tv Russia and Social Media Hearing CSPAN October 31, 2017 2:28pm-5:01pm EDT
ontinued by the democrats. what we know from the research is that donald trump or the trump campaign and the russians were talking to each other. and their conversations, again, as we know from yesterday, were about serious things. nterfering with our democracy. it was done by the law firm that was hired by the campaign. i don't know how much of the opposition research was the fusion opposition research. i have not disaggregated that amount. but what i do know is the issue at hand is whether the trump administration, trump campaign, and the russians were conspiring to affect this election.
papadopoulos' r. guilty plea yeand the indictments that came down yesterday is serious stuff as it relates to collusion. [captioning performed by the national captioning institute, which is responsible for its caption content and accuracy. visit ncicap.org] [captions copyright national cable satellite corp. 2017] >> we are live on capitol hill this afternoon as congressional committees are holding three hearings today and tomorrow. looking at russian influence on the 2016 presidential election. this is the first hearing that we're about to show you. this is the senate judiciary committee here on c-span. officials from twitter, facebook and google will be explaining how russian organizations purchased ads and they spread fake news.
order. here's sort of the order of battle today. i'll make an opening statement. senator whitehouse, senator feinstein would like it make one. i think senator grassley comes, he'll certainly be allowed to do so. the title of this hearing is extremist content and russian disease information online, working with -- disinformation online, working with tech to find solutions. that's exactly what we want to do. we're here to try to find solutions. let me describe, i think, the challenge. i doubt if i would be here if it weren't for social media, to be honest with you, president trump told fox news on october 20, 2017. so this is the president of the united states saying that from his point of view, social media was an invaluable tool to help him win an election. i would dare say that every politician up here today asking you questions uses your service and we find it invaluable to communicate with our constituents and get our message out. not only do we use it, not only
does the president use it, millions of americans use your technology to share the first step of a grandchild, to talk about good and bad things in our lives, and i would like to say to all of you, you've enriched america. we have more information available to us because of what you do. we can find the answer to almost any question. we can share aspects of our lives with those who mean the most to us. and we can talk amongst ourselves in 140 characters. some people are better at that than others. some people should probably do less of it. t the bottom line is these technologies also can be used to undermine our democracy and put our nation at risk. the platforms that i've just described that add value to individual american lives and to our country also can be used by terrorists to recruit in the cyberworld people of their cause. can be used by foreign governments.
we've seen an example of that in 2016, to create chaos within our democracy. information is power. ideas are the essence of democracy. the exchange of ideas, being able to criticize each other is one of the things that we cherish the most. but what we have to be on guard as a nation is having people who want to undermine our way of life using these platforms against us and i think this is a national security challenge of the 21st century. here's what general petraeus said about jihadists online. jihadists have shown particular facility in exploiting ungoverned or even inadequately governed spaces in the islamic world. they're also exploiting the vast largely ungoverned spaces in cyberspace. demonstrating increasing technical expertise, sfuftcation and media production and agility if the face of various efforts to limit its access. it is clear that our
counterextremism efforts and other initiatives to combat extremism online have, until now, been inadequate. i think that's a fair statement. and the purpose of this hearing is to figure out how we can help you. i believe that each of you in your own way are taking these problems seriously. the one thing i can say without a doubt, what we're doing collectively is not working. we had a foreign government apparently buying thousands of dollars worth of advertising to create discontent and discord in the 2016 election. you have foreign entities going to websites to create fight among americans. like we don't have enough to fight about on our own. so the bottom line is these platforms are being used by people who wish us harm and wish to undercut our way of life. if you're a man like putin, democracy is your worst nightmare. if you live in putin's russia, the idea of exchanging information about what's good and bad about your government
is something you dare not do. because you won't last very long. so, to those who wish to undermine the american way of life, they found portals into our society that are intermingled with everyday life and the challenge of this earing and of this focus is to how to keep the good and deal withed bad? we'll never be 100% perfect, but the goal is to be better than we are today. and to the extent that legislation can help, we'd like to know about what we could do to help. to the extent that the status quo is acceptable, we all want to be on the record and say it is not. so with that i'll turn it over to senator whitehouse. white white thank you, senator -- whithewhithe thank you, senator graham, for -- mr. whitehouse: thank you, senator graham, for organizing this hearing into russia's meddling in the 2016 election. i'm very proud of the work we
are doing on this issue. i hope it will continue and i hope that you and your team see me and my team as loyal partners in this effort. understanding what happened, how russia applied the varied methods in its election interference tool box to interfere with our democracy, is an important step toward protecting the integrity of future elections and of our democratic process. each hearing the subcommittee holds gets us closer to that understanding. but our -- at our first hearing back in march, we talked about the subcommittee's intent, and i quote, to begin a public conversation about the means and methods russia uses to undermine democratic government. we heard testimony from expert witnesses who outlined the various tools through which the kremlin exerts influence abroad. from traditional intelligence methods like compromising corrupt business and political figures, to hacking and leaking
stolen information, to disease information, propaganda -- disinformation, propaganda and provocation through both tradition alameda and social media networks. at a subcommittee hearing in may, i went through a list to see which had been deployed in 2016. we'll learn more today about one of those methods. propaganda, fake news, trolls and bots. from representatives of some major american tech companies and from outside experts. the russian government exploited social media platforms as part of a wide ranging disease information campaign targeted -- disinformation campaign targeted against america and american voters. as we explore how that campaign works and how we might better ininsulate ourselves in the future, let's recap what we've learned in our hearings and what we still don't know. we certainly sought hacking and theft of political information by russia, something no serious person can dispute. timed leaks of damaging
material were the fruits of that crime. we know they happened. but we still don't know how the decisions were made about what to leak and when and who made them. it's been reported that trump confidant roger stone, we learned last week from a press account that the c.e.o. of a data analytics firm that worked for the trump campaign offered assistant to julian assange. and of course we now have the statement of the offense prepared by the mueller probe. but we don't know the full story of who coordinated with wikileaks or even directly with russian hackers. another method we've heard about is the ex employ tavings shady business and financial ties. we've heard testimony from a number witnesses, both hear in the subcommittee and at hearings of the helsinki commission, that the u.s. has become a haven for secretive shell corningses that can allow foreign influence schemes to channel funds to compromised individuals and exert political
influence. we still know next to nothing about the president's business dealings in russia or with russians, except that he's long chased after deals there. the president's tax returns would clarify a great deal and hopefully put an end to some of these questions, but those tax returns have not been made public. paul manafort's long history of suspicious business relationships with russian and ukrainian oligarchs has now yielded his indictment. the indictment exposed gaping holes in enforcement and in picking up on false statements and international money laundering. if you can use his alleged scheme to buy property, why not use it to make unanimous political expenditures? or spend money to influence elections? we still don't have answers about the president's curious relationship with felix, who was chasing russian business in consultation with senior trump organization executive well after the presidential campaign had begun. we haven't been title of the
bill speak with him or cohen, so he we still don't have -- so we still don't have answers on this front. we know the ruppingses -- russians try to corrupt political figures. we don't know to what extend tenth that happened here but we do know the trump campaign and administration has had a very bad habit about forgetting about meetings with russians. michael flynn is still the only person to have been held accountable for hiding improper contacts with russia, even as more and more such contacts have emerged in the intervening months. paul manafort, jared kushner and the president's son with -- met with a russian lawyer sent to give damaging information about his opponent on behalf of the russian government in 2016. urb in her reflects -- 100 foreign contacts kushner left off, including meetings with am bass doer and the head of a major russian bank. the leaders of the judiciary committee sent letters to the
white house in june and july of this year with questions about the status of mr. kushner's clearance to. to this day, those questions have been ignored. nearly six months after we first ran through that check list, we still have more questions than answers. my sincere hope remains that we'll find those answers so that we accomplish this subcommittee's primary purpose, which is to help us learn how to protect the country from foreign political influence in our elections. today we have an opportunity to learn more about how russia exploited social media as part of its disinformation campaign and to share some of those details with the public. i appreciate the cooperation of facebook and twitter and google in sending representatives here today and in working with our staff over the last several weeks to voluntary produce information. the intelligence committee assessment published in january reported that, and i quote them here, moscow's influence campaign followed a russian messaging strategy that blends
covert intelligence operations such as cyberactivity with overt efforts by russian government agencies, state-funded media, third party intermediaries and paid social media users or trolls. ussian state-backed networks are an important disseminater of messages designed to undermine confidence in the legitimacy of western institutions and governments. social media troll armies like the one operated by the st. petersburg-based internet research agency helped to amplify those meages, often posing as americans on facebook and twitter, to launder russian propaganda messages and obscure their russian origin. according to iranian scholar, russian media, quote, implant propagandist narratives in the international media sphere and they do so with the express intent of having them picked up on social networks. in russia's best case sent year, traditional meetia --
media will then give it legitimacy. when narrative laundering is uccessful, according to him, propaganda narratives can become part of the mainstream media sphere. how can western democracies interrupt this vicious cycle while respecting our commitment to freedom of speech? greater transparency and disclosure about the source of information, especially paid political advertising, is a necessary first step. but our adversaries have access to tools well beyond traditional advertising. they are using our own social networks, our friendships, our families and our biases and viewpoints against us. to achieve their political ends. i look forward to hearing from our witnesses about the ways we can work with the tech community to ensure that we are prepared to confront russian disinformation in the future. and again i express my appreciation to our chairman, enator graham.
