Transcripts For CSPAN3 Discussion On Securing Elections Combating Social Engineering 20240713

Our first speaker that we have is madeleine, are you over there . Im going to let you stand up here at the mic if thats good four. Shes the principal director for cyber policy for the office of the secretary of defense. Without further ado lets get the party started. [ applause ] so ill apologize, we actually had thought our colleague from the fbi would go first, but im happy is that okay . If thats all right. Okay. Its not the program, but so it is my distinct honor and pleasure to introduce david kay porter who is the assistant section chief with the Foreign Influence Task force, counterintelligence division, joining us from fbis headquarters. Thank you. [ applause ] sorry about that. So good morning, everybody. My name is david porter. As anne said, i serve as the assistant section chief of the fbis Foreign Influence Task force. Or fitf as we are more broadly referred to within the Intelligence Community. I appreciate the opportunity to be here this morning to represent the men and women of the fbi. Men and women who are working tirelessly on behalf of the American People to protect the integrity of our democracy to include the 2020 elections. Ten minutes is not a lot of time. Im going to get right to it. What i want to do this morning with my time with you is to speak to three different things. First, i want to define what we in this space refer to as the malign foreign influence threat. Secondly, i need to articulate some of those objectives and ttps or tactics, techniques, and procedures, pertaining to the threat. Lastly ill describe a little bit about what we do in this space and how we mitigate this threat. First, to defining the malign foreign influence threat, i think whats important to do here is draw a distinction between normal foreign influence activity and malign foreign influence activity. So the former would be, you know, normal diplomatic activity carried out by every country, usually conducted through diplomatic channels. The latter, though, malign foreign influence activity, its operations are actions by a foreign power to influence u. S. Policy, distort Political Sentiment and public discourse, undermine confidence in democratic processes and values, and this is important for us because this is the focus of our investigative efforts at the bureau and with ftif, it is the subversive, undeclared, criminal, or coercive nature of these activities that serve as the basis to have our investigative interest. Our adversaries frequently use a whole of government approach here. The vectors, including the official, outwardfacing component of a foreign government, intelligence services, cyber actors, staterun media, businesses close to government officials, and social media actors. There is a Broad Spectrum of foreign influence activity. When a country moves from normal and official diplomatic engagement to conduct in this subversive, undeclared, criminal or coercive conduct, that is when we see the malign foreign influence activity. It can show itself through economic coercion, bribery, honey pots, covert media reports, social media exploitation, blackmail, to name a few. So to move on to the second point, objectives and ttps. There are two main objectives for malign foreign influence actors. The first is to push foreign goals and policies at the expense of the United States. Some countries, lets say china, use influence activities driven by priorities associated with their National Development or stability. For example, chinas primary objective is to strengthen and perpetuate the rule of the Chinese Communist party. Beijing also seeks to ensure sustainable economic development, protect against perceived threats to its states sovereignty, territorial integrity, and maintain the stability of its political system. Second objective, and one that well talk a little bit more about, is create division and muddy public discourse. We refer to something called information confrontation in this space. These operations seek to weaken an adversary from within. Countries like russia use information confrontation to target the perceptions of their adversarys population. These methods erode confidence in Democratic Values and institutions, to encourage negative sentiment, apathy and mistrust of government. Election interference is one of the vectors in this space. Its designed to degrade confidence in the very foundation of our democratic system, in our leaders ability to govern. Its also designed to weaken the adversary from within by identifying existing dissonant political and social issues and driving wedges into those fracture lines, and amplify them through online disinformation in an effort to create an environment of permanent cacophony, unrest, and conflict. Its also designed to undermine the publics confidence in the credibility of an established free and Independent News Media to create an environment of public mistrust in the narrative reported by traditional news organizations. This environment is then exploited to push consumers toward alternative news sources on social media where of course it is much easier to introduce false narratives. Its also designed to sow doubt and confusion about true narratives, by exploiting the media landscape to introduce conflicting storylines, undermine credible sources of information and saturate the Information Space with unreliable narratives. To be clear, the goal here is to exhaust our ability to think critically and to separate truth from falsehoods. The primary objective is not to create a particular version of the truth but rather to cloud the truth and erode our ability to find it, creating a sentiment that no narrative or news source can be trusted at all. Some of the ttps in this space wont necessarily come as a surprise to you, but were talked about intrusion into u. S. Government networks and political organizations, hack and leak operations relating to compromising or embarrassing information, Cyber Attacks against our voting infrastructure, the targeting of u. S. Persons or elected officials by social media and disinformation, suppression of voter turnout by spreading misinformation regarding polling and voting, manipulation of media through the injection of false stories and news reporting, and then subsequent amplification of disinformation to shape public discourse. Sorry about that. There are a few overarching similarities regarding the manner in which our two principal adversaries, china and russia, execute these two objectives. Both countries use a whole of government approach. And they use sophisticated and aggressive efforts to advance their national priorities. However there are clear differences in this space as well. China and russia vary in the extent of their aggressiveness and risk tolerance. We see russia is willing to conduct more brazen and disruptive influence operations because of how it perceives its conflict with the west. In some ways, however, china contains its malign foreign influence operations to its strategic goals of developing a modern National Economy and building its geopolitical prestige to be respected as an equal if not superior rival in its eyes to that of the United States. To put it simply, in this space russia wants to watch us tear ourselves apart while it seems china, on the other hand, would rather manage our gradual economic decline over the course of generations. So what do we do in this space . The director established the ftfa, Foreign Influence Task force, in october of 2017 to bring together the fbis National Security and traditional criminal investigative expertise under one umbrella, to identify and counteract malign foreign influence operations. At the time we were primarily focused on the russian malign foreign influence operations. However, the scope has expanded since then to address other global adversaries. China, iran, north korea. Our structure bears some resemblance to the fbis joint terrific task force which exists at fbi headquarters in very large fashion but also in all 56 of our field offices. However, while we do work side by side with detailees from various agencies in the fitf, our structure is different. Fitf combines expertise from across the enterprise, including analysts and other professional staff from the fbis counterintelligence, cyber, count terrorism, and criminal divisions. Our