The very ne day i want to say that i recognize the challenging position that social Media Companies are in. Your platform host more t the world vws to stay informed, so youre on the frontine of both domestic and foreign disinformation campaigns, and you have to balance american ideals like freedom of speh, ou have to eliminam limit hate you have to limit meaningless information. Thats a significant challee. I saw one of the people w came hereo demonstrate last weekend in washington saying theyre there because they found out that china had, e minute before th polls closed, dumped llions of votes for joe biden. Somebody sa, well, what do you mean . They said, ell, it was on the internet. Its got to be curate. But now your platforms have taken some positive eps. I mentioned that one thing and i heard that people are tang seriously some of this misinformation, sometimes dangerous misinformation, theyve gotten it from your platforms. I often think you can and mus dobetter. Our security, i think even our democracy, the basic truth depend on you doing abetter job. President obama described this erosion of the acceptance of facts and scnce as clear evidence as tooth decay you know, its hard enough to get people to assist and your platforms cant bring people together. I think oftenhey act as a form of driving pple apart. Now, during this election, esident trump has emerged as the most prominent spewer of misinformation. Hes still doing it. He claims the election was rigged,ays the states actually cheated and fixed results, even claims that millions of trump votes were deleted, and hes doing that while his own department of Homeland Security is saying the election was the most secure in american history. And theres no evidee in any Voting System that it was in any way compromised. Thats what our own u. S. Government is saying, when the head of government is saying just the opposite. It may make him fel better about the fact tat he lost dly, but wehouldnt have to put up with it. I have a question for bot of you. Has facebook orwitter conducted an indeptpost momortm review of eltion information at how this information ads . Have you done that kind of postmortem . Senator,e will do that analysis and also ware commissioning and working with independent academics to enale them to do the studies themselvesnd to publish what they find without any intervention or permission required from facebook. Thank you. And we are doin the same, incding opening up our apis to researchers to makeure that others a able to see what we may not see oursels. Will that be made available to us . Can other people see the results of that study youre going to do, both of you, i would ask . Senator, yes. The Academic Research is going to be public, and the academics are going to be able to publish this themselves without even having to get facebooks approval over what they publish. Thank you. Mr. Dorsey . Well make our reports and findings public as well so everyone can learn. I look forward to reading them. Im actually one member of the senate who will actually read them, so thank you. You look at some of the things, i know senator blumenthal wanted me to raise the question of putting them on a video. Think of what it did. It called for the murder, the beheading of dr. Fauci and directly the fbi christopher wray. Think what that does. The fbi director travels with security all the time. Dr. Fauci is has family or private citizens. They were calling for his beheading. It was seen by thousands of people on facebook. If youre going to have somebody threatening to murder somebody, what do you do about that . I was a prosecutor. I prosecuted murderers, and we didnt have to face this kind of threat at that time. But what do you do when hundreds of thousands of people see a rea threat, go murder somebody. In that case, the threat violated our pocies. We took it don. As has been the subject of some of the other questions, if someone had multiple offenses like that, we would remove their whole account. Im sure the threat that if they multiple times, say, go out and murder somebody, cut off their head, were going to face a real problem. Facebook will take down our posting. Oh, my goodness, what a deteent. Senator, what we try to do is identify content that violates our policy fore anyone in the cmunity has to see it or report it to us. For someategories like terrorism, like ie cited before, about 98or 99 of the content that we take down, our ai and Human Systems find it before anyone haso report it to us. On hate speech, were up to 94 of the connt that we take down, our ai systemsnd content reviewers find it before people have to report it to us. What w try to drive on more effectiveness is basically finding more and more of that harmful content earlier before it is seen broadly across our system. Let me ask you about that, because, you know, weve had these discussions before. Im deeply concerned about facebooks role in spreading hate speech in miramar, hate speech that help fuel a genocide against the Muslim Rohinga people. Youve made some progress about this since you and i talked about this last. But my understanding is facebook shuts down specific accounts that violate policy, but then users can just create a new account. In miramar, for example, on october 8, facebook took down 38 inauthentic accounts created and controlled by members of the Miramar Military prior to promote rohingan content. I commend you for doing that. But in the meantime, they turned around and created new accounts that held the same content. So in some ways you have a whackamole problem here. But is there a way that you can stop these things, not just at the account level, at the user level . I use that