>> thank you very much, mr. chairman. i very much appreciate your courtesy in permitting just regular members to be here and participate. it's very much appreciated. fine fine i had a briefing -- ms. feinstein: i had a briefing last week by outside technical experts and i saw really for the first time how effectively russia has harnessed the tremendous and, quite frankly to me, frightening power of social media. they showed us how millions of americans are reached and how russia's successfully used fake accounts to imbed itself, to shape and manipulate opinion and actions. so it shouldn't be news to anyone that russia interfered in the election. what is really staggering and hard to fully comprehend is how easily and successfully they turned modern technologies to their advantage.
russia used covert cyberattacks to obtain and release information to impact the election, as well as propaganda campaigns that relied heavily on r.t. formerly russia today. the state-run television network. and the internet research agency. a group of professional trolls reportedly financed by a close putin ally with ties to russian intelligence. documents and information that we have received from facebook, 2013 and google confirm this -- twitter and google confirm this. just a few more facts. facebook has identified 470 accounts tied to the internet research agency. itter has identified 2,752 i.r.a.-related accounts. and almost 37,000 russian linked accounts that generated
automated election content. from what we have seen so far, russian-backed trolls use fake accounts on facebook for more than 3,000 paid advertisements. nd those ads sought to sow discord and amplify racial and social divisions among american voters. they exploited hot button topics such as immigration, gun rights, lgbt and racial issues, to target both conservative and progressive audiences. so, mr. chairman, and ranking member, this is really a critical hearing. because it's the first time we will have heard, at least to my knowledge, from the three agencies about exactly what is going on. and most importantly, what they are prepared to do to stop it.
thank you very much. [inaudible] mr. graham: i want to thank you all for coming. you've been very helpful and cooperative and we appreciate it. to my left, mr. collins is the general counsel for facebook. thank you very much for coming. acting general counsel for twitter. and the director of law enforcement and information security at google. i'm sure you got these jobs because you're very good at what you do. >> chairman graham, ranking member whitehouse, and distinguished members of the subcommittee, thank you for the opportunity to appear before you today. my name is collin stretch and since july, 2013, i have served
as the general counsel of facebook. we appreciate your hard work as you continue to seek more effective ways to combat crime, terrorism and all other threats to our national security. mr. stretch tpwhrverage we are deeply concerned about all -- mr. stretch: we are deeply concerned about all of these threats. at facebook we create innovative technology that gives people the power to build community and bring the world closer together. we're proud that over two billion people around the world come to facebook every month to share with friends and family, to learn about new products and services, to volunteer or donate to organizations they care about, and to help out in a crisis. being at the forefront of technology also means being at the forefront of new legal, security and policy challenges. our teams come to work every day to confront these challenges head-on. thousands of facebook employees around the world work to make facebook a place where both personal expression and personal safety are protected and respected.
i'm sure today to address two important issues for our platform and others like it. the threat of extremist content and the efforts by foreign actors to interfere with the 2016 election. keeping people safe on facebook is critical to our mation -- mission. and there is no place on facebook for terrorism or hate. we remove terrorists and posts that support terrorism as soon as we become aware of them. and in the rare cases when we uncover evidence of imminent harm, we promptly inform authorities. while there are challenges to fighting terrorism, we think technology and facebook can be part of the solution. we also believe we have an important role to play in the democratic process. and a responsibility to protect it on our platform. when it comes to the 2016 election, i want to be clear. we take what happened on facebook very seriously.
the foreign interference we saw is rerp henceable. that foreign actors -- rerp henceable. that foreign actors behinding -- hiding behind fake accounts used our platform to sow discord and try to undermine the election is directly contrary to our values and goes against everything facebook stands for. we build tools to help people connect. and we recognize that facebook has become an important tool for political engagement and debate. our goal is to bring people closer together. these foreign actors sought to drive people apart. in our investigation, which continues to this day, we have found that foreign actors used fake accounts to place ads on facebook and instagram that reached millions of americans over a two-year period, and that those ads were used to promote pages which in turn hosted more content. people shared these posts, spreading them still further. many of these ads are
inflammatory. some are down right offensive. and much of it will be particularly painful to communities that engaged with this content believing it to be authentic. they have every right to expect more from us. and we are determined to do better. in aggregate these ads were a very small fraction of the overall content on facebook. but any amount is too much. all of these accounts and pages violated our policies and we removed them. going forward, we're making significant investments. we're hiring mourad reviewers, doubling or security ennearing efforts, putting in place tighter ad content restrictions, launching new tools to improve ad transparency, and requiring more information from political ad buyers. we're building artificial intelligence to help locate more band content and bad actors. we are working more closely with industry to share information on how to identify and prevent threats so so that
we can all respond faster and more effectively. and we are expanding our efforts to work with law enforcement. we know bad actors aren't going to stop their efforts. we know we'll all have to keep learning and improving to stay ahead of them. that's why i want to thank you for this investigation. we look forward to the conclusions will you ultimately shear with the american public and i look forward -- share with the american public and i ook forward to your questions. >> members of the committee, twitter understands the importance of the committee's inquiry into extremist content and russian disinformation in the 2016 electionment avement and we appreciate the opportunity to appear here today. the events underlined in this hearing have been deeply concerning to our company and to the broader twitter community. mr. edgett tpwhrverage we're committed to providing a service that fosters and facilitates free democratic debate and that promotes positive change in the world.
we are troubled by reports that power of twitter was misused by a foreign actor for the purpose of influencing the u.s. presidential election and undermining public faith in the democratic process. the abuse of our platform to attempt state sponsored manipulation of elections is a new challenge for us. and one that we are determined to meet. today we intend to show the committee how serious we are about addressing this new threat by explaining the work we are doing to understand what happened and to ensure it does not happen again. at the time of the 2016 election, we observed instances and acted on them of automated and malicious activity. as we learned more about the scope of the broader problem, we resolved to strengthen our systems going forward. elections continue all the time. so our first priority was to do all we could to block and remove malicious activity from interfering with our users' experience. we created dedicated teams
within twitter to enhance the quality of the information our users see, and to block malicious activity wherever and whenever we find it. those teams continue to work every day to ensure twitter remains a safe, open, transparent and positive platform. we have also launched a retrospective review to find russian efforts to influence the 2016 election through automation, coordinated activity, and advertising. while that review is still under way, we have made the decision to share what we know today. in the interest of transparency and out of appreciation for the urgency of this matter, we do so recognizing that our findings may be supplemented as we work with committee staff and other companies, discover more facts and gain a greater understanding of these events. my written testimony details the methodology and current findings of the retrospective review in detail. we studied tweets from the period september 1 to november
15, 2016. during that time we did find automated and coordinated activity of interest. we determined that the number of accounts we could link to russia and that were tweeting election-related content was comparatively small. hundredth of a percent. 1/3 of one percent came from russian linked automated accounts. we did, however, observe instances where russian linked activity was more pronounced. and we have uncovered more accounts linked to the russian-based internet research agency as a result of our review. we also determined that vertising by russia today in seven small accounts was related to the election and violated either the policies that existed at the time or that have since the been implemented. we have banned all of those users as advertisers and we will donate that revenue to academic research into the use of tritter during elections and
for civic engagement. we are making meaningful improvements based on our findings of the last week we announced industry-leading changes to our advertising policies that will hecht our platform from un-- protect our platform from unwarranted products. we're sharpening our tools for stopping malicious activity and increasing transparency to promote public understanding of all of these areas. these improvements will further our efforts to fight both terrorist content and disinformation. we will continue confronting these challenges for as long asthma lishes actors seek to abuse our system -- ma -- as malicious actors seek to abuse our system. we have heard the concerns about russian actors' use of twitter to disrupt the 2016 election and about our commitment to addressing this issue. twitter believes that any activity of that kind, regardless of magnitude, is unacceptable. and we agree we must do better to prevent it. we hope that our appearance
today and the description of the work we have undertaken demonstrates our commitment to working with you, our industry partners, and other stakeholders to ensure that the experience of 2016 never happens again. cooperation to combat this challenge is essential. we cannot at the feet this evolving shared threat alone. as with most technology-based threats, the best approach is to combine information and ideas, to increase our collective knowledge, working within the broader community we will continue to test, to learn, to share and to improve so that our product remains effective and safe. i look forward to answering our questions. >> chairman graham, ranking member whitehouse, distinguished members of the subcommittee. thank you for inviting to us participate in today's hearing. and for your leadership on these challenging and important issues. my name is richard. as director of law enforcement and information security at google, i work with the thousands of people across
teams at google tasked with protecting the security of our network and user data. mr. salgado: previously i had the honor of serving with the computer crime and intellectual property section at the department of justice, focusing on computer networks, crimes, and other crimes such as hacking. google's services provide real benefits to our society. we recognize, however, that our services can be mused. state-sponsored attackers -- can be miss used. state-sponsored attackers are particularly pernicious. they are well resourced. they are sophisticated. they are patient. and often by design, they are difficult to recognize. protecting our platforms from state-sponsored interference is a challenge we began tackling long before the 2016 presidential election. we've dedicated significant resources to help protect our platt forms from such a-- platforms from such attacks by remaining cutting edge defense systems and building security tools directly into our
consumer products. we also have a range of tools to detect and prevent bad acters from engaging in artificially amplifying content on our platform. youtube, for example, uses an array of signals to catch those who try to artificially inflate the view counts their videos or the number of subscribers on their channels. with respect to the 2016 election, we've been looking across our products to understand whether individuals who appear to be connected to government-backed entities were disseminated information in the u.s. for the purpose of interfering with the election. . this was based on the investigatory work of our security teams and on leads provided by other companies. our review included a broad review of all ads from june 15 until the election last november. they were categorized as potentially political by our systems and had even the loosest connection to russia such as a russian i.p. address,
bill address or russian currency. we found two accounts that appeared to be in engaged in activity suspected with government-backed entities. the two companies spent roughly $4,700 in connection with the 2017 elections. our investigation also focused on other platforms. on youtube we found 18 channels with approximately 1,100 videos that were uploaded by individuals who we suspect are associated with this effort and that contained political content. these videos mostly had low view counts, just 3% of them had more than 5,000 views, and instituted only around 43 hours of youtube content. while this is relatively small, people watch over a billion hours of youtube content a day, 400 hours of content are uploaded every minute, we understand any misuse of this platform can be very serious.
the youtube videos were not targeted to any segment of the u.s. population as that's not a feature available in youtube links did observe that were posted to other social media platforms. we believe that the rarely limited activity we found as a result of the safeguards we had in place advance of the election, google doesn't have the viral dissemination these actors seem to prefer but we are committed to continuing to improve our existing security measures to help prevent that kind of abuse. as part of our commitment, we are making our political advertising more transparent, easier for users to understand and even more secure. in 2018 we will release a transparency report for election ads and pair that with the library of election and election ad content that will be accessible to researchers. going forward, users will find the name of any advertiser running an election-related ad
on youtube with one click on the icon above the ad and we will be increasing the safeguards in place to make sure users are in compliant with this covering election ads. on the topics of extremist content, we developed rigorous policies and programs to make sure the use of our platform to spread hate or insight violence more prohibited. we use a mixture of technology and human review to enforce our guidelines and continue to invest in this approach. we are committed to doing our part and recognize that we must work together across government, civil society and the private sector to address these complex issues at their root. we look forward to continuing to work with this committee as it takes on this important issue. thank you for your time and i look forward to your questions. senator graham: thank you. we will do five-minute rounds and obviously stay as long as necessary. so i'll start here. what nations do you worry about
other than russia interfering in our elections? anybody comes at the top of your head there, mr. stretch? mr. stretch: thank you, mr. chairman, for the question. we worry about nation state actors really from around the globe. starting in 2014, we stood up a threat intelligence team that was dedicated primarily to reviewing and monitoring four attacks from threat actors tied to nation states. that work mostly was directed at traditional cybersecurity, account compromise, surveillance, dissemination of stolen information. it's really only recently that we've seen this threat evolve into what we were talking about -- what i was talking about in my testimony, this dissemination of misinformation. in terms of specific countries,
it really is a global threat that we think of it, and we'd certainly be happy to come back to the committee and provide more details on specific actors. senator graham: is that true for the rest of you? mr. edgett: we said in our written testimony see a disproportionate amount of spam or automated accounts coming out of russia but it's agnostic. senator graham: could iran and north korea potentially do his? mr. stretch: certainly potentially. the internet is borderless. senator graham: let's talk about time period. you said you started picking up foreign interference two years ago, is that right, mr. stretch? mr. stretch: we've been tracking threat actors for several years, yes. senator graham: before the 2016 election cycle? mr. stretch: yes, that's correct. senator graham: did you find
activity after the election? mr. stretch: yes, we did. senator graham: what happened after the election? mr. stretch: following the election, the activity we've seen really continued in a sense if you view the activity as a whole, we saw this concerted effort to sew division and discord. in the wake of the election and now president trump's election, we saw a lot of activity directed at fomenting discord about the validity of his election. senator graham: so this continued after his election? dr. stern: it continued until we dis-- mr. stretch: it continued until we disabled the accounts. senator graham: what about you, mr. edgett? mr. edgett: we saw the activity drop off after the election. but these automated accounts continue and so we're continued to focus on making sure they're
removed from our platform. senator graham: mr. salgado? mr. salgado: the same is true for google. the limited use of our platforms certainly decreased once we terminated accounts and we expected that. senator graham: did you see any activity in the primary, mr. stretch? mr. stretch: the activity that we've now attributed to the internet research agency really started in 2015 and was ongoing through the primary, yes. senator graham: were these ads pro-clinton, anti-clinton or could you tell? these activities? mr. stretch: viewing the aggregate, the activity, again, really appears to address a wide range of hot button topics and appears directed at fomenting discord and inflaming discourse. graham in terms of volume, again, what -- how much volume are we talking about?