multidivision task force has the authority and mandate to bridge all fbi programs and equities to combat this threat. In addition to investigative operations and broad intelligence sharing across the usc and with our state and local partners, the fitf also works hard to build strong partnerships with the private sectors and academia. The fitf Management Team meets regularly with social media and tech companies. When appropriate, the fbi provides actionable intelligence to social Media Companies to help them fight abuse of their platforms by foreign actors. And work in this area, one thing that is very important to us, particularly on the cyber side of things, is that attribution is key. We do not run around chasing content, in addition to having a host of First Amendment issues, that approach would be inefficient and ineffective. We dont focus on what the actors say. We spend our time concentrating on who they are. Attribution is key. When were able to identify and track foreign actors as they establish and use their infrastructure and mature their online presence, the fitf works with social Media Companies to illuminate and disrupt our adversaries activities, including at times through actions taken entirely by the companies themselves to voluntarily remove accounts that violate their terms of service agreements. Ive run out of time, so ill just say that our adversaries are actively engaged to influence Public Opinion and our electoral processes. It is our responsibility to take the threat seriously, evolve our methods of disruption, and maintain our fierce determination and focus. Thank you for giving me the time to speak to you today. [ applause ] ive already introduced you. Yes, thank you so much. So good morning. My name is madeleine wartelmans. Were responsible for developing guidance and providing advice to the secretary of defense about what the department of defense must be capable of. Im here to talk about dods role in defending elections and in support of and complementary to our federal government partners and the fbi in countering foreign influence and the department of Homeland Security in supporting elections infrastructure security, as well as state and local partners. From my viewpoint, in dod speak, we frequently talk about supported and supporting. In the context of defending elections the department of defense is principally in a supporting role. That was why it was important for you to hear from my colleague from the fbi first. So in that, i want to just jump right in and try to get to the gist. The National Defense strategy concludes that for the past decade, to offset our military superiority, our adversaries are increasingly using actions below the use of force to undermine our National Security and national interests. Nowhere is this more true than in and through cyberspace where we see adversaries using coordinated, long term campaigns of malicious cyber activity to harm the United States, our allies and partners, and to undermine international order. Their objective is within is to win without war. And in the event of conflict, to leverage their accesses and capabilities priority hostilities in order to achieve strategic advantageous. China, russia, iran, and north korea are using and will continue to use cyber to steal information and to prepare to disrupt Critical National infrastructure. But duties interest is not limited to military operations. Our adversary is also seeking to influence our citizens and to undermine Democratic Institutions in order to achieve that strategic advantage that will allow them to win for their national interests. The Intelligence Community assessed that they are capable of and may seek to interfere in our voting process, the infrastructure that we use, or to covertly influence our citizens in order to achieve an outcome. The department of defense has now determined at the president s direction that helping to defend elections is an enduring mission. But we are part of a broader whole of government effort, an unprecedented level of coordination. And in that way the department of defense is playing a complementary and supporting role to our domestic partners. In this way, we are looking to leverage our specific comparative advantage and authorities to defend forward. Just as the department of defense projects power in the physical domains, land, sea, and air, where we seek to understand our adversaries activities, to shape the environment, and to address threats before they reach the homeland, were seeking to do the same things in and through cyberspace in support of our elections. Building on our activities in 2018 in defense of the 2018 midterm elections, the department of defense is conducting complementary activities in support of civilianled efforts defending forward in order to generate insights, first generate insights about adversary intentions and activities. In this way, were collecting and analyzing data about foreign threats. It includes things like identifying malware and networks, covert influence identities, and to better understand whats happening outside of the United States. We do this with partners and throughout the global environment. Second, were enabling our domestic partners to better defend elections. This includes sharing those insights that we generated from outside of the United States with our domestic partners to enable their Network Defense activities. An example would be in 2018, we conducted what we call hunt forward operations, but what is in fact partnered Network Defense operations with our traditional allies and partners to gain insight about how our variations are using their malicious activities in other countries that can in turn be used to better defend our own networks. And finally, thirdly, when appropriate and authorized, the department of defense will conduct military operations to degrade, disrupt, or defeat foreign interference or covert influence. This can take any number of forms. It includes operations that would seek to put the sand in the gears of our adversaries attempting to accomplish their outcomes. We can do this for instance through exposing their malware or their Network Threats indicators. To date, u. S. Cyber command has publicly exposed eight different examples of how adversaries are seeking to conduct malicious cyber activity against the United States, which enhances our overall defense. We can also take other operations that seek to slow them down. The department of defenses efforts to defend forward and to defend elections are spearheaded by u. S. Cyber command and the National Security agency. This enables us to combine our unique comparative advantage, expertise, capabilities, and capacity. Their efforts are also complemented across the dod with u. S. Northeastern command, u. S. India pacific command, National Guard bureau, all prepared and ready to support our dhs and fbi and appropriate state and local authorities. Thank you. [ applause ] thank you, madeleine, and thank you, david. So in listening to what we just heard, it was really i want to highlight a couple of things i heard david say, this idea of mistrust to the government, and these are threat actors, malicious actors doing this, pushing consumers toward alternative news sources, sowing doubt and confusion about true narratives, destroying our ability to think critically, clouding the truth, well, how does that happen . Im going to step back for a minute and step away from elections and talk for a minute just about data and the world we live in, right . We are surrounded by censors everywhere. Everything we do, were all carrying these around, theyre embedded with sensors, we have fitbits. Every environment in which we exist has sevnsors. Hey, alexa, what are the news results . We have health and fitness sens sensors, automobile sensors, theyre all connected to satellites. Were physically tracked. So much data is being aggregated about us. I always like to give an example, it gives a good perspective for the audience. This is from an actual Patent Application from walmart for biometric data via sensors in a walmart shopping cart handle. These sensors would give heart rate, temperature, force against the handle, cart speed and location in your walmart shopping cart. Shoutout and thanks to r. A. Wisenberg for alerting me to this. We have sensors that are recording us, everybody has their fitbits on, their apple phones. What