you can sort of start weaponizing these accounts and sort of use this accounts for more divisive and more political content.

“And so do we have any sense of what this campaign might have been about like what it might have been gearing up to do.”

Yeah, it's it's always kind of tricky to say, oh, here's what would have happened with this.

But here, if we take the IRA's previous campaigns who are focused on the same audiences as a template, then you start having posts that are more divisive more political that perhaps go to voter suppression, although I want to be really clear, we hadn't seen that yet, right, so it's sort of also very important that space to be clear in describing what you're seeing versus sort of where you think this could go. We can make the assumption that over time, those posts would have been increasingly divisive and would have been used to increase division in American society, which as as you know, the long running goal of information operations emanating from Russia.

There's a lot of ground here, we've talked about like different kinds of actors, state actors versus covert operations, we've talked about political disinformation versus health disinformation.

We've talked about some operations that don't even spread particularly problematic or, you know, divisive content initially, I guess something that we're sort of dancing around is, how do you define what is a problematic campaign, you know, there's a lot of inauthentic behavior online, that's kind of what the internet is.

“There's a lot of memes online, and so one of the things that you've been working on and you've been, you know, key in developing is this ABC framework to analyze online disinformation campaigns, could you maybe talk us through what that is?”

Yes, of course, that ABC framework looks at what makes different things disinformation, and I crafted it with the goal of informing concrete responses and enforcement and areas of law that we could come to bear to tackle this information. And so, you know, as I said, a lot gets tangled up in this disinformation bucket, right, a lot of people put together like bots and Russian campaigns and my uncle says crazy stuff on Facebook and, you know, in reality, a lot of these different problems don't belong together and get often referred to as disinformation as a big mix.

It sends for actors, behaviors, and content. I'm going to start by content because often when people think about this information, they have this idea that it's the content of a message that makes it disinformation.

But when you look at, for instance, terms of services, you realize that content is not the bulk of this information, and there are very few policies that focus exclusively on content.

In the age of COVID-19, it's actually quite different, it's evolving rapidly. We see a lot of terms of services focusing on the content of messages saying, if you have these following content, then we will consider the message to be disinformation and then we will take action.

“So that's for the content vector, which is kind of the most straightforward. Be is for behavior. So you can have a campaign in which the content is totally fine, right?”

I could be again a plifting empowering content, but the way the content is being disseminated on the platform is what makes it disinformation. So bots is a good example of that, right? If I have a large bot network that is designed to make people believe that my campaign is much more popular and much more viral than what it really is, that I'm deceiving.

I'm deceiving the audience is not because of content, but because of the behavior by which the campaign is being spread.

And a finally is for actors, sometimes you have one person manning one account, right? So it's not a deceiving behavior, and it's not acting in the large network, and you have content that's totally fine, but really the problem is who's the actor behind it, and some of the information operation sort of are that, right? If I have an account like Jenna Abrams, which was one of my favorite trolls from the original IRA campaigns, it can be one person is manning one account, and it can be the Jenna puts out a lot of content that's not on its face problematic.

But here, the problem is, behind the account, you have a Russian operation that's designed to deceive American audiences, and so here, that's a good example of a campaign that that is disinformation because of the manipulating actor, manipulative actor behind it. The idea in this framework is also to say, when we want to address the full of disinformation, perhaps in government policy, for instance, there are different disciplines that are coming to help and sort of weigh on these different aspects.

For actor-based campaign, a lot of this is also a cybersecurity issue, and warrants the type of collaboration and information sharing that you see in the cybersecurity space.

And this is where you really need to think through a content moderation issues and issues that have to do with freedom of speech and the type of issue that we're more familiar on the content side. So this is sort of like where the ABC frameworks comes from and what it's trying to do. To clarify, so it is basically the idea that if you break it out into these three vectors, you both get a clearer understanding of how disinformation and misinformation can be disseminated, and the people who are trying to address understand counter that material, have a clearer sense of, you know, how they should be thinking about each specific issue is that right.

“And I think it's also meant to show that there is not going to be a one-size-fits-all easy response to the whole of the disinformation problem.”

That paper was really written up to sort of highlight that if we want to tackle this information, we will need to bring to bear different types of instruments, different types of disciplines and legal framework, because at the end of the day, we have different types of problem entangled into one another. So I think you're right that content sort of an actor even seems quite simple and intuitive in many ways. I'd love to sort of dig down on the behavior vector a little bit, because it seems to be one that really gets a lot of focus.

“You know, these platforms often when they do an announcement saying they've done a take down, they say, and we're not taking this down, not because of the content that they pose, but because of the behavior, right?”

So that it avoids the dreaded arbiter of truth problem, like this is not about the politics, this is about some sort of deceptive behavior.

Like do we have a really good set of criteria for evaluating that yet?