mr. stretch: about approximately 90% of the volume we saw on the ad side appears to be issues based, primarily a much smaller proportion were particularly at particular candidates. senator graham: in terms of the particular facebook, one in 23,000, i don't know, maybe that was another company. mr. stretch: correct. in terms of total volume of material on the site it's a small percentage. we estimate the internet research content was approximately .004% of the content in news feed during the time period in question. senator graham: to sum this up, and i'll come back with a jihadist in round two, russia as a nation state started interfering in the election cycle back in 2015. and they continued after the election. during the election they were
trying to create discord between americans, most of it directed against clinton. after the election you saw russian-tied groups and organizations trying to undermine president trump's legitimacy, is that what you saw on facebook? mr. stretch: sir, that's an accurate statement. mr. edgett: that's an accurate statement. mr. salgado: not sure i can characterize on our network which way the content went. senator graham: thank you. >> thank you, mr. chairman. so i take it we can all agree that the russians did in fact interfere and meddle in the 2016 elections. your observations on that are consistent with what our intelligence community reports, is that correct? senator whitehouse: mr. stretch? mr. stretch: that's correct. mr. edgett: that's correct. mr. salgado: that's true. senator whitehouse: and i gather that all of your
companies have moved beyond any notion that your job is only to provide a platform and whatever goes across it is not your affair? mr. stretch: senator, our commitment to addressing this problem is unwavering. we take this very seriously and are committed to investigating as necessary to prevent this from happening again, absolutely. mr. edgett: absolutely agree with mr. stretch. this type of activity creates not only a bad user experience but distrust of the of platforms that we are working at trying to get better solving this problem. mr. salgado: that's the same for google. we take this seriously and made changes and continue to get better. senator whitehouse: and ultimately you are american companies and threats to american election security and threats to american peace and order are things that concern you greatly, correct? mr. stretch: that is certainly correct. mr. edgett: i agree. mr. salgado: that's right. senator whitehouse: what i would like to do, and i don't have the time for it here, is to ask you all to answer a
question for the record that i will ask now which is -- give us the key benchmarks of how you have improved at deal with this problem in recent months or recent years? what your goal posts are ahead that you've not yet achieved but are slated or intend to achieve to deal with this problem? that's two. and three would be, what does success look like to you? what can you come to us and say, we have accomplished x and therefore you as a congress don't need to worry about legislating in this space or creating regulations or holding more hearings because we have now got america's back? can you do that for me? mr. stretch: yes. senator whitehouse: you are also corporations that have, i believe, headquarters and significant operations in the state of california. california has a state law
regarding disclosure. presumably you comply with that state law with regard to customers in california. are there lessons and recommendations that you would have for us in evaluating the effectiveness of the california disclosure law? i am given short amount of time i have -- give me just a very brief, do you follow that law and a very brief response to it and then we can flesh out in a question for the record how much of a model that might be for this committee to look at. mr. stretch: senator, we comply certainly with all applicable law. in terms of disclosure going forward, we made an announcement last week that really drew on some of the ideas from the honest ads act which senator klobuchar had introduced. intended to bring ads
transparency really into the political realm, creating a repository of searchable ads, provoiding innovative ways for advertisers to meet their disclosure requirements and requiring documentation and information so that we can ensure that advertisers are not running political ads on facebook in violation of federal election law. senator whitehouse: so let me ask will probably be my last question of this round, anyway, which is that you are all prepared, as i understand it, to undertake to make sure that you can trace content that goes across your platform that qualifies for concern in this area back to a legitimate source so you know if it's a russian who's actually running it, so you know it's an imaginary entity running it. how do you deal with the
problem of a legitimate and lawful but phony american shell corporation, one that calls itself, say, americans for puppies and prosperity has a dropbox as its address and a $50 million check in its bankbook that it is using to spend to manipulate election outcomes? start with mr. edgett this time because we got mr. stretch last time. mr. edgett: i think that's a problem. we are looking into how do you get to know your client. we are also, like mr. stretch said, proud of the work we've done around ads transparency and the ads transparency center we're building and think that center really allows the american citizen to be educated about who is running for an ad, who is paying for the ads, what other ads are they putting out into the world and what they're targeting and believe we will have to figure out a good process to understand who those customers actually are, signing
the contract for twitter to run in. senator whitehouse: let's call it americans for puppies and prosperity, and it's actually a shell corporation. you don't know who's behind it. it could be vladimir putin. it could be a big powerful american special interest. it could be the north koreans or iranians. you need to be able to penetrate the obscurity of the shell corporation, correct? mr. edgett: we're working on the best approach to getting to know the clients and getting to know who's behind the entities that's signing up for it for advertising. senator graham: thank you. senator grassley. senator grassley: thank you, mr. chairman. thanks to both of the senators for leading this discussion, nd thanks to the corporations. the press has reported that russian government placed ads in facebook that were largely aimed at influencing the election. so i want to highlight what i consider inaccuracy in that
reporting. the committee is reviewing the ads that facebook produced. overall the ads do not support a specific candidate, either republican or democrat, and about half of the ads, my staff have reviewed, were placed after the election. the large majority exploits controversial issues in our country and effort to further divide us as a country. for example, some ads target users in ferguson, baltimore and cleveland. these ads spread stories about abuse of black americans by law enforcement. these ads are clearly intended to worsen racial tensions and possibly violence in those cities. it might be true that these ads were intended to influence elections but it's important to be clear that the nature of the ads. russia does not have loyalty to a political party in the united states. their goal is to divide us and discredit our democracy. so a question for you, mr. stretch.
the ads that facebook produced are all about -- all from internet research agency. what is facebook doing to identify ads and contents placed by other bad actors? stroich -- mr. stretch: thank you, senator, for the question. we are focused broadly on addressing questions of authenticity around the content that's placed on our platform and investments we're making around security and around transparency sweep across the entire platform. so for example, the learning that we gain from the 2016 election and from our expansive investigation into it now informs the automated tools that we use to detect and remove fake accounts from anywhere. senator grassley: ok. has facebook produced all the ads and contents it has located from russian sources that were placed prior to the election and if facebook has not, will
facebook focus on producing those ads? mr. stretch: yes, senator. we have produced everything that we've identified that is the product of what we call coordinated inauthentic activity from facebook and we are continuing to investigate and we commit to keep the committee up-to-date on any further progress in our investigation. senator grassley: so that would include all ads placed from russian sources? mr. stretch: all ads from russian sources that -- senator grassley: as well as others? mr. stretch: that are inauthentic and directed at these political issues. there are many advertisements cross-border for legitimate purposes that we have not produced. senator grassley: overall facebook identified more than 3,000 ad purchases worth $100,000 during the 2016 election that had links to russia, russian agents posted messages that reached 126 million users. the ads, accounts and posts that facebook found attempted
to amplify divisive political issues across the political spectrum. twitter identified around 200 accounts linked to russian groups. identified by facebook, russia today or r.t., as sometimes called, spent approximately $27 o0,000 targeting u.s. markets, 2016. russian agents published more than 131,000 messages on twitter. google found tens of thousands of ads by russian accounts that used youtube or google. russian agents uploaded about 1,000 videos on youtube. so questions each of you and a short answer on these two questions i am going to put together. to each company -- starting with mr. stretch -- have you completed internal investigations to identify all accounts, advertisements and posts with connections to russia that purchased ads in the leadup to the 2016
election? and if not, what is the timeline for completion? mr. stretch: senator, as i stated in my testimony, the investigation continues, and we expect to keep the committee up to date on any further discoveries. mr. edgett: the same goes for twitter. we are continuing to work with your staff on both -- our relevant period was september 1 and november 15. we are working with your staff on other investigations. mr. salgado: and for google, the answer is similar. as our investigation continues we will keep the committee up-to-date. senator grassley: and one i would like you to write -- give me answers in writing, can you give us an update on what your internal investigations found? be specific with regard to the number of accounts and total value of the advertisements. y time's up, mr. chairman. >> thanks, mr. chairman. mr. stretch, we know that
russian operatives built misleading pages like black matters u.s. and united muslims of america to attract facebook users. they then exploited a powerful facebook tool called custom audiences to track down those users and send them targeted messages. senator feinstein: can you explain who was targeted using facebook's custom audiences tool? mr. stretch: thank you for the question, senator. as a threshold matter, you're correct that much of the content we've seen is social lly imitative of causes, very meaningful ones to many members of the community, in the facebook community. it's what make i think this content so vial, so upsetting,
so cynical. its attempt to exploit divisions in our society. in terms of the advertising tools that were used to promote these pages that were masquerading essentially, the advertising targeting that was used in the main was a combination of very broad geographic targeting. most of the ads, about 75% of the corpus we've given you was targeted to the united states as a whole and about a quarter of the adds were targeted -- ads were targeted to states. and they were targeted to interest groups. so we have various, what we call like-based or interest-based targeting that was apparently intended to attract people who were following the causes you identified to subscribe to those pages. senator feinstein: thank you. and what have you done with the tools since? mr. stretch: thank you for the question. it's an important one for us
because we do believe these tools are powerful and yet we have a responsibility to make sure they're not used to inflame divisions. so what we're doing is making a number of changes to our ad targeting policies. we're tightening the restrictions on hate speech in ads generally. we're adding additional layers of review where people use potentially sensitive categories for targeting. and we're also limiting the ad content permissions so that here ads are directed at potentially divisive issues, we're trying to tighten our standards to make sure that they're not targeting individuals or communities. senator feinstein: thank you. i appreciate that. mr. salgado, why did google give preferred status to russia today, a russian propaganda arm on youtube? mr. salgado: there was a period of time where russia today
qualified really because of algorithms to participate in an advertising program that opened up some inventory for them. subjective standards around popularity and some other criteria to be able to participate in that program. platforms or publishers like r.t. drop in and out of the program as things change and that is the case for r.t. they dropped out of the program. senator feinstein: why did you sdnt you revert r.t.'s status after the i.c. came out january, 2017? it took you to september of 2017 to do it? mr. salgado: the removal of r.t. from the program was actually the result, as i understand it, is the result of some of the drop in viewership. not as a result of any action otherwise. so there was -- there's nothing out r.t. or its content that
meant that it stayed in or stayed out. senator feinstein: a quick one for twitter. twitter produced images from tweets that contained false voting information. example, telling voters they could vote by sending a text message. all targeting likely clinton voters. just before the election. twitter initially responded to complaints saying twitter had, quote, determined that it was not in violation of our rules. twitter has said there was no obvious russian origin. the posts were removed only after twitter's c.e.o. was directly notified by a twitter user. that's the facts as i understand them. why was this false content allowed to remain in place? mr. edgett: my understanding is once we had user reports of the content we began to remove it as a legal voter suppression. and the interesting thing about the text to vote tweets that we shared with your staff was,