happens is how these sensors are interacting with all of these apps that we download and are using on our phones, right, we have numerous companies, private companies, buying, selling, aggregating this data, and its intimate data. Im not just talking about who is in this room right now, which is being aggregated, but really, data that can be used to manipulate our choices. If we think about that idea, clouding mistrust, part of that is foreign actors, but part of that is ourselves. And part of that is, all of this content we create allows us to be so easily manipulated. So just a screenshot here of facebook, advertising on facebook makes it easy to find the right people, capture their attention, and get results. You can do that on any budget. How does that happen . How is it effective . Its effective because its use and sale of your data for microtargeting ads. So we understand microtargeted advertising. Microtargeting enables an ad tailored just for you, only one person in the room would receive that version of an ad. This is right off of facebook. Facebook isnt alone here, they just happened to always be such a prominent user of your data. So audience insights, knowing your audience like never before. Aggregate information, right . So facebooks collecting information from all of our apps. And so an example is some women or persons use apps to track their cycles. And some apps track womens cycles without letting them know. So what does that matter . How can that be used politically, and why do i have a picture up when were talking about periods . That information is shared with facebook. And so what does that mean . Where does that take us . Well, lets think about something weve heard about in the news or at least women have heard about and are aware of, taxes on feminine hygiene. So lets think about microtargeting for a minute. If you are in a particular period or time in your cycle, you might feel more strongly about seeing which states tax your period. You might be much more susceptible to content like this. And this content can come not i mean, its the ability to manipulate you based upon very intimate details about what youre doing, what youre experiencing, where you are. And so again, were not just talking your location, who your contacts are. Were talking very intimate, intimate information. So we see politicians take advantage of this, right . We have one representative tweeting out about how he wanted to be charged by the house administrators because he wanted to provide feminine hygiene items for his staffers and people visiting his office. Thats taking political, you know, use of this in an advertisement. And yes, thats absurd, thats not considered a necessity for women. But it raises this question, when youre getting messaging, what is real . So we heard from david porter about, you know, the sowing of mistrust. And part of that is this question of how do we determine what is real. We have so much information coming at us on social media, our news feed, its even worse on your mobile phone. Research shows youll scroll through your mobile feed much more quickly. In this question of what is real, the technology continues to advance, and that takes us to deepfakes. We have an indian politician who is using deepfakes to win new voters. How did he do it . He didnt speak a particular language for part of the population. He just used ai, lip syncing technology, and had his, you know had a deepfake of himself speaking in a language he didnt speak to atune to an audience. Weve seen these deepfakes. Weve heard about the pelosi, weve heard about the obama. Theres lots of this. Its not just something coming in the future. Its here, ready or not. How is the law handling this . We heard a few people mention First Amendment. So the question is, is this protected speech, right . Are deepfakes, are disinformation or misinformation, are they protected speech . Is this protected content . Is this something were going to say our First Amendment, no law shall abridge the freedom of speech, no law shall abridge . Fortunately, lawyers, a little plug for the profession there, lawyers recognized that the First Amendment isnt in fact omnipotent. Its not that all speech must be protected. So whats interesting is, in the preinternet era, this was so much more clear. We knew that publishers were liable for content and individuals were liable for content that was defamation. If you Say Something about someone thats not true and it causes actual harm to their reputation, they can sue you. And theres court decisions, its hard to do this in a flash talk, but hate speech is not protected. Burning a cross on someones yard is not protected by the First Amendment. Incitements to violence are not protected. Putting up pictures of physicians who worker at an Abortion Clinic with crosshairs, thats not protected speech. And so invasions of privacy were also recognized, we could protect our privacy when its violated if we had a reasonable expectation of privacy in content. Remember, this is important. Publishers were liable before the internet came along. Then something happened. And all of these platforms that are Interactive Computer Service providers have a shield now. And so if we think about that source, your information source, and weve heard a lot about, Pay Attention to credible sourcing, so if you saw something in the new york times, you knew, im reading this in the new york times, or the wall street journal, as a publisher, they would be liable for things they repeated. So if there was defamation or hate speech or something that wasnt protected, the law had an enforcement mechanism to get rid of it. And then along came something that is extolled as the real, you know, strength of the First Amendment, and thats the Communications Decency act, section 230 bars Interactive Computer Service providers, i. E. Facebook, google, twitter, from liability for content posted by third parties. It destroyed liability as we knew it. It destroyed the ability to control speech as we knew it. So we see this often referred to in Court Opinions as this wild west that happens online. These are platforms where content is being posted and yet theyre completely immune from liability, right . So it decouples, the cda decoupled that historically recognized publisher liability from publishers, i. E. In this case Interactive Computer Services, from online content. So as a litigator who has dealt for a long time with cda cases where people are victims of revenge porn or horrific other content, the cda means, sorry, you can go to facebook, you can go to twitter, you can go to tumbler and say, hey, can you take that salacious photo of this client who didnt authorize this to be there, and theyre like, sorry, third party content, were not liable and you cant make us take it down. So we saw the law come along and try to give us revenge porn. But at the end of the day this is allowing invasions of privacy, misinformation, and those deepfakes i talked about. If a deepfake of someone is online and it makes it appear theyre doing something they werent doing, they have a recourse against that in the precda era, right . Prior to this. Whats interesting is, the platforms, though, are permitted to make money off of this content. And a lot of money is made off of misinformation. Its not just money thats made, though. Weve heard from how much we are being distorted by actors that are putting content out there thats not real. Really quickly, we do have states, as i mentioned, revenge porn, and revenge porn laws, states were the ones who said were going to pass laws to protect victims of revenge porn. We now see california and texas trying to jump in and address this problem of deepfakes. Both states just passed laws that went into effect in october of 2019 where theyre making it a criminal offense if a person with an intent to injure a candidate or influence the results an election creates a deepfake video, causes the deepfake video to be published or distributed within 30 days of an election, and how do they define deepfake . Very simply, right . Theyre trying to address something that will cover a wide net. It means its a video created with ai with the intent to deceive and appears to depict a real person performing an action that that person that did not occur in reality. So we see states trying to jump in and fill this joyvoid. Thats i think part of the conversation