example because people are being murdered in our systemat genocide. Please answer senator lees question, then well need to move on. Go ahead. Im sorry to take long, but the previous questioner took all his time plus the time allotted to me. No, were at 2 1 2 minutes. Lets just go ahead and answer the question. Senar, youre correctly pointing out that wedid disable certn gerals in the Myanmar Military as dangerous figures, and they are not allowed to sign up for accounts, but as you pointed out, these technical roblems are not ones that there is a Silver Bullet or you can ever fully solve them. We will always be working to help minimize the prevalence of harm in the same way that a city will never eliminate all crime. You try to reduce it and get have it be as little as possible, and thats what we try to do through a combination of building ai systems to identify harmful content up front, hiri thousands of people, tens of thousands of people to do content review, and partnering with organizations, whether its in the intelligece community, law enforcement, Election Officials or in myanmar local Civic Society to help us flag things we should be aware of and on high alert about. Thank you. Mr. Chairman, ill have some questions for the record for both of our witnesses. Thank you very much, senator lee. I appreciate that. Senator cruz. Thank you, mr. Chairman. Febook and twitter and google have massive power. They have a monopolyn Public Discourse inthe onlinearena. It is dismaying to listen to the questions from our democratic colleues, because consistently the message from Senate Democrats is for facebook and twitter and google to censor more, to abuse their power more, to silence voices that Senate Democrats disagree with more. That is very dangerous we want to maintin a free and fair democracy, if we want to maintain free speech. There was a time when decrats embraced and defended the principles of free speech. There was a time when democrats embraced and defended the principlesof the free press. And yet there is an absolute silence from democrats speaki up for the pre otlets censored by big tech. There is absolute slence for the democra speaking out for the citizens silenced by big tech. Instead there is a demand, use even more per to silence dissent, and thats a totalitarian instinct that i think is very dangerous. The same time that big tech exercises massive por, it also enjoys masve corporate welfare. Rough the effect of secton 230, a specia immunity for liability that no one el gets. Congress has given big tech, in effect, a subsidy while they become some of the weahiest corporations on the fac of the planet. Mr. Dorsey, i want to focus primarily on twitter and ask you initially, is twitter a publisher . Is twitter a bub lipublisher . Yes. No, were not. We distribute informion. So what is a publisher . Entity that is publishing under editorial guidelines and desions. Well, your answer happe to be contrary to the text of federal statute, particularly section 230, which definesan information content provider is any person or entity that is responsible in whole or i part for the creation or development of information providedhrough the internet or any other interactive computer service. Let me ask you, was twitter being a publisher when it censored the New York Post . No. We have very clear policies on the conduct we enable to the platform, and if there is a violation, we take enforcement action and People Choose to commit to those policies and to those terms of service. Except your polies were applied asuspectivemanner. You didnt talk about President Trump talking about his tax returns, even though it is a crime to submit your tax returns without consent. You didnt block that, did you . Did you block the discussion of the president s tax material . In the New York Times cas we interpreted it as reporting about the tax material. Did you block edwrd snowden whene illegally released material . I dont have the answerto that. The answer is no. You havent used this in a selective matter let mesk you, were you being a publisher when you forced politico, another journalistic outlet, and when you took down a topic you deemed inadmissible. A New York Post author said, i went to twitter to see if there was an answer. I wish i had given this story a closer read before i answered. Twitter locked it. He got a letter that said, im sorry, your information was blocked. He said, my goal was to raise questions about the story. Oh, my lords in silicon valley, i was attacking the New York Post, you dont understand, i was attacking them as i did in subsequent tweets and see how the administration responded. They responded, and shortly after, he comes back with, my account is no longer suspended, i deleted the tweet. When twitter is editing and censoring and silencing the New York Post, one of the biggest read newspapers in the country, is twitter considering that when they decide what stories to be published or not . No, we realized there was an error in that policy and the enforcement. We corrected that with 24 hours. Imooking at the tweet from twitter that sas, your account has been loked. Youre telling me this is not an accurate thats a lock and can be unlocked when you dete the material. I understand tha you ve the star chaer power. Your awer is always, once we silence you, we canhoose tool lou you to speak. But you are engaged in publishing decisions. Let me shift to a different topic. Mr. Dorsey, does voter fraud exist . I dont know for certain. Are you an exrt in voter fraud . No, im not. Why, they be, is twitter right now putting purported warnings on virtually any statement about voter fraud . Were simply linking to a broader conversation