there was a small amount of tweets relative to the size of the platform, but impressions of tweets, calling out those things as fake, were eight times as large. we had 10 times the amount of retweets calling those things out. senator feinstein: impressions -- i don't quite understand? mr. edgett: thank you for your question and so we are all on the same page. impression is the metric to determine whether or not a tweet has been in view on our product. potentially seen by a user. but the interesting thing about the text to vote tweets where we saw the saw the complete counternarrative, the twitter community seizing on them to let everyone know that they were fake. twitter did action those tweets and remove them from the platform. senator feinstein: thank you. thank you, mr. chairman. >> thank you, mr. chairman. i'd like to ask questions about metrics. for twitter, can you walk us to how many accounts there are, how many are anonymous, how many are fake as a subset of
fake or purportedly real? mr. edgett: we have 34u7bd monthly active users. we reported a couple weeks ago. we track things internally like -- senator feinstein: did he say 330 million active users? mr. edgett: that look at things like compressions. we estimate less than 55% of twitter users are false accounts or spam or automated. mr. sasse: less than 5%? can you distinguish between page and automated? because wouldn't there be accounts that real people would run but you could imagine business cycle purposes where you have accounts that are automated but not fake? mr. edgett: great. thank you for letting me distinguish that. we look at whether or not spam or automated whether we think there is a human behind it or not. we can't calculate in some instances whether someone is pretending to be someone they're not. we have a pseudo none muss
policy that allows you to come on and create your own name. it doesn't have to be the one that you have. in that instance we can't trust those metrics. mr. sasse: in political twitter it looks like there are a whole bunch of issues they're following 5,000 and they've got 5,000. they're following 5,000. they have 5,000 followers. it looks like many of these accounts that are the sort of most laughably fake are self-rerchingsal. they're retweeting back and forth at each other. when you evaluate impression numbers, do you have some way of quantifying which numbers are likely inside the universe of the 95% that you think are real users and that's how you pry or tries trying to figure out where -- prioritize trying to figure out where the big problems are? mr. edgett: it's a way malicious actors get their voice act. the instance you discussed where it looks like there's just a lot of retweeting among the circle of followers, some of that can be positive in
natural if you have friends you're following, interested in the same thing. some can be malicious or automated. our tools are getting better each day what is malicious, what is not real. we have a distinction how a real human tweets versus a robot tweets. so we're working on redoubling our efforts on that record. mr. sasse: i want to go to facebook quickly, too. it would seem there is a pretty distinction between verifiablely fake things, text to vote or voter location or voter hours. and things that are narrative-based, where people have competing world views and differing interpretations how some facts and overarching narrative of actual good versus evil and merely political versions that are the subset of those debates. how do you rank order what you should focus on and what is the human capital that you have doing this? so russia and china and potentially north korea, iranian examples are sort of
straightforward in the context the way we've been debating to in the u.s. election. but in the context of potentially jihaddi accounts, there's a whole range of theological interpretation of people who don't quite believe in violence in the name of reoig and people think that's a threshold that the certain theology requires of them. who are your people who do this work? mr. edgett: so we prioritize safety and abuse. it's the number one priority of the company. earlier this year we repivoted all of our engineering and design teams to -- as a subset of things, things like automated accounts being used by malicious actor to amplify their voice. we have hundreds in terms of the entiring engineer organizations, sometimes thousands, we have a company of about 3,800 employees and half are focused on this problem at certain times throughout the life cycle. mr. sasse: and understanding the intory cassies of theology
is something that engineers don't do? who do you hire for that? mr. edgett: we have a very respected trust and safety team who has to research those issues around the world. global platform being used everywhere except for a few places. so we have teams that are researching these issues and trying to distinguish what you're talking about between violent groups and groups that may have some connection to them but are more political arms. we've seen many instances of that. there are teams who has to tee up the nuances and understand how these groups are acting and coordinating at sometimes but there are teams that help us refine and implement new policies around them. mr. sasse: mr. stretch, i'll save some metrics questions for you after the hearing. can you tell us about facebook's human capital solution to the same problem? mr. stretch: yes. thank you, senator, for the question. so today across our safety, security and product and
community operations teams we have about 10,000 people who are working on safety and security generally and we're committing to investing more and doubling that number by the end of 2018. on the question of extremist content generally, i think you raise a really important point which is we need to understand the behavior and we need to have the capacity both as a company and as an industry to be able to track it and eradicate it. so we have thousands of people who as part of their job on a regular basis are attempting to keep terrorism off of facebook. we have 150 people who do nothing else. that's their job, and across that 150 people they have, as mr. edgett suggested, in our case as well, significant expertise in understanding jihaddi threats. they cover about 30 separate languages. one of the things that each of us has done as a company has worked together to make sure that the industry is sharing threat information and sharing
expertise and also providing that information to other smaller companies that may not have the same level of resources. we all agree, not just that terrorism doesn't have a place on facebook, terrorism has no place on the internet and we're trying to lead the industry to make sure that we're all doing our part to address that threat. and the last point i'll make, it also requires an ongoing dialogue with law enforcement, with the government because there's a great wealth of information in the government as it tracks these issues that they can share with us and that in turn gives me some optimism as we address the question of foreign interference in the election. we know how to work together to address a threat on the internet, both as an industry and working with government and i think if we bring that same concerted behavior to bear looking at this threat of foreign interference in the election i think we'll make some progress. senator sasse: thank you. i ran past my time but i'll follow up with your team. >> thanks, mr. chairman. thanks to all on the panel.
last night, 19 major civil rights organizations sent a letter to facebook which explained their, quote, deep concern regarding ads, pages and hateful content on your platform used to divide our country and in particular to promote anti-muslim, anti-black, anti-lgbtq animus. senator durbin: they had a russian facebook account that, quote, not only promoted anti-immigrant messaging online but also managed to organize an in-person anti-refugee rally in twin falls, idaho, in 2016. the letter also detailed a situation where facebook offered its expertise to a bigoted advocacy group by creating a case study testing different video formats. advising how to reach the anti-refugee camp campaign in the swing states in the final weeks of the 2016 election. is it true facebook assisted in an anti-muslim effort?
mr. stretch: thank you, senator, for the question. let me start by saying that the content that we produced to this committee and that was run by these fake accounts masquerading as real authenticated identities is vial. and it's vial for precisely the reason you say. it's particularly exploit ative in so far it was directed at groups that have ever reason to expect us to protect the authenticity of debate on facebook. in terms of what we're doing in response, we are -- we are reviewing and tightening our ad policies and there's two particular changes that we're making. one is, we are -- we are tightening our content guidelines as they apply to ads with respect to violence.
so much of the content that is so disturbing is -- involves threats of violence towards community and that has no place on facebook and it certainly has no place -- senator durbin: regardless of source? mr. stretch: yes, regardless of source. regardless of source, exactly. we want our ad tools to be used for political discourse, certainly, but we are -- we do not want our ad tools to be used to inflame and divide. senator durbin: the point i am trying to get to i read the set of facts to you. the trigger word was a russian facebook account at which point most of us would say -- what is russia doing promoting anti-immigrant, anti-refugee sentiment in the united states? i'll take the word russia out of a facebook account that promotes anti-immigrant, anti-refugee sentiment in the united states? i don't know if you would characterize that as vial. i sure would. these groups would. what i am trying to get to is
this. when we start with the word russian, fake, trools, botz, so forth, we know the starting point is a trigger, something needs to be done. the second thing we know is if it includes a reference to a political candidate or a party. hen it's a category two of electioneering. i'm sure senator klobuchar will get into that. and what you get into this case as vial content. how are you going to sort this out consistent with the basic values of this country when it comes to freedom of expression? mr. stretch: it's a great -- it's a great question. i don't suggest it's easy. we do value personal expression when that is the purpose of your subject there is objectionable beyond objectionable. where we are trying to draw the line is with respect to
advertising content and using our tools to promote messages and -- senator durbin: i will stipulate they are all ads. they are being purchased to affect an outcome of an election or a voter sentiment or to mislead voters. i'd like to ask your questions to address this as well. is twitter what saying? mr. edgett: our ads policy address those things. if there's inflammatory content that some even would find to be upsetting, that's not the type of ad we want to run on twitter. we establish organic tweets where you and i can tweet from their phone or computer from advertising. advertising our tweets that are serving someone who hasn't asked to follow the content, hasn't asked to follow that, so we draw a hard line on making sure advertisements aren't
inflammatory. senator durbin: i certainly commend and endorse that. i agree with senator sasse, when it comes to drawing those lines, it's a challenge for us and we do it for a living and i think it will be a challenge for you as well. mr. salgado, would you like to comment? mr. salgado: i agree it's a real challenge. we have policies to keep our s and high quality and the proposals we made and we'll be implementing around election ad transparency will reflect that as well. senator durbin: thank you, mr. chairman. >> gentlemen, i am very proud that the three companies you're representing here today are american companies. and i think you do enormous good. power kennedy: but your sometimes scares me. mr. stretch, how many
advertisers does facebook have? mr. stretch: we have approximately five million advertisers on a monthly basis, senator. senator kennedy: did china run ads in the last election cycle that tried to impact our election? mr. stretch: not that i'm aware of, senator. senator kennedy: not that you're aware of. did turk minute stan? mr. stretch: no, senator, not that i am aware of. senator kennedy: how about north korea? mr. stretch: i am not aware of other foreign actors running the same sort of -- senator kennedy: how could you be aware? you got five million advertisers, and you're going to tell me that you're able to trace the origin of all of those advertisements? if i want to hire a lawyer, if
i wanted to hire you when you were in private practice -- you have an incredible resume -- and say, let's go through four shell corporations, i want to run some ads and let's go through four shell corporations because i want to hide my identity, you're telling me you have the ability to go -- to trace through all of these corporations and find the true identity of every one of your advertisers, you're not telling me that, are you? mr. stretch: senator, the commitment -- senator kennedy: no. i am just asking about your ability. not commitment. can you do that today? mr. stretch: we are not able to see beyond the activity we see on the platform. the technical signals we get from an account. we do think the technical signals we see can be used to help us eye departmentify inauthentic behavior. senator kennedy: i am trying to get us down from lala land here. the truth of the matter is you have five million advertisers that change every month, every
minute, probably every second. you don't have the ability to know who every one of those advertisers is, do you? today, right now? not your commitment, i'm asking about your ability. mr. stretch: to your question about seeing essentially behind the platform to understand if there are shell corporations, of course the answer is no. we cannot see behind the activity. senator kennedy: let me ask you something else. if i came to you -- i don't mean to just pick on you but i don't have enough time to do all three of you so you two gentlemen get to skate. if i came to you and said, i want to buy an ad that everybody sees on facebook, that's going to be cost prohibited, can we agree on that? mr. stretch: that's likely the case, senator. senator kennedy: so i got to narrow it down. and you can help me narrow it down because that's your
business model. you collect data and lease it out to companies who use that data to sell people products, services and candidates, isn't that basically your business model? mr. stretch: senator, we do provide targeted advertising. we don't share the data of individuals. senator kennedy: do you have a profile on me? mr. stretch: senator, if you're a facebook user, we would permit you to be targeted with an advertisement based on your characteristics and likes along with other people that share -- senator kennedy: let's suppose your c.e.o. came to you -- not you but somebody who could do it in your company. maybe you could. and said, i want to know everything we can find out about senator graham. i want to know all the movies he likes. i want to know the bars he goes to. i want to know who his friends are. i want to know what schools he goes -- went to, you could do that, couldn't you?