i look forward to hearing from our panelists coming up next. Without further ado im really excited to introduce anthony robinson, who is a professor at penn state. And his maps and work with that. Ill let you talk more about that. And you can just, if you want to introduce maria so we can keep it moving. Thanks. [ applause ] thanks very much. My name is anthony robinson, im associate professor at the department of geography at Penn State University and im really happy to have this chance to talk to you today about mapping in the context of u. S. Elections. I want to that anything governor ridge and everybody else who helped organize this meeting. Its an important gathering for to us consider. So the next few minutes, i want to talk to you about mapping. Im a cartographer. Ive been working on research into how and why some maps go viral. Not all maps are easy to verify. No matter what, maps simplifications of our complex world. There arent any 100 truthful maps, that might come as a surprise to you. Public trust in maps is perhaps greater than ever. As an example of that, one of the reasons that i think this is the case is that millions of people every day are interacting with navigation apps like google maps, for example, and they work really well. We get from point a to point b and so our trust is reified by that. Election maps are messy. Maps that tell a story about data, what we refer to as thematic maps, its not as simple as navigating from point a to point b. My work at penn state is exploring what happens when public trust might be misplaced in that, such as when maps come out in an election, like in way, sorry, john king. Shortly about the 2016 election, nate silver shared a hypothetical Electoral College map exploring what might happen if only women were to vote. What i noticed is this map seemed to spawn the creation of hundreds of other maps all competing for peoples attention and coming with completely unknown sourcing. The example on the right was not created by this person on twitter but was shared by somebody with a blue checkmark who has a sizable audience. This is a sample of about 500 what if only blank voted maps that i collected from twitter. There were hundreds more that i could have kept collecting. One map in particular that went viral on its own and generated its own media access and multiple stories was this one. It took a fair bit of digging to figure out its purported creator who posted a blog post on medium after it went viral and claimed credit for the work. It wasnt easy to see where it came from. At penn states were exploring the ability to take Machine Learning techniques to take maps like that one and know where they come from, to know whether theyre derived from previous sources or represent something new on the internet. The results here show partial matches to the original viral map as detected by googles cloud vision platform. One variation simply asked the question i had, which is, where did this map come from . The others here appear to be accurate translations of the original into four other languages. But its not hard to imagine an inaccurate translation. And it would be trivially easy to completely change a text and change the message the map conveys. Leveraging reverse image platforms led us to look at other viral maps. Laura trump posted this in september. I wanted to know more about the map and the background. I made the assumption that someone in either the campaign or the white house borrowed the underlying map from another source. What we found is not only, yes, the underlying map was borrowed from another source, but also that it had already been used in some previous meming with a similar intent a couple of years prior. So its possible that this more recent map was inspired by some predecessor maps. We also noticed wihat many in te media noticed which is that more counties are colored red than are supposed to be. Really small changes in map design can lead to huge changes in how theyre interpreted. It would be easy to prepare a map that shows different primary results, maps that show the wrong locations for polling places, and for a lot of readers its very hard for them to tell when something is deliberately misleading them. Visualizing viral maps, a postdoc at penn state and i have built a prototype called macroverse which do reverse image source and helps us identify potential source materials for viral maps. It also lets us look at the derivative stuff that happens downstream. Youre welcome to try the prototype software. For example, with the impeachment map we were able to build an overview which shows dozens of related map images. We can review details about where and when those things were shared, how they emerged over time, filtered by similarity, and characterize the keywords that appear with them on different web pages. So were trying to attack this from multiple different angles. And reflecting on this, i would like to come back to a few. Some of the most crucial points, first is that maps are always simplifying our infinitely complex world. Thats actually a fundamental power of mapping. Thats a good thing. We cant handle all the detail that were surrounded by all the time, we need to simplify it. Second, maps have been used for disinformation for a long time, well in advance of the internet area. Thats not actually something new. The difference today, though, i would argue, is that its easy for virtually anybody to create and share Something Like this. I would argue also thats potentially a good thing as much as it is potentially a bad thing. In the past, maps were made by professionals and shared by media sources. When there was disinformation or persuasion happening with mapping, at least we had an idea where it was coming from. The technology to produce and share maps was available to very few people compared to the total number of map readers. Today, though, its possible to quickly refute one map with another. In some cases it doesnt help, in other cases it can be helpful, such as this example from the 2011 japan earthquake. This map was widely misused for a proxy for radiation emissions from fukushima. A similar example emerged with the coronavirus outbreak. Several media sources published and shared a map that purported to share the places where resist from wuhan had fled. This is actually a map from a few years ago which shows airplane routes everywhere in the world. Im curious to know how this got in the discourse. Is that just the product of random social charter or is there something underlying and intentional behind that . Potential new research direction, for those in the room who support the work or support funding and legislation, what can we do to make sources and modifications of maps more obvious to readers . I dont think we can stop them from being created, but can we make people enabled with better tools to see whats underlying them. Can we identify organic information sharing versus dmib acts deliberate misinformation. Im dealing right now with static things and its even more compelling to make something more animated. My colleague maria will talk about each of these issues in a broader context. [ applause ] hi, my name is maria molina. I want to start off by highlighting something that anthony mentioned and that is false news is not a new phenomenon. What is new is the source and the modality of false news. So the reality is that Interactive Media brought a shift in communication. Think traditional media. We had a source that delivered a message through a medium to a receiver. But now the receiver has also become the source. So what do we mean with that . Well, receivers can share, they can retweet, but they can also create their own content for dissemination among their networks. Do people distinguish between a professional source and their friends tweeting at them . What we see is that they actually dont. Another shift with Interactive Media is media grammar. Take for instance the newspaper. We had a clear sections and labels that distinguish opinion from entertainment to news. And we also had boxes. And those boxes helped us differentiate an advertisement from the actual information. And the same happened on tv. But lets look at what happened with social media. This is an example from my instagram account. You can see the new york times, sponsored content, and a brand that i volume. Virtually they look exactly the