“A lot of stuff online is authentic, as I sort of said before, and I think, you know, we saw the Bloomberg campaign in particular trying to push the limits and boundaries of this instead of testing where platforms are going to draw the lines.”

Is that something that sort of establish or is it still in evolution? I think it bound to be constant evolution, and the other reason why this is a great question is, you will see platforms define problematic behavior on their platforms differently from one another. Part of this is because manipulated behaviors will look different from one platform to the next, and other parties because they have different types of terms of services and philosophies that they're applying in their moderation practices.

So Facebook, for instance, coined this term coordinated and inauthentic behavior to get to the idea that a series of pages, for instance, who are pretending to not be all related to one another, but who are acting in concert to deceit users is a behavior problem, right?

So you will have different platforms talking about different ways in which their own platforms can be abused to do deceptive behavior and sort of coining their own terms, and you're right.

“We're going to see again a little bit of this cat and mouse dimension with new campaigns, trying out new behaviors to try to see, okay, what, what, you know, what are the limits of that?”

So we're almost out of time, but before we close, I wanted to ask you about the sort of elephant in the room, which is the 2020 election in the United States. What is your biggest concern? What are you watching for around that? Yeah, it's a tough question for me because as I work on these issues, I do think about this a lot. There are a lot of things that bring me comfort in seeing how much progress we've, we've made since 2016. We're in a very different position in 2020 than in 2016. As I said, we have a clearer understanding of what interference looks like. We have a clearer understanding of both the foreign and the domestic groups.

We have clearer policies. We have clearer teams who are in charge of enforcing this. We have a great set of reporters who are uncovering this more researchers paying attention to this face. So, you know, in general, I, I measure how long, how far we've gone since 2016.

Now, I do worry about a few things that I don't things we fully explored yet from previous incidents. And here, you know, several things come to mind.

I've worked with people who were sort of personally targeted by information operations and who shared the type of direct messaging that they had going back and forth with trolls and sort of being sort of an unwittingly enrolled in campaigns like this. We really talk about how much of these information operations when they're sophisticated target individuals directly and use direct messages. So that's a vector that's top of mind. There's something that we haven't studied enough yet and discussed it transparently enough yet. So I wonder, are activists who are likely to be targeted journalists who are likely targeted equipped enough with the type of evidence we need them to have seen to to see it again if it comes again in 2020.

I'm curious about the manipulated media and synthetic media, right? We talk a lot about deep fakes, but in my work, I've seen a few campaigns using a gang generated profile pictures.

“And so I think synthetic images is a significant threat. Similarly, since 2016, we've had much, you know, much, much stronger synthetic text capabilities.”

But with GPT2, when I can see that the, you know, the threat of synthetic text, which, which can be very powerful and large this information campaign is something that we're going to have to deal with this time around. This is the part that it will read fakes because I have a terrible sense of humor. And so I think, you know, there are a lot of different vectors and a lot of different technologies that make this problem, a continuous evolution of things that we're going to have to catch up with that,

yeah, keep on keeping, I don't like keep me up at 9, because I, you know, I'm trying to sleep at 9, but yeah, it, it makes for for a few things. I don't know if to be worried about is the right framing, but for sure, a few things to be focused on as we head towards 2020.

I guess just one final question then is if you could ask for one thing, whether it be from platforms or from governments or some sort of reform in this space that would make your job easier or more impactful or sort of help you achieve what you're trying to achieve.

“What's the one big thing? What's the one big frustration that you have or or the one big ask that you would have from anyone in this space?”

It's a great and top question. I think I have a lot of smaller ask. We still have a long ways to go and being more transparent in that space.

Again, I think we're making fast progress, but I think the faster we can get to robust transparency principles as we uncover and enforce against these operations.

“The faster we'll make progress and actually tackling the issue in a way that ensures that we're not creating more problems by by solving those.”

So it's a bit of a unsurprising answer or specifically, you know, yeah, it's an unsurprising answer, but I think like we need to continue working towards transparency and accountability in that space, and I think that we'll make a whole world of the difference.

All right, Camille, thank you so much for joining us. Yeah, thank you so much for having me. This was a really fun conversation.

You've been listening to Arborters of Truth. The Lover podcast is many series on disinformation.

You can find past episodes in the Lover podcast feed and we'll be back for another episode next Thursday.

The Lover podcast is producing cooperation with a Brookings Institution. Our music is performed by Sophia Giann, our audio engineer this week was Ian N. Wright, and our producer is Jen Pachie Howell. Please, rate and review the Lover podcast and whatever app you use, and thanks for listening.

“Do your current managed services really help run your operations or are they just running in circles?”

Running isn't enough anymore. With PWC's managed services, your operations don't just run, they evolve continuously, powered by AI embedded directly into your workflows. So instead of maintaining yesterday's model, you're building tomorrow's advantage. PWC's managed services. We've run your operations with tech and talent, so you can run faster, scale smarter, lead stronger.