[laughter] mr. stretch: it is a very good question. the answer is absolutely not. we have limitations in place on our ability to use personal -- senator kennedy: i am not saying about your rules. you have the ability to do that? don't you? mr. stretch: again, senator, the answer is no. we're not -- we're not -- senator kennedy: you can't put a name to a face to a piece of data? you're tell me that? mr. stretch: so we have designed our systems to prevent exactly that to protect the privacy of our users. senator kennedy: i understand but you can get around that to find that identity, can't you? mr. stretch: no, senator, i cannot. senator kennedy: that's your testimony under oath? mr. stretch: yes, it is. senator kennedy: i am about out of time. i will take one more minute or one more 30 seconds. are you a media -- let me ask google this to be fair.
are you a media company or neutral technology platform? mr. salgado: we are -- the technology platform primarily. senator kennedy: that's what i thought you'd say. you don't think you are one of the largest news -- the largest newspaper in 92 countries? mr. salgado: we are not a newspaper. we are a platform for sharing of information that can include news from sources such as newspapers. senator kennedy: isn't that what newspapers do? mr. salgado: this is a platform for which news can be read from news sources. senator kennedy: i am way over. thank you, mr. chairman. senator klobuchar: thank you very much, mr. chairman. i won't ask what bar you go to. don't be concerned. and thank you to both of you for this important hearing. so i come at this just with the simple idea that our democracy was formed to be self-governing and that means we don't want foreign entities influencing decisions that our citizens make. and they have a right -- right
of freedom to make their own decisions and i think that was interfered with by influenced by russians and also by others when there's no rules in place to tell us where these ads, paid ads are coming from. and so i do appreciate the efforts from these companies but i don't think it's enough. we will have a patchwork of ads from different companies. some won't be doing any. one won't be doing another. and we won't have actual enforcement. management can change and decisions changes so i think it's important we have the same rules of the road for these issue ads as well as candidate ads that we have for tv, radio and print. it's that simple and that is the bill that senator warner and i have and that senator mccain is our co-sponsor and we're very pleased with that. he looks at this as a national security issue.
so my first question is simply, will you support our bill? mr. stretch? mr. stretch: thank you, senator, for the question. so first, we're not waiting for legislation. as i explained earlier -- senator klobuchar: if you could answer that if you support your bill and if not why not? mr. stretch: we have drawn much of what's in the bill to inform our announcement on friday related to ads transparency and disclosure obligations and we stand ready to work with you and your co-sponsors on that legislation going forward. senator klobuchar: mr. edgett. twitter. mr. edgett: the same goes for twitter. mr. salgado: we certainly support the goals of the legislation and would like to work through the nuances to make it work for all of us. senator klobuchar: just to clarify where you are trying to take pocket and make some changes, there won't be an outside enforcer of those
policies, it would be you, is that true? not you personally but your company? ok. can someone answer? mr. stretch: that's correct. senator klobuchar: and also no one has said yet they will include issue ads, right, and that's what we heard from mr. stretch, 90% of the russian id political ads and somehow the radio in deep river falls, minnesota, is able to find what an issue ad is under the guidelines because i hope your company can do that because the way our election works, campaigns are able to see each other's ads and also the journalists and the public are able to see the ads. so that has been a stumbling block so far. is this a problem you won't be able to figure out those rules like tv, broadcast and radio? and i'm just trying to understand that.
mr. edgett: issue ads are 2.0 and something we're trying to operationalize and what rules to put around it and determine what is an issue ad on a platform likeous like one that's related to an election. we are working hard. we don't have anything to announce today but we hope to announce something soon. senator klobuchar: mr. stretch? we have to have clarification on this. it's f.c.c. clarification. encompasses political issues that are the subject of continuing controversy or discussion at the national level, national debt, defense. people see ads all the time on energy. they see ads on immigration policy. on all kinds of things. and they're able to simply -- we see disclaimers who's paying for them and we know where they're running. as brilliant as your companies are, and they are incredibly brilliant and employed so many people, we happen to believe if you make these things public,
which are the goals of your company is to share information, you will have other sets of eyes that are able to look at those ads. so i want you to look at it from that perspective. i had one question, mr. stretch. i appreciated you guys put out there that the 126 million people had access to those russian ads. were those originally paid ads and they got shared and liked and went through the system? how did that work? mr. stretch: thank you, senator. there's really two categories of content. there were ads, about 3,000 of them, that appeared -- senator klobuchar: paid for by -- ruble ads? mr. stretch: many were paid by rubles. senator klobuchar: what about others? mr. stretch: what you are referring to is organic content. the ads would drive follow
ership and others had unpaid content that people who followed the page are eligible to see. and the 126 million number refers to the latter. it refers to the unpaid or organic posts over the approximately two-year period. senator klobuchar: ok. last question because we need to move on, my colleagues. isn't it often the case candidates or issue groups will get organic looking ad, it doesn't have a disclaimer and they boost it, right? they pay? they are really doing a paid ad but on tv you would know it's a paid ad but not on your platform? mr. stretch: it's a great question, senator. everything that's boosted or promoted in that way is designated as sponsored today. that's the case. so it would have shown up as sponsored if it was paid regardless of whether it related to can candidate or an issue. whatary' doing going forward would capture that in the sense we are creating a transparency mechanism for all pages. you go to the page and you can
see every single ad that's running. senator klobuchar: right now you are just doing candidate ads and not issue ads and 90% were russians? mr. stretch: with respect to what a user could see if they navigate to the page, they would see all of them. every single ad from every single advertiser. so it does sweep much more broadly than that. and we are also focused on making sure that in the political space if there are disclosure obligations for the advertiser that relate to issue ads, as you described, we're going to enable that. so the innovative ways we're working on to make sure that candidate can satisfy their obligations we'll cover that category. senator klobuchar: we'll revisit this on a second round. obviously be easier if everyone had the same rules and we would all know. thank you. >> thank you. thanks for holding the hearing. thank you for your testimony. sorry if i'm -- missed the first part of it. with regard to threat
detection, just to find out what's out there, i believe facebook, you have algorithms that do the first run and humans that if it kicks to a set of humans to look at it and if i understand you increase the number of humans who do look at these kind of things. senator flake: can you explain the process that you have? mr. stretch: thank you, senator. it's certainly true that in ensuring the security of the site we rely on systems and people. we have invested and we are investing heavily on both fronts. with respect to the sort of what i'll call the particularized threat actors that are typically associated with nation states, that's a highly manual process so we will, for example, have a threat indicator that we're tracking and if we see activity, that's a highly
intensive manual effort to police and try to understand the activity and essentially fan out from there. the systems are quite effective it entifying more -- call run of the mine abusive behavior. so fake accounts that are springing up quickly with the purpose of spreading spam can be recognized by our systems. usually pretty readily and disabled without that sort of human intervention. senator flake: what's twitter's policy there? how does that differ? same kind of thing, algorithms and then kicks it to humans? mr. edgett: that's right. we're very focused on the behavior of the account. a lot of the signals we can see behind the public facing platform. so we look at activity like mr. stretch said that looks very apparent. not opposite what a natural person would be doing. signing up for a lot of accounts within seconds. liking or retweeting things
within seconds and that's where we can identify a lot of this automated activity and take it off our platform. mr. salgado: the same is true. the desires to move as much as this as we can because of the scale we operate at to machine learning, over time it gets smarter, we get signals. we educate the automated portion of the process so it's faster and higher quality. we always have human review on the other side of it to handle the issues that are novel or where there are some gray area. senator flake: obviously you've undertaken these means without all three of you without government telling you you have to do it. is there base model that says, let's take care of this? is it in your economic interest to take care of this, all of you? starting with facebook. mr. stretch: senator, we believe authenticity is really a cornerstone of what we do and preventing the platform from
being abused -- used for abuse is our responsibility and we're committed to meeting that responsibility. mr. edgett: absolutely the same at twitter. we believe that we shouldn't have automated malicious actors on the platform. it's a bad experience for the user if you ask about the business case. we want to be known as a platform for promoting debate and discussion and having interferences of automated accounts is not something we want on twitter. mr. salgado: yeah, the philosophy for google, if it warrants an ethical and moral imperative, which it is, it's certainly a business imperative. senator flake: thank you, mr. chairman. >> thank you, chairman graham and ranking member whitehouse and i'd like to thank our three witnesses for joining us today. i think what you presented today is truly troubling evidence of the scope and reach of russia's interference in our l.a.x. election and the ways in which americans who typically expect to know when they're consuming a political
advertisement were misled and what are very troubling, slow, halting steps of your otherwise compelling and innovative companies to come forward and to work with us and to help us understand the scope and consequence of this. senator coons: mr. stretch, let me start if i could with a political ad from facebook. this is an ad that was run on facebook in may of 2016. a key moment in the primary campaigns of both hillary clinton and donald trump when both were closing in on the nomination. a group that claimed to be part of texas -- you can see in the upper left -- but was in fact id for by russians in rubles used this ad to target americans based on their professed characteristics like an interest in patriotism or supporting veterans. the ad claims that hillary clinton is, quote, only one politician except barack obama who is despised by the overwhelming majority of american veterans and says if clinton were elected president
the, quote, army should be withdrawn from her control, according to amendments to the constitution. this ad is nothing short of the russian government directly interfering in our elections, lying to american citizens, duping folks who believe they are joining and supporting a group that is about veterans and based in texas when in fact it's paid for in rubles by russians. should facebook be allowed to be a platform that foreign adversaries can use to run political ads, sir? mr. stretch: senator, that advertisement has no place on facebook, and we are committed to preventing that sort of behavior from occurring again on our platform. it's something we take incredibly seriously. i think you're right to surface it. it's upsetting. it makes me angry. it makes everybody who works at the company angry. when i said we are doubling our teams from 10,000 to 20,000 in
order to address safety and security on facebook, that's exactly the sort of thing. senator coons: mr. stretch, thank you for that answer. let me show you another example. there's been a lot of attention on ads but i think we also ought to focus on events. russians also used facebook to make up and promote political events. a group called being patriotic, upper left, shared their event. a miners for trump rally to users in pennsylvania. but, again, this political event was in fact a fraud. organized, funded and supported by russians. russians trying to influence our election duped americans in pennsylvania into coming to an event that was nothing but a fake. help me understand, if i might, mr. stretch, you said these things are vial and upsetting and cynical and that you take responsibility for changing, yet, i am concerned we are now
nearly a year after the election, 10 months after the election, september 6, facebook acknowledged $100,000 worth of ads were linked to russians linked to the kremlin. but if i understand your testimony today, 80,000 posts by the research agency were seen by 29 million americans and may have reached an estimated 126 million people. why has it taken facebook 11 months to come forward and help us understand the scope of this problem, see it clearly for the problem it is and begin to work in a responsible, legislative way to address it when former president obama cautioned your c.e.o. literally nine days after the election last november that this was a big problem and facebook needed to come forward? .