same. We see an absence of these categories in social media. But how do we evaluate content in the absence of these categories . We rely on heuristics or rule of thumb. Take for instance the bandwagon heuristic. This is triggered by the number of likes that you see in the news shared in your social media. And this gives us the idea that if others think this article is good, it should be good for me too. We text this premise and we gave participants in an experiment, half of them a real story and half of them the false version of the same story. So what happened in social media is when someone publishes a real story such as this one, theres always someone who comes up with a false version of the same story. We also gave them, half our participants had the article with a low number of likes and half with a high number of likes. What we found is that those participants who received the false news with high number of likes versus those that received the false news with low number of likes or real nice with high or low number of likes were actually more likely to comment or share this particular news story. See, what happens is that when false news has received a lot of likes, it elicits agitation and that persuades users to actually their this content. So in answering do other users matter when evaluating content online, the answer is yes. Yet another hue iie iis heurist have is that seeing is believing. We looked at the context of india using the whatsapp messaging service. For participants who sought story, thought it was more credible than those in the audio or text, they also were more likely to share it more. Now, this is important with deepfakes where we can create a video and have anyone say whatever we want. So in social media, we rely on the heuristics to evaluate content. We also know there are those absences of categories. So its our responsibility as communication practitioners to be able to more accurately convey these categories for communication purposes. Thats something im currently involved in in a project where were creating a Machine Learning algorithm to be able to classify the taxonomy of information where we identify the different types of content. You can distinguish these based on unique features. So for instance, here we see features of real news. The features include linguistics, sources, structural, network features. Linguistic features are Something Like this. The word should in this example is an indicator that this is likely to be an opinion piece. Here, the seattle tribune, this does not even exist. An indicator that this is actually a false website. We also have structural characteristics. This seems to be an abc news article but if we look at the url, that ending is a giveaway that this is a false website. Finally, we have network characteristics. Look at the diffusion of information. All of these are unknown sources. On the other hand, we have this diffusion network. Here we have the presence of a news organization. So real information is more likely to be picked up by news organizations. So what do we do . We have these features and what do we do . We feed them into a supervised learning model. What we do is feed them, features from false news and deep news and we train it based on previously classified articles. These articles have been classified by organizations such as splpolitifact or snopes. We test the model to see if we can distinguish between false and real news. These are binomial classifications. Our approach is actually an monomial classification. So there is a little bit of how our approach will work. I want to point out that this is a probablistic indication. Our algorithm will first figure out if this article has narrative writing, say the answer is no. It will go to the next route. Source verification. If it has no source verification it will go to the next route, and so on until an actual categorization or classification has been reached. Furthermore, we also provide an accuracy score and allows us to further assess the accuracy of our classification. Can we detect misinformation . Were currently working on these algorithms and we see some promising results. Were very thank you. [ applause ] were going to get started. Give us a second to call up our panelist for this next conversation about social engineering. Im excited to introduce our panelist. If everybody can come up, that will be fantastic. Thank you. Let me introduce our panelist here. I want to thank the speakers we had. Kevin and maria, i think you gave us such great examples of both how were being manipulated in believable data we think is trustworthy and to see the work youre doing with other researchers to identify this in ways thats not just that are sophisticated. Not just short. I have Carrie Cordero sitting to my right. Shes general counsel to the assistant to the attorney general. Shes a legal analyst with cnn. Also, shes an adjunct professor at georgetown law and cofounders of checks and balances. Org. Sitting next to her is emma llanso who is the Free Expression project manager for cdt. Next to emma we have josh who is the executive director for Georgetown Institute for constitutional advocacy and protection. We have kevin munger next to josh. Were going to do the panel as a conversation in the same way that we did the other one. Were going to try to have time for q and a in the audience. Carrie, if i start with you, we heard from some of the speakers earlier a bit about what were learning about russian and other influences. Can you give us what we know about our elections from the Senate Intelligence committee . Sure. Thank you to ann for hosting us today and bringing together this terrific program. I want to start out because we do with some information about the Senate Intelligence Committee Report because the senate has been doing an indepi an indepth investigation into the 2016 election interference. News has focused a lot in terms of what we are previewing to happen in 2020 and what type of election interference activities might take place over the course of the next several months but the Intelligence Committee report has done a valuable service in showing us a little bit of a road map of what we could see again and i think it hasnt gotten enough of a billing. I want to highlight a few things it does. So far three volumes of the Senate Intelligence committees work have been released. Those three volumes have been declassed and their investigation, they are producing these volume ons a ro rolling basis. As soon as they finish, they put it through the declassification. Volume one of the report focused on russian efforts against election infrastructure. Really getting to a lot of the details that we heard about on the earlier Panel Discussion today. The russian efforts in 2016 to probe election Infrastructure Systems down at the state and local level. Interestingly, the report says there was no it says it several time which is is why im focusing on it. It says there was no evidence discovered that actual vote tallies for Election Results were affected by the russian interference efforts. It also says and it appears several times tlouts volume one that the intelligence communities view into whether that is the case was have limited. One of the things that i hope we can think about Going Forward is whether or not there are ways we actually in the future to verify that information because there doesnt seem to be an auditing capability. The second volume of the Senate Intelligence Committee Report focused on russian use of social media. This is much of the information that in our Public Domain we heard come forward throughout the special counsel muellers investigation. In other words, the activities of the russian intelligence services. The entity they formed known as the Internet Research agency which was conducting online activities, disinformation, fake pers personas, interacting online with americans in the efforts to affect the election and the election discourse. Its described as information warfare. This third volume which ill highlight and well open the discussion focuses on the u. S. Governments response to the 2016 election interference efforts. This is the area im most interested in us doing better this time. We know from the comments of the fbi representative earlier in his talk and the department of defense and also just from public reporting that there are certainly some types