so people have more information. No, youre not, you putup a page that says, quote, voter fraud of any kind is exceedingly rare in the uited stats. Thats not linkingo a broader conrsation, thats taking response as a publisher when you do that. You can take a policy position, but you dont get to pretend youre a publisher and get special benefit under section 230 as a result. That link is pointing to a broader conversation with tweets from publishers and people all around the country. Mr. Dorsey, would the following statement violate twitters policies . Quote, absentee ballots remain the larger source of potential voter fraud. I imagine we would label it so people can have more context. How about this quote . Quote, third party organizations, candidates and political activists, voter fraud is particularly possible where, quote, Third Party Candidates and Political Party activists are involved in handling absentee ballots. Would you flag that as potentially misleading . You doni dont know the specf how we might enforce that, but i would imagine a lot of these would have a label. Youre right, you would label them because youve taken the political position right now that voter fraud doesnt exist. I would note both of those quotes come from the carterbaker commission on election reform. That is former president jimmy carter and james baker, and the question is simply voter fraud does not exist. Are you aware that just two weeks ago in the state of texas, a woman was charged with 104 counts of Election Fraud . Are you aware of that . I am not aware of that. If i tweeted that statement, would you put a warning on it that said voter fraud doesnt exist . I dont think its useful to use hypotheticals, but i dont think so. Were going to test that, because im going to tweet it and well see what you put on it. Yesterday, mr. Dorsey, you and i spent a considerable amount of time on the phone, and you said you wanted to embrace transparency. So i want to ask you, ive asked twitter, ive asked facebook multiple times, how many times have you blocked republican candidates for office, their tweets or their posts in 2016, in 2018 and 2020 . How many times have you blocked Democratic Candidates for office . How many times have you blocked Republican Office holders . How many times have you blocked democratic Office Holders . Twitter has repeatedly refused to answer that question with specific, hard data and cataloging the examples. In the interest of transparency which you said you want to embrace, will you commit in this hearing right now to answer those questions in writing . And well let that be the last question. Im sorry, mr. Dorsey, i didnt hear you. Thats exactly what were pushing for as we think about building upon 230. Is that a yes, that you will answer those questions in writing . Transparency not just of outcomes bu also our process as well. Is that a s, that you will answer those questions in writing . Well certainly look into it and see what we can do. And actually answer them and not give lawyerly double speak about why youre not giving spifics. Will you commit to this committee that you will answer those questions . Were going to work to answer broader transparency around our customer. Thats a no. Mr. Zuckerberg, how about you . Will you commit that facebook will answer those specific questions, cataloging the number of instances in which democrats in 2016, 2018 and 2020 have been silenced in many instances the way republicans have been silenced on facebook . Senator, im not sure if we have that data available, but i wil follow up with you or your team. Ill tak that as a yes. And twitter, well see if the answer is a yes, or transpency is bogus and we dont intendo provide it. We live in a dangerous world. Issues of national security, the orst Pandemic Public Health crisis in modern times in america, and w are being challenged as to whether there is going to be a peacel transition of power in america of the presidency. At that momentin time, we decided none of those topics were important, and what was important was to determine whether or not social media w discriminating against republicans. Ts an interesting question. I think there are more important and timely questions. We have an ection underway in georgia. We have timely obligations from lection officials there, public Election Officials, where they have faced literally death threats. We are trying to determine whether or not the sial media nstruments of america are fair to the republican party. Im tryi to struggle with this issue because i want t put it in a context, and maybe i nt. Maybe this is unique. Wcertainly know what the constitution sayshen it comes to free speech. And we know whatit meant over the years. We certainly didnt suggest at anyone that used a telephone line for nefarious, illegal, banned activity somehow talked the Telephone Company into it by its nature. Then came radio and tv and we had to come up with new rules in terms of equal time, fair content and so forth. And now we have this new, relatively new, mechanism of communicating information, and were trying to determine what to do with it, whether to tre it like a Newspaper Publishing or treat it like some sort of a Communications Network alone. Section 230 sis an attempt to d that, and im sure everyone finds fault wth it. I would like tosk the two witnesses if they would comment on the historical aspects of this particula debate, if they have any thoughts . Mr. Ckerberg . Senator, one of the points in the discussion that i fnd interesting is people ask if the regulatory model should be more li kind of the news industry or re like telco is. But from my perspeive, these platforms are a n industry and suld