>> one clarification. mr. stretch: when they met and spoke they were talking about acebook generally, not foreign countries but i think to understand what we now see as a systemic and sophisticated effort to interfere in the selection one i want to address. we published a white pain for the april of 2017 that detailed our findings to that point. when the office of the director of national intelligence issued its assessment in january, we weren't sitting around. our threat intelligence team, which i had, which i mentioned earlier, had been looking at what we could learn if the 2016 election and on the basis of that assessment started looking hard at the question of disinformation on facebook and identified a number of practices we thought would be helpful for the industry to be aware of and for the public to be aware of.
we publicly issued that white pain for the april. as you roll the clock forward and we continued our investigation, we did further analysis and we did then discover the ads associated with the internet research agency. at that moment, we brought those advertisements and our learning to congress. we issued a public blog post telling the public what we had found and we committed to continuing our investigation and continuing to commit to share what we learned with congress. i'd like to make one additional point if i may. on that content that you exhibited, what to me is so interesting about that is it reflects the sophistication , in my view of course what we're dealing with. this is not just an online attack. this is an online attack that affects multiple companies, multiple platforms and it's also paired to offline activity.
this is a national security issue and it's one that we are taking very seriously. i know my colleagues here are taking very seriously. we do need to work together to make sure we understand the scope of the threat and we need to continue to work with law enforcement to make sure we're sharing information and expertise to address this thoroughly. senator coons: thank you for your answer. i'm grateful to the chairman for holding this hearing but gentlemen, i wish we had the executives of your three companies before us today and i look forward to hearing in more concrete ways the steps your organizations are taking to address these very real threats to our democracy. thank you. >> for members information, i think there are two votes, one just started, we're going to keep marching on here and take turns voting and senator cruz. senator cruz: thank you, mr. chairman. welcome. i appreciate each of you coming and testifying today.
i recognize that a lot of folks in the media are, and even some members of this committee, are praising your companies for taking active steps to police some of the content on your sites. but i have to note that doing so raises troublesome concerns at the same time. particularly given the percentage of news and political information that americans receive online through social media or through other online avenues. the prospect of silicon valley companies actively censoring the is ch or the news content troubling to anyone who cares about a democratic process with
a robust first amendment. take one example. which is google. in december of 2015 a professor at northwestern university conducted a study analyzing google search results. he searched for the names of all 16 presidential candidates at the time. and discovered that democrats on average had seven favorable search results among google's top 10 and republican candidates had 5.9 positive articles. and indeed, of the major candidates at the time, hillary clinton had five positive search results and only one negative on the first page. donald trump had four positive and three negative on the first page. bernie sanders had nine positive results without a single negative resulten the first page. and a final candidate, the junior senator from texas had a total of zero positive results on the first page.
[laughter] you may well have been citing my colleague from minnesota on that page. >> that is outrageous. if tor cruz: i will say there were frankenfilter that might be popular. the same profess joran a second study and found the vast majority of news outsets that were represented in google searches for -- google searches were left leaning. it's not just google. 2016 it was revealed that facebook was, quote, curating the list of trending news stories on their website. facebook workers were artificially spiking facebook stories including about former i.r.s. official lois lerner, former navy seal chris kyle and
positive stories about conservative politicians. the reports revealed that reports by conservative outlets like "washington examiner" and newsmax that were popular enough to be picked up by facebook's algorithm were excluded until "the new york times" and cnn began covering the same stories. last month, twitter barred marsha blackburn from advertising her campaign launch video because it deemed a line about her efforts to investigate planned parenthood to be inflammatory. the susan b. anthony list had a video advertisement against a political candidate blocked on twitter because it referred to partial birth abortion as being akin to infant cide. -- to infanticide. those are all political positions that people can take in our democratic society. but it is disconcerting if those
political positions become a lens through which the american consumers consume news. so i want to ask each of you, do your sites, mr. edgett and mr. stretch, to be neutral public fora? >> senator, we think of facebook as a platform for all ideas and we have boundaries in the sense that we don't permit certain cat goifers content such as hate speech but within those guidelines we don't in any way discriminate on the basis of viewpoint or ideology. senator cruz: is that a yes or no whether you consider yourself o be a neutral public forum? mr. stretch: we don't think in
terms of neutral, we're trying to provide each user a personalized news feed that will be the content that's most interesting to that user but we do think of ourselves as, again, within the boundaries i described, open to all ideas without regard to viewpoint or ideology. senator cruz: mr. edget? . edgett: spree speech and ex-sat the core of the twitter mission. we do what we can to embrace that other than violent threats, abuse or harassment but we believe allowing the public an open platform as twit serves the community is one that's important to debate and discussion. senator cruz: if i can ask one more question. >> i'm not going to object but i would note that you and i are the only two who have sat through all of this today and i would like to have a chance to ask a question.
of course i'll let the senator finish his question. senator cruz: how to you respond to the public concerns and growing concerns that your companies and other silicon valley companies are putting a thumb on the political debate and shifting it in ways consistent with the views of our employees? mr. stretch: we think of ourselves as a platform for all ideas and we aspire to that. we are acutely aware of the possibility of unconscious bias across a range of issues, not just politics, and we train our employees on that, for that precise reason, we want to make sure that people's own biases are not brought to bear in how e manage the platform. mr. edgett: we are spending a lot of time training employees on organic tweets.
we have strict rules about advertisements, where since we're serving those ads to folk who aren't following the accounts we want to make sure it's a positive experience but even there we're making tough calls and learning from mistakes and revising policies and procedures going forward but our goal and one of our fundamental principles of the company is to remain impartial. senator cruz: thank you. >> mr. chairman, you and i are the only two who sat through all of this i must admit i'm -- with all due respect to all your companies, i hear a lot of johnny come latelies. there's a lot i think you could have done earlier, i suspect departmentdvertising has watched the profits go up and wish they'd spent some of those profits earlier looking at what the content wasism do know,
we have to be very careful not o be censors is but i'll start with you, mr. stretch. an estimated 126 million people were exposed to misinformation posted by russia's information saget on facebook. the vast majority of this was not associated with advertisement but free russian propaganda. that is spread like wildfire. , these, e show you these strongly resemble places you've already linked to russia. at minimum these pages are inflammatory. when i mentioned johnny come lately, these were on today. today.
it's a problem. now can you tell me with certainty none of these pages were created by russian-linked organizations? they're very similar to some we've seen before. mr. stretch: i can tell you with absolute certainty that none of them are linked to the accounts we identified as coordinated inauthentic activity. because we've removed all of that all those accounts from our site. >> pretty similar to some of them? mr. stretch: the core problem with the accounts we identified was a lack of authenticity. so it wasn't so much the content, although to be clear much of that content is offensive and has no place on facebook. but the real problem with what we saw was its lack of
authenticity. the fact that it came from fake accounts masquerading as authentic individuals on facebook. we would have to look at that content to understand if it suffered from the same, or the accounts associated with that content to understand if it was the same sort of activity. >> we have, in virginia we have , a candidate ce calling protect confederate statues. in alabama we have a candidate for senate who called being gay reprehensible and said muslims shouldn't be able to serve in congress. these seem to be similar to what russians did is there indication the russians are doing that now on these two rations? mr. stretch: we have not seen
evidence of that in connection with those two races. i will say that the -- >> you are looking for it? mr. stretch: absolutely. we are focused on addressing this behavior going forward not just in connection with those two races but throughout the country and indeed around the world. each time there's an election we face a challenge and a responsibility to ensure that the platform is not used for abuse and we're investing heavily to make sure we meet that challenge. >> let me ask about that. cebook's fastest growing markets are in the developing world. there, consequences are divisive information can be dire not just election, it's people's lives. for example, facebook is being used today as a breeding ground for hate speech against rohingya refugees in myanmar.
these are especially vulnerable people. they are being violently persecuted. the leadership of that country is not doing a darn thing. even though it includes a nobel peace prize recipient. not doing a darn thing to stop this persecution. cambodia, the authoritarian government is exploiting social media to smear dissidents. hat are you doing? you're increasingly monetizing information from users in the developing world and you have a right to do that. but what are you doing to make undermine ot used to nascent democracies especially and the undermining is not losing votes, it's losing lives? mr. stretch: senator, thank you for the question. it's really an excellent point and a very challenging topic.