of activities being conducted by russia, potentially other countries but we know russia and two, play some role in interfering or affecting or taking an interest in the 2020 election. Much of what volume three of the Senate Intelligence reports describes is the difficulty that the Political Administration in 2016 had with figuring out how much information should they make public. There was a volume three lays out was this intense deat a time that took place within the administration about weighing whether they should make information public. Whether that would further undermine the confidence in the election. Those are the three things they laid out so far. Whats important is it was a bipartisan report. Several members did write additional views at the end of the report. Question what types of activities were seeing. Im grateful the economy done this work in order to establish a historical record. Ill pause there. Its promising for how we live. Its ubiquitous. Misinformation flow freely and normally without consequence. You have the content providers. You have to buy the account. It had an army of trolls. It had a couple thousands accounts on facebook and twitter. Where do we go in terms of trying to regulate content. Other challenges associated with this kind of thing, freedom of speech, First Amendment. Theyre not yelling fire in the crowded theater if theyre disrupting the process. Thats the main question. That is several gigantic questions. Ill try to do the three to five minute answer to that. We can get into this a lot more. There are a couple of key things to think about when thinking about the role of law in all of this, particularly here in the United States. Someone the question about the First Amendment and how much protection it provides to speech. Its three or four different kinds of speech that have maybe different First Amendment backgrounds to them. You have this question of commercial speech. Commercial speech in the u. S. Does receive a lesser degree of protection than say core political speech but typically when talking about commercial speech were talking about speech proposing a transaction. Buy this brand of toothpaste. Hire me to clear your house. Thats not what were talking about when talking about political advertising which gets very high level of First Amendment protection. Folks may remember the Citizens United case from about ten years ago where the Supreme Court found that independent expenditures were really crucial in their view to people being able form associations and promote their political things. That gets about as high First Amendment protection as say the press does. Its also important to understand that even lying has some First Amendment protections. Theres a case that dealt with the stolen valor act that tried to restrict people from falsely claiming they had gotten Different Military honors or served in the military on the basis that was really damaging to the valor acts of true military service. The Supreme Court found that law, if applied only because to people lying without any intent to defraud or sever military benefits or some other component to the act was unconstitutional. Just restricting someone from falsely claiming they served in the military hindered their First Amendment expression. We start diving into slanting reporting and fraudulent statements with that. Quickly, a couple of things on the online speech component of all of this. One of the key factors to keep in mind when ever were talking about online speech is the sheer scale of what were dealing with. You may have heard the facts of 400 hours of video are up loaded to youtube every minute. Theres millions or billions of posts and tweets up loaded on social media. Our traditional publisher liability that ann was talk about in her presentation really does not apply the same way in the online speech as it did say in the newspaper publication days. The law that ann referenced, section 230 that sets out the liability frame work for hosts of third party con tetent onlin was passed with specific reason. There were a couple of Different Cases trying to apply traditional publisher liability to online speech. Compuserve did no mods ratierat. The court said you are just districting this content. You didnt know about it. Prodigy did moderation. They took down comments that were off topic. Shay rr trying to shape their Message Boards into totally open free willing environment. That meant they had more liability for the comments that were posted because pay were acting more like a traditional publisher. It made sense as a way to apply traditional publisher doctrine. Youre probably going to end up with a lot of spam and other material that maybe going to interfere with constructive communication. Also out there because i know we have a lot of other things to get to. Thats a bit of the ground work and not answering. Ill take it from there to ask josh a question. The original purpose of the cda was to ensurge sites to cull through for content. They didnt want sites working to take down unlawful content. Its a very important thing. Josh, we have heard all about the election infrastructure, the technical infrastructure. We have specific examples about what were doing to improve that infrastructure. I really like this term you have that i came across or that you emailed about what are we doing as a nation to upgrades the nations cognitive election infrastructure . You want to talk about that . Sure, happy to. Grateful to be part of this great gathering today in these interesting conversations. In some ways i think about the threat to Election Integrity as roughly divided into two buckets. One you might call threats to Data Security and one you might call threats to discourse. Data security you might think about as threats to the voter roles and somebody reaching and changing that or the vote themselves. You heard a lot about how our election infrastructure, the machines you go vote on or the databases that, oh, information about who is registered to vote, theres been progress. I think its fair to say. Not as much progress but theres greater awareness of the problems and the threats. I think its fair to say there is some work thats been done through some parts of the country to address those. Thats the election infrastructure to deal with data intent. Then theres the discourse intent. The discussion of candidates, views, who they are. Even discussion of whos allowed to vote, when theyre allowed to vote, whether they should bother to vote and that requires a cognitive structure. It attacks vulnerabilities if people are willing to believe what is essentially junk but not just any junk. Junk thats been tailored made for a particular audience and m microta microtargeted them to convince them the pope has endorsed a particular candidate but the pope never evendorses a candida. As carrie said, in 2016, the threat to data integrity was the one that seemed to have been probed to use exactly the right word. The word i think the Senate Intelligence Committee Use is probe but we dont see the data having been changed. The discourse integrity was a mess. We all know that. Thats clear from the senate reports. Its clear from media reporting. Its clear going back to what was circulating especially online but amplified in other settings in 2016. On that front i worry the cognitive infrastructure has gotten worse than better. We see that wide portions of the audience, the americans, especially voting americans but even other americans who talk to voters they continue to digest and share and reshare things ge appealing in 2020 than it was an 2016 because a lot of it is originating right here at home. Let me give you one concrete example. I was curious when the Inspector General of the Justice Department had released his long anticipated report on some of the 2016 activities. I was curious how that was being reported by rt, essentially a propaganda arm of the russian government. They will skew this in some way, but i wonder how. I looked at the rt story online but i wonder how. I looked at the rt story online within hours of the report beine released and it began with a big block quote that did skew what was found. That big block quote came from the attorney general of the , b United States, bill barr. Its what he had said but it wa. Disinformation. It was not an accurate rendering of what was concluded. That strikes me as more concerning. Interesting to point out to that. One of be specific recommendations was about education. You want to address thothat . Sure. The volume three of the Senate Intelligence report focus on the considerations that were being given within the administration about what information they should put forward. You may remember it wasnt untie october of 2016 that the u. S. Government, the department of Homeland Security jointly with the Intelligence Community released a public report that described what the election interference activity the put ans were engage ing was out. October 2016 was late. There had ban lot of hang wringing within the T Administration whether to even do that or whether theyhe t cou have done more. Woul theyre judgment, at the time, was more informationn publicly would undermine confidence in the elections. Incumben i have a different perspective e forward. Oing i think that it is incumbent upon the Intelligence Community and the u. S. Government more ao broadly to inform, certainly congress, and as much as possible, the public about whate it is that is their best assessment of what the threats are to our Election Integrity. The reason is that because we dd see what just happened last week. Normally intelligence oversight isis conducted behind closed dot to protect sources and methods. H we have theat we intelligence communities that are set up to do that. Clos in the environmentedd were in right now, what we saw last weet is we saw one of those closed door briefings take place. N immediately, bits andot tel pie from the briefing came out and now i couldnt tell you sitting here today what actually was the truth of what was the mem repor that brief frg ting from the Intelligence Community. Ive been encouraging that the government be more forthcoming of the information thats indicative of election be thre. There was far too much information that was redacted. Key findings of the report are redacted. At a key paragraph that is the heading, Intelligence Communityk awareness about whatin the thre was is redacted. That doesnt help us looking forward. There i might just observe based on my experience theres a lot of over classification of what is considered to be secret or top t secret information that i think, unfortunately, put a restraint on the kind of information the public needs know and deal with without compromisingg sources. Thats an observation ive made, during my time in government. Kevin, youre not going to get off the hook here, pal. I remember back in 82, when i first ran, if you were going to with a literature drop or radio ad or tv commercial, it was still a large area. You can get down to individual e street or the individual constituent. You got this microtarget ability. This information follows to that information or group of attent individuals. Io not paying too much attention. This creates very pervasive impact on that. Id like you to comment on it. Sure, thanks. Ask a i really like the use of that term digital literacy. Obviously something is going onh now. Something omnovel. If you look at the postmortem of what sciences have shown from 2016, theres really significant in terms of was 4 consuming and sharing fake news on facebook. It seems like people over 65 th were four to seven. Were looking at times more like to share misinformation on racy. Facebook than younger people. Wh age is simply a proxy for or the digitalir literacy. There are people who have been using the internet for their entire adult lives. If we look at who is joining facebook, a group of people most likely to be joining facebook were people who i think what weal have come to realize is thn the internet andd social media and the systems were designed for and by the young and tech iv savvy. If you at werena an early adop the internet, you happen to be p the kinder of person using technology. As theype are pen tratsing deep into our society were realizing there are some people who donth have this erpath. D how th they didnt get the training. Rt i think its tempting to put a lot of the blame on facebook. Sh. I think there are certain things they have done to make it too a easy to access, too easy to share. They move too quickly and change the world too quickly. Their whole arrogant slogan of t move fast and. Break things. They have succeeded in that. I think its the way the world is. You have seen some of the legacn information of the systems more top down. Now that we have almost every e one in u. S. Society on the e internet on facebook. Were seeing the presence of a small number of peoplepo who do not have access to high levels r literacy pose a pretty significant threat to the rms of Overall Health of our t information system. We have to think of these people in terms of weaknesses in our et collective socialio verificatioe of information. Yone u we do some kind of quick fix in media a literacy, getting every one up to speed and how the fuy internet operates. Moving forward if were going to live in a world where we have democracy and the fully mature l internet that we have evtoday, have to realize we need every tl one in our society up to speed and the level of informational education that we have in the u. S. Today, its not going to be acceptable. How do we embrace the First Amendment and address this information . Who is the gated ever keeperu truth . Okay. The short answer is i dont think youre ever going to get people to agree on who should bn the gate keepero of truth. Thats the fundamental problem. T some of us might think no one ever should be that. Some of us might have some at candidates in mind. Is theres not going to be a consensus definition there. I think thats part of what the First Amendment iss trying to protect for all of us the has be freedom to form our own opinions but so much of what has been discussed has been about this question of like whatalook athorough authority authoritative questions. Ion i think any answer to misinformation online is going to have to be multifacetted. Pi theres no silver bullet. Theres no, you just put this ab label onel content and everybod understands that label to mean the same thing, responds to it the same way and your problem it solved. Ots of i think not to try to do an entire summary of all of that research but there are lots of different contexts online. Of t you talk about facebook as oneh platform as its one thing. There are hundreds, thousands, millions different of groupingst of people on facebook who will f respond to some of the interventions facebook is doing in different ways. , whether theres labels of information by third party facti checkers. Whether its amplifying information to ho or removing i whether its addresmicrotargete. The fundamental answer to how do you start addressing or do o better at addressing this information is actually more tos the points on transparency and r being able to get more independent researchers access to information from the different Online Platforms so Different Research studies can be done. What we have is a real lack of trust in society. Of more ability to do verifiable research on dynamics and impacts of proposed solutions. I do think the Major Companies facebook, twitter, we youtube are different ideological orientation as they approach this problem in 2020 from where they were in 2016. Over simplifying for purposes oa a succinct discussion, it would be fair to describe them as dme roughlydo hands off. First amendment doesnt apply to them. They made an institutional choice, commercial choice to be hands off. That has not yielded them great applaud as we have digested what happened in 2016. That strikes me as something that needs to mature very g quickly. They make some decisions and gem criticized within the decisions look bad in principle or implementation. Notice when facebook was over in europe they may decide in response of european inquirienq maybe a toe or a foot or up to the ankle in because they were threatened that if you dont sor a little more content aware, be agressive to take this misinformation off, we may have to regulate you. Questions . He had his hand up. Im michael nelson. Where should our priority be . Professor mckenna made a case that section 230 is causing a lot of this problem. Is a ould several of usus believe theres bigger problem. Is there a third thing we should focus on first . Hat the what sense of priority here should we have . To quickly describe what the third thing