as anyone who has followed the news is aware, the tragedy that's unfolding in myanmar is horrifying. we view our platform in that sense as a vehicle for providing greater visibility into what's going on around the world and greater visibility into human rights abuses. we do have an only fwation to make sure it is not misused. senator leahy: we're talking about lives. myanmar is an example of that, cambodia. mr. stretch: i don't disagree we have teams with language competence working with local organizations to understand the particular challenges associated with operating in those res to to make sure we get it -- in those regions to make sure we get it right. we have a role to play but not be used as a tool to, for
example, foment hatred or glorify violence in any way. senator leahy: you can understand our concern. russia's internet research agency set up a paid twitter -- fake twitter account for the republican party, sent out a stream of fake claims including allegations of voter fraud. people know it was. yet it was retweeted by kellyanne conway, donald trump jr. and president trump even though everybody knew it was fake this is what happens. it is frustrating. i don't say that as a democrat, i say that as an american, as one with has visited countries around the world, trying to protect the right to vote and then we see russians coming here. so speaking of vote, i will now go and vote but please admireand, you have -- i what you've been able to do in
reaching people. but off great responsibility. not only can elections be swayed this way by people who don't are not favorable to the united states but people can die. thank you. senator graham: senator blumenthal. senator blumen that'll: thank you to our chairman graham and ranking member whitehouse. we don't need to lecture you on social responseability, i'm convinced you are well aware of it and i'm convinced you understand the perniciously malign effects of abuses such as we've shown you. and you've provided to us. in fact, they are a cancer on ur democracy and they will
metastasize into distortions of our democratic process unless we throttle them. by disclosure. in this realm, the cure for untruth is in fact more truth and disclosure. that's why i've joined the bill that's been offered by senators klobuchar and warner but also am crafting my own bill to provide for even broader disclosure, not limited to the political realm because just as you said the internet is borderless, it is also largely anonymous and it is ill defined in terms of subject matter. o i want to first show you something that i find maybe the most kind of reprehensible sign
of what can go wrong in this ealm and it is from twitter. provided to us. it shows in effect a deliberates mis-- a deliberate misleading of people, you're aware of it, that they can vote, in effect, online. and my question to you is, using , aziz ebrity's image ansari, he's well known to the group likely to believe it, prompted some people to think they voted, avoiding the line, save time, avoid the line, when in fact their votes were completely negated. do you know how many people voted in quotes, in way. thought they voted but in fact were fooled?
r. edgett: we aren't able to quantify that, what we were able to do before we otook those votes and others like them down as voter suppression is there's an outpouring of tweets refuting these tweets as being false and illegal vetter suppression. we saw eight times as many tweets seen on the refuting tweets. we saw ten times as many users retweeting tweets that warned other voters about this. senator blumenthal: but there's no question in your mind that this kind of image is voter suppression. i have 20, 30, 40 of them. there may have been people discounting them but they kept reappearing and you're telling us you have no way of noting how many voters in effect wasted their efforts believing this
false image, correct? mr. edgett: we were focused on removing the content. senator blumen thaul: i'm not sure we're able to -- senator blumenthal: i'm not sure but we'll try. if we still have the image from the heart of texas it's been taken away i'm told. so we can do without the image. let me ask you, mr. stretch, that post or ad, whatever you call it, was in fact targeted to an audience, correct? mr. stretch: each of the post was targeted to an audience. i confess i'm not sure precisely which one you're referring to. senator blumenthal: the one that , it had an e of
image of a soldier, referred to hillary clinton as a 69% disapursuant to the rule rate among veterans. mr. stretch: i'm familiar with it. senator blumenthal: this ad and many others were provided to the committee. i've been through a lot, or most or perhaps all the ads. what struck me is not just the ads, and how misleading they are, and you used the word sophisticated, absolutely right, they are sophisticated in their malignant distortion but they're also targeted. in fact, the information that you provided us indicates this ad was targeted at people in the state of texas, 18 to 65 years old, it had various other
characteristics, interests, independent -- independence or patriotism. that's true of most of the ads you provided us. they have a target which also was extremely sophisticated, correct? mr. stretch: the interest based targeting appeared to reflect an understanding of the type of audience these actors were trying to reach. senator blumenthal: there's a professional activity that helps target ads, correct? mr. stretch: there are certainly many, many companies and individuals who work on targeting digital media, yes. senator blumenthal: do you know who helped the internet research saget in doing this targeting? mr. stretch: senator we're not able to see behind the accounts, all we can see is the activity hat's on our platform.
all we have is the targeted information. senator blumenthal: let me ask you to give us information about how the internet research agency and any other russian sponsored activities were able to sponsor -- target these ads to specific groups, demographic areas and age groups. thank you. senator graham: senator franken. senator franken: thank you ranking member and chair for holding this. we've been looking at the russian attack on our democracy and questioning government officials past and present and campaign people for the trump the ign but now this is --
extent to which the russians exploited your -- exploited your platforms is bringing the question in, maybe this isn't something just the government has to do. this is something you guys have deal with and fix. and you were kind of the canary in the coal mine in 2016 and at was me time russia conducting cyberespionage against american political organizations they deployed this propaganda program on your platforms, in some cases paying for it in rubles. i want to understand why no one seems to have caught on to the russian effort earlier. mr. stretch. how did facebook, which prides itself on being able to process billions of data points and
instantly transform them in the personal connections for its user somehow not make the ads ction that electoral paid for in rubles were coming from russia. -- from russia? those are two data points. how could you not connect those dots? mr. stretch: you mentioned one aspect of the russian threat that was so viz nble 2016 which was the question of account compromise, stealing contents and disseminating them and that is a threat our security team was intensely focused on and we
think effectively addressed, i think in hindsight we should have had a broader lens. there were signals we missed and we are now focused -- senator franken: people are buying ads on your platform with rubles. political ads. you put data points together, that's what i hear the plat frms do, google has all knowledge that man has ever developed. you can't put together rubles with a political ad and go, hmm. these two data points spell out something bad.
mr. stretch: it's something we should have caught but didn't and we'll pay more attention in the future. senator franken: will facebook commit to not accepting political ad, paid for in rubles yuan? the north korean mr. stretch: our requirement is that all advertisers provide documentation that they are authorized to advertise. i understand the currency signal. senator franken: please answer yes or no. i'm asking you a question, just answer yes or no. can you do that? you're sophisticated. you're the chief legal counsel for facebook. please answer yes or no.
mr. stretch: we're not going to permit advertising to permit political advertise big foreign actors. the reason i'm hesitating on foreign currency is it's relatively easy for bad actors to switch currencies, it's a signal but it's not enough. senator franken: then why would anyone use a north korean yuan? why would they go, i'm going to trick facebook and use this currency? mr. stretch: our goal is to make sure we're addressing all forms -- senator franken: my goal is for you to think through this stuff better. can i have a little more time? ok. senator hirono, you're next. senator hirono: since all of you commit to doing better, do each
f you have a mission statement regarding your platforms to prevent the promotion of discord or fomenting of discord? is there a statement that you can tell us that says ok we're committed that this is not going to happen again? mr. stretch, we'll start with you. mr. stretch: so our mission is to give people the power to build community and bring the world closer together. our policies prohibit hate speech of all kinds as well as other forms of bullying and harassment. senator hirono: one thing you said, mr. stretch , in response to another question, you said authenticity is our responsibility. i like that as a mission statement for all three of your companies. authenticity. mr. stretch: at facebook we do require people to use the
service by the name they're known by. we believe that's a very important part of the service facebook offers. it is, to us, the cornerstone of authentic dialogue. we want to make sure people when they come to facebook can trust the content they see and it's our responsibility to make sure we enforce that by policing bad actors. senator hirono: or a simple mission statement like, you can trust our platform. o do your companies have a mission state wment regard to this particular concern? mr. edgett: one of our underlying missions is to unite and inform. the type of activity we're talking about today is intended to divide. so we are working very hard on this. we have a policy around anonanymore toy allow free speech and expression, especially in more difficult part theefs world, to enable
political dissidents or embedded journalists or human rights activists to take on a different persona to speak truth to powerful individuals. so we are always trying to balance that with the ability to make sure that we are not trying or vide through political paid for acts like we're talking about today. nator hirono: i'll let mr. salgado respond. mr. salgado: we put out a public statement, i think it was yesterday, about wanting to do bet for the this area and being committed to do so and set ought specific steps. senator hirono: we are interested in specific steps. mr. stretch you said there are 150 people at facebook focused on the content of what's on your plt form. how many people do you have, mr. edgett at twitter to concentrate
on the content and ferreting out the kind of content that would be deemed unacceptable, divisive -- i realize there are a lot of first amendment issue bus how many people do you have? mr. edgett: we harness the power of ma heen learning and alg rhythms and a large team of people, the safety team, and our user services team, it's hundreds of people. we're at a different scale than facebook and google, fble but we're dedicating a will the of resource to make sure we're looking at user reports act activity on the platform think they is violent or illegal and prioritizing that accordingly. facebook has 150, you say you have hundreds? mr. edgett: we have hundreds. mr. salgado: we have thousands
of people working on them. we also get a good deal of leads on content that we need to review for, whether it's appropriate or not that come from outside the company. senator jairo noah: you have thousands just focused on content? thousands of people dedicated? mr. salgado: we have thousands of people dedicated, google is -- has many different properties within it, but yes the answer we have thousands that look at cob tent that's been reported to us as inappropriate. senator hirono: in view of that, mr. stretch, do you think 150 people is enough people? mr. stretch: to be clear the 150 people i mentioned earlier is people whose full-time job is focused on addressing terrorism content on facebook. in terms of addressing content
on the site generally, we have thousands and indeed we have a community operations team that we announced earlier that we are -- earlier this year we were going to be adding addition thousand -- additional thousands to the several thousands already working on the problem every day. senator hirono: i think it's clear that this is a whole new use or misuse of your platform and you may have various ways to address terrorist content but this is a whole knotter thing. i do have, one more question. this is a short one for mr. stretch because you indicated that there were 126 million people who saw content associated with the internet research agency and that may be just the tip of the iceberg because that was just from one source and there may be plenoy of -- plenty of other dark source out there. in an election with a toal of about 115,000 votes would have changed the outcome, can you say the false and misleading
information people saw on facebook didn't have an impact on the election? can you say it didn't have an impact on the election? mr. stretch: we're not well positioned to judge why any one person or an entire electorate voted as it did. the content that we've provided to the committee was a very small fraction, .004%, of the content served in the united states over the period in question. the point i do want to emphasize is any amount, however small a fraction, this sort of content has no place on facebook and that's why we're investing to address it. senator hirono: i don't think anyone can say this content did not have an effect on the election outcome. >> chairman graham authorized a second round of three minutes if you guys are good to continue then i'll turn to senator klobuchar for three minutes. senator cleb char: thank you very much. i have a followup question
actually about voter suppression efforts. mr. edgett we appreciate the ads we saw and would like to see all of them but some of the ads have been discussed contained misinformation telling voters that they could vote online which of course wasn't true and in fact here's one of them, targeted of course, telling people that they could just text hillary to that number and that's how they vote. i mean you have clearly stated this would be wrong, in fact, illegal, right? mr. edgett: we took these dun as illegal voter suppression. senator klobuchar: i appreciate that you did that but i want people to understand what this is. efforts like this are actually criminal. they're illegal. i'm not talking about your company here, i'm talking about people running these kinds of ads. i was thinking back to new hampshire where some of their people there had actually
engaged in activity where they shut down a line so people couldn't get people to the polls, it was voter suppression. people that did that, one of them went to jail, three people got charged with crimes. so my point here is that these kinds of ads, how serious this is, this is voter suppression. it is illegal to do this it is criminal. i say it so that people understand why we need to have another kind of law in place which is to police this conduct so that we don't have these kinds of things going on that are so very serious. that's why senator mccain and senator warner and i have come together. thank you. mr. salgado i was just a remaining minute here. our intelligence agencies reported that the kremlin has spent nearly $200 million a year spreading propaganda through outlets like r.t. in the leadup to the election,
president putin sought to delegitimize the u.s. electoral system by intensifying critical coverage on rt. it's one of the most popular channels on youtube? mr. salgado: it is a pop lohr hannel on youtube. senator klobuchar: and part of the earnings go to sponsors of those popular youtube channels. is it safe to say they make a significant amount of money in shared ad revenue from you? is google actually paying this kremlin hoff -- kremlin-owned entity? mr. salgado: i think rt is making money on ad revenues. senator klobuchar: is google paying them? mr. salgado: i'm not sure how
the money flow go bus we would be involved in the remuneration to youtube channels. senator klobuchar: that sounds like you're paying them. mr. salgado: i don't know the actual flow. senator klobuchar: we'll follow up in writing. >> i have a question about a couple of ads she we showed, you can tell me how many people obeyed the instruction and texted hilly to 59925, you don't know what other behavior they engaged in and you would know o tweeted clinton with the #between 8:00 a.m. and 6:00 a.m. mr. edgett: we would know the
second one but we wouldn't know if someone picked up the phone and texted that number. >> there's been a bit of atmosphere here that in terms of small dollar numbers it's not a big deal because you have such big platforms. vice news talked with the own over a facebook page who uses facebook ads to use content. after spending $34,100 he controlled pages with 1. million likes. he was able to push out content that could hit some pretty serious numbers. quote work a few advertising dollars, one april video received more than 27 million views and other 450,000 shares. spreading so pervasively into the conservative media yube verse that donald trump's official fatebook page shared it two days later. would you agree that the percentage share of your revenues that an intrusion like this is is not relevant in terms of harm ott public?
mr. stretch: we're trying to provide facts, the information we know about our investigation and we leave it to you and your investigators with viz to believe the a number of different sources to draw the conclusions you will. >> i urge you to stop making the argument that because it's a small number we shouldn't be concerned. >> senator, if i may, we're trying to provide facts. i do not want to suggest in any way we don't think this is a big deal. we think this is a huge deal. >> terrific. last two questions. botnets. good things that are use to feel you or terrible things that are a me nass? in this context? >> i would describe automated fake account creation as a me nass. zwhrilled have to agree. >> me nass.
>> shell corporations that prevent you from looking through them and seing who the true beneficial owner is in this case a help or a me nass? >> anything that prevents us from policing the authenticity of our users is a me nass. >> i would agree. >> i agree with that >> all right, i'm done. thank you, chairman. chairman graham: ok. make it clear are any of you in the content business? do you make your own content? do you generate content? mr. stretch? mr. stretch: the vast, vast, vast majority of the content you see on facebook is user generated. senator graham: what percentage of content do you generate? mr. stretch: a min us kuhl percentage. chairman graham: less than 1%? >> we're not in the content
business. >> we're not in the content business. senator stretch: this is a mr. smith, this is a post brother vided to the committee by mr. smith it's on the islamic state's niger news telefwram messenger channels. i hope they're out of business now. it encourages islamic state supporters to distribute islamic state propaganda on facebook, twitter and google. s to these are your domains, supports of the caliphate. can you give me your -- do you all agree it's bad for business that the american public perceives you as being able to have your platforms hijacked by terrorists to radicalize meshes? that's bad for business, right? >> it's beyond bad for business, senator. there's no place for terrorism on facebook. >> we would agree with that. our technology take downs 95% of terrorist accounts, 75% of them
before their first tweet. >> i agree with that proposition. senator graham: on may 22 in manchester, there's a suicide bomb, the man in question killed 22 people. there's been isis bomb making instructional video on youtube to build an explosive device, you took it down, it came back up. how do you prevent that happening? mr. salgado: there are a few techniques to keep videos from coming back up. one is to of course take action on the account but the more sophisticated way is to generate a digital fingerprint of that video and then block future attempts to upload it. there are some -- there are sometimes ways to evade it but in general it works very well and when it works, it works perfectly. the other is to make sure we have fast flagging processes so that when it comes back up again it somehow evades us, we're notified of it by others or
ourselves quickly and take it down again senator graham: in short order, what have you learned today, mr. stretch? mr. stretch: i've learned the seriousness of this committee and its approach to this topic. senator graham: mr. edgett. mr. edget: we have a lot more work to do. mr. salga d do: this is a project that's going to take the work of all the companies, law nforcement and n.g.o.'s to solve. senator graham: ok we're going to do two more second rounds and then we have to move to the second panel. >> facebook would not commit to not accepting political ads paid for with foreign money. i'd like to ask the same question of you, mr. edgett and you mr. salgado.
twitter and google commit today to stop running electoral ads on american political campaigns paid for by foreign actors? mr. edgett: i don't believe we do. part of the initiative around transparency on who is paying for an ad is to help educate who is behind the advertisement but don't believe we take rubles. senator franken: yes or no, will you not take foreign ads paid for with foreign money? mr. edgett: yes. mr. salgado: i would want to make sure it's a good signal. yes. if it's not a good signal, it's not a good approach. but the intention of it is consistent. senator franken: foreign companies actually can legally do that. mr. salgado: foreign companies
can. the trick is to make sure it's a signal that gives us the right hit. it's a good signal and may be the right one to use. senator franken: foreign owners can't use money in our campaigns you know that, right? it's illegal. so you want to know if it would be a good signal whether to do something illegal or not? mr. salgado: if it's a good enough signal on illegality. there's not going to be -- mr. franken: -- senator franken: propublica reported that up until last month, facebook allowed those using the self-service ad buying to target people who expressed interest in the topic of ewe hater and other anti-semitic team. theerp the categories were created by an algorithm, not a
human being, they were remove bud only after the company was notified about it. i understand that facebook has taken a number of steps to prevent targeting based on self-reported interest bus i question how such categories would be generated and allowed to persist without any human oversight. is it possible facebook didn't know that these categories existed until the media told you? mr. stretch: senator, these cat gores which of course are deeply offensive and alarming were as your question suggested algo rit. ically generated and we don't have reason to believe they were used. the mere possibility that our system permitted them to be generated is ahn acceptable and that's why we not only remove themmed but launched a comprehensive review to make sure we have adequate guardrails in the terms. senator franken: you don't know
how many people saw those ads? mr. stretch: we're not aware of evidence they were used. senator franken: but you don't know they weren't? mr. stretch: we don't know for sure. what we want to do is learn from this and make sure that with respect to the interest categories we permit advertisers to target against obviously are ot so vile and so offensive. senator franken: i know oim out -- i'm out of time but you don't know how much revenue you generated from ad campaigns targetting jew haught irs. mr. stretch: we're not aware of any. senator franken: but you don't know whether you did or not. the answer to my question is, i don't know. mr. stretch: the answer is i'm not aware of any revenue that was generated, we have no reason to believe they were used but i cannot say that would
equivocation. senator graham: senator kennedy. senator kennedy: i'm proud that you are here representing companies that are american companies. and i do believe you do enormous good. but i do find your power breathtaking. i don't believe that you have the ability to determine the identity of all your advertisers. you're good but you're not that good. i don't believe you. and i'm not saying you're not telling the truth i just don't think you can because the change so quickly. i also agree with senator durbin, we've got to be careful here. if we tell you, go forth and don't run any more advertisements that are divisive , well that's going to require you to edit and censor content. that kind of bothers me too.
here's my question on an unrelated subject. though, it is related because none of this would be possible without your ability to target people. if i came to you, let's suppose i'm a liquor advertiser. and i come, let me start with fwoogle, i didn't mean to pick on facebook all the time. i came to google and i said, i want to run some liquor ads can you put together for me a list of everybody who is depressed? could you do that? mr. salgado: i don't think we could do that or anything close to that. senator kennedy: could facebook do that? mr. stretch: absolutely not.
senator kenky: if i said i sell diet pills, could you put together a list of teenagers who think they're overweight? mr. stretch: no. senator kennedy: i'm looking at a report from "the australian" where it reported facebook put together a 23-page memo for an advertiser demonstrating how it could mike are ro target 6.4 million teenagers during moments of emotional vulnerability. when they were stressed and anxious and insecure. was that reporting wrong? mr. stretch: that reporting lied -- relied on a internal document that was overstated and in the wake of that reporting we have put in place safe guards to prevent any sort of ad targeting on that basis. senator kennedy: i'm not accusing you of anything, if i
can have 30 extra seconds, i'm not accusing you of anything. i understand your internal rules, i respect and admire those. but you have the ability, you have the ability to give me a list of people, people who are using facebook or google who are teenagers who are insecure about their weight. now you may not sell that to an advertiser but you have the ability, just like i believe you have the ability to go look specifically at senator graham's or my profile. you may have a policy against that. but i believe you can do it, can you not? mr. stretch: again we have architected our system so that i may not. senator kennedy: but i could if you wanted to. mr. stretch: any user on
facebook will share information related to their own interests and activities. it is precisely because that information is personal and why -- senator kennedy: i'm not asking you what is moral or immoral or what your rules are. i'm saying if you wanted to, is it not the case that you can go to john kennedy's profile and see things about john kennedy as a result of my activity on facebook? you have that ability, do you not? mr. stretch: certainly, any user, including myself could navigate to your profile and ow -- see what you've chosen o share. senator kenky: does google have that ability? r. salgado: if you went to
accounts.google.com. senator kennedy: if i'm president of your company and said here's an order. if you don't do it, you're fired and i'm taking all your stock options. ok. i want you to go find senator graham's account and tell me everything you know about him. i'm not saying you would do it or not do it i'm not getting into ethics, not talking about internal rules. i'm saying you have the ability to do that don't you? mr. salgado: we have the ability to look at a user's account. senator graham: thank you to the panel, we'll have the second panel. thank you all very, very much. [captions copyright national cable satellite corp. 2017]
[captioning performed by the national captioning institute, which is responsible for its caption content and accuracy. visit ncicap.org] senator graham: thank you both for coming. could you raise your right hand. do you swear the testimony you are about to give the subcommittee is the truth, who whole truth and nothing but the truth so help you god? our second panel consists of mr. foxx, from the policy research institute, he's been dealing with child pornography and terrorism issues on the internet for a very long time.
mr. michael smith ii is a terrorism analyst who has dedicated a lot of his adult life to dealing with jihaddist efforts to radicalize on the internet and other issues. mr. watts could you please lead us off. mr. watt: thank you, mr. chairman, members of the committee, for having me here today. a decade ago, al qaeda in iraq littered youtube with violent s and a few years later, twitter became a platform for al-shabab and its tirades. >> we're going to leave this hearing at this point but it continues live on c-span3 and online at c-span.org. live now to the u.s. house where members plan to debate four bills from the agriculture department. votes and speeches at 6:30 p.m. eastern. questions at a later time. for what purpose does the gentleman from arkansas seek recognition? >> mr. speaker, i move the house suspend the rules and pass h.r.