might be, ill point to frame work that a research has talked about, the abc frame work of actors, behavior and fit content. If 230 is about content and liability for content can you lke rules or t t laws, thats extremely difficult to truly judgege is this a lie . Is this intentional inte misinformation. We can talklk about deep fakes nown too. The texas law is really speciall interesting. Scale. Content is kind of a known really thorny issue especially to evaluate its escascale. Thats anan interesting question although when you start getting into ideas of regulation arounda limiting peoplesme ability to speak anonymously online. Also m one, you have First Amendment of concerns or Free Expression concerns more broadly but you es also may run into the risk of fe crafting a regulation that of prevents kind of regular people from taking advantage of the mt benefits online and the way that can express people to express s things they mightth face reprisr for. The important element but thoseo regulations still be circumvented by highly eaves resourced, especially foreign actors to kind of circumvent those protections. H that that leaves us with b, behavior and looking at this is an appropriate that all the of the . Major social Media Companies have taken at is looking at what can you see among patterns of behavior online. I think facebook calls it esearc coordinated and authentic perti activity. Latfor twitter looks at drawing from their research and expertize t e around fighting spam on the f t platform. Re there ways that areusly characteristic that tend to look more like previously identified patternsns of misinformation thg spreading rather than kind of naturalified human behavior. A lot of this get sbos into is information being amplified by bo bots. Is this designed for a specific purpose and more quote unquote natural cl natural collection of people. A lot of the research is around this behavior question. Is there something kind of more discreetly measurable in the patterns of sharing of ck information that could be t targeted. E gipeople dont understand the world is so different today than it was 20 years ago. I give that background. I do, to emmas point, and the researchers we heard today, i think the reason were having en this suchch an interesting focu. Is this will take an interdisciplinary fix. I think thats requiringg technology. I think its requiring Significant Education of people of all ages like i guess if cric youre 50 youll have toal walko the right and kevin will teach you. I think thats to critical her. Kevin wants to make an observation as well. The majority of misinformation on facebook is r financially motivated. That is an area where theres potential for regulation which could not file with any First Amendment concerns. He was an entrepreneur. Hegood said i was in california. It was hard to start a business otherwise. This was a good way tome get on the internet and make some money. If facebook can put some more frictions in this t seemprocess would be more restrictive abouty who can make money from advertising. Thatsi seems like an ease y way reduce overall misinformation. Shoo shes bringing you the mike. be there in one second. A fel thank you. Lo is there a chance to address this issue with some russian actors either in the governmentl either in the business, either in the Civil Society because, well, russian political system is complicated. Thank you. Theres space for governmento actors to have a role in this. I think a lot of our conversations is focused on theh role of the private sectors bute theresct a role for government. So far what we have seen is frs government has underestimated the role of modest diplomatic nn efforts russiag could take in terms of dissauading russia fro engaging in these maligned activities. It waswhole lo not so effective extent they took them. I think the public doesnt have a whole lot of clarityt the cu regarding, what, if any, activities the currenT Administration is doing to dissuade russia maligned e activities. I think its a very valid point theres a role of government. Pa one thing to keeptf this mind wn it comes to the role of the Technology Sector and the of oto content platforms is there are all sorts of other industries that the u. S. Government regulates and takes on when they play a significant role in the s u. S. Economy or society. For some reason and i think theres probably a variety of se factors that we wont get into here, congress has been incapable of doing anything in this space. A whether its limits on abilities to prevent corruption and the way American Companies do bout t businesshe overseas and thinkine about the foreign corrupt compan practices act that weie regulat the actions of governments of t companies to engage in activities over seas. When we look at the traditional Telecommunications Infrastructure in the United States. Outwe look tacross anyhat n o industry, we see the u. S. Congress has been capable in the past of doing something about it. This is the last question. Well take a quick break. Theres lunches out there. That is part of the fix thats needed. This question is for carrie. I think you made a point about. Being able to get information out in a timely manner with regards to election threats and being moresome forthcoming. Im interesting in your thoughts on that in a practical sense considering the events that thi transpired last week. I for folks who are inss governme i think theres some really p interesting questions and issueh and imarin sbreinterested on yk background and perspective in terms of how folks ought to share that information and what that would look like. Th ill give you one quick answer. My First Response of something e that can be done immediately is for the Intelligence Committees is to schedule the worldwide threat. This ise an annual briefing thai thety Intelligence Committees d every single year. Its the one opportunity that thehe act intelligence communit leaders, that would be whoever is the acting director of National Intelligence of the dan plus the other heads of the intelligence elements and they s appear before the Intelligence Committees in an unclassified en section and there can be a classified section after ssion. They have to prepare a written statement so the white house th would not be surprised of what goes in the statement which is what we heart about perhaps what happened last week. There would be a vetted written y woul statement. It would coverd of t allheit so threats but threats to Election Integrity could need to be one i and it could givety members of congress to ask the Intelligence Community leaders questions about the threats to the 2020. Election integrity under oath. I was going to suggest two i things. That was one of them. D the other not mutually exclusive. It would be for either Intelligence Committee or both committees to call the hearing a specifically onll the state of 2020 election related threats. Means that not everything can b shared but it means something can be shared and to ensure the people and 2 joe mcguire shoulde among those called to testify. Current dni, whoever it may be at that moment in time, also should be called to testify butf to have talking points prepared for them by the rank and file oo the Intelligence Community who i firmly believe continue to do their very, very best to identify and track theseabout and help inform policy makers about how to address them, whata can be shared at the eserving unclassified level should be. Exo were all sitting here wanting know and really deserving know in the context of our functioning democracy. Okay. Thank you. Thank you kevin, josh, emma. All right. Lets grab some lunch and come on back in. You can bring your lunch in here. Thank you. If you missed the m before the Senate Committee you can watch tonight beginning at 8 00 p. M. Eastern. Centers for Disease Control and prevention director and other federal officials will testify on the coronavirus response. Watch that on cspan3. Coming up, governors delivering their state of the state addresses. Well begin with tennessee governor billie o lee. Well hear from delaware governor. Hes proposing a 50 Million Investment in clean water infrastructure. Then later, georgia governor discusses ways he wants to combat Human Trafficking and how he wants to make health care more affordable. First, here is tennessee governor bill lees state of the state address