have a different regulatory model that are distinct from eith of those other two. I thk it is not the case that we are like a telco and that clearly some categories of content, whether its terrorism or child exploitatn, that people expect us to moderate and address, but were also clearly not like a news publisher in that we dont choose the content. We dont choose up front what to create, we give people a choice for content. Perhaps there should be me viability for whats on the platform, but i do not think some of these industries cated previously will ever be fully the right way toook at this. I think it deserves and needs its own regulatry framework to g built here. Thank you. Would the other witness care to respond . Kb from a historical perspective, 230 has created so much innovation, and if we had had that when we started 14 years ago, we could not start. Thats what were most concerned with, is that we continue to enable new companies to contribute to the internet, to contribute to conversation. And we do have to be very careful and thoughtful about changes to 230, because going one direction might box out new competitors and new setups. Going another might create a demand for an impossible amount of resources to handle it, and going yet another might encourage even more blocking of voices or whats being raised here, which is censorship of voicesnd changing the internet dramatically. So i believe that we can build upon 230. I think we can make sure were earning Peoples Trust by encouraging more transparency around content moderation and our ocess of it. I think we need much more straightforward appeals, and i think the biggest pnt to really focus on Going Forward is alrithms and how they are managing andreating these experiences and being ableto have choice in how to use those algorithms on platforms lik ours. Let me get into a specific, mr. Zuckerberg. October 10, Detroit Free Press reported 13en charged thursday in the conspiracy to kidnap michigan governor gretchen whitmer, used apps to add more. In 209, according to an affidavit by brian russell, michigan sergeant, Michigan State police. They encrypted aplatform. Facebooklerted the fbi about the kidnappers Online Activity veral months before the arrest. Thank goodness. However, in august, a Facebook Page for the kenosha guard militia which advocatediolence in the aftermath of the shooting of jacob blake was reported flagged over 55 times to facebook. However, the page was deeed nonviolating and left up. More than 4,000 people respond to that event. Hundreds of med militia showed up. A member of this group, a teenagerrom illinois, late shot and killed two people on the strts of kenosha. Mr. Zuckerberg, you describe facebooks handling of this militia page as an operational mistake. Can you explain the exact reason why the kenosha militia pagewas not taken down . Senator, yes, and first, what happened in kenosha was obviously terrible. What happened here was we rolled out a strengthened policy around militia pages in general. Whereas before that, we would have allowed a group that was a militia as long as it wasnt planning vionce directly. Lding up to the election, we strengthened the policy to disallow more of those groups because we were on high alert and were teaching these situations as highly vulnerable around theelection. We ju put that policy in place. For a number of reasons, itad not yet beenully rolled out, and all of the content reviewers hadnt been fully trained on that. So we made mistakes in assessing whether that group should be tak down. But upon appeal, when it was escalated to a more senior level of content review folks who have more specific expertise in these areas, we recognize tat it did violate the policy and we took it down. Was a miste. It was certainly an iue, and we debriefing and figuring out how we can do better. However, one more piece that i would add is that the person who carried out the shootings was not in any way connected to that page or linked to any of the content there from anything that we or others could tell. Yesterday the fbi released a hate crime report they found that more people were involved in hate crimes since they started collecting hate crime datan 1990. Document and religionbased hate crimes and hispanic hate cmes and hate crimes targetg individuals based on gender identit given these statistics, its cle to me that its more important that social media combat this more than ever. This is not antifa. These are documented haterimes frombi. Usl muslims have rched out to you manytimes, mr. Zuckerbe, about this issue, relating to published content that reflects on certain religious groups. You said in a hearing, you do not alow hate crimes on facebook. Yet in may 2020, the tech transparency proct found more than 100 american white supremast groups, manyof them explicitly antimuslim, acting on the platform of group pages as wl as other content. Facebook altered some ofhe content but the hate groups largely raid. Are you looking the other way, mr. Zuckerberg, at apotentially dangerous situation . No, senator, ts is ncredibly important and we take hatete crimes of violence very seriously. We banned 150 white supremacist organizatio and treat them as fairly as other hate organizations around the world. Weve ramped up our capacity to identify hate speh and incitement to violence bere people even see it on the latforms. Our ai and human review teams, you can track our results in the transparency report that we issue. Now take down aout 94 of the hate speech that we find on our platforms before anyone has to even report it to us, which is a dramatic amount of progress from where we were a few years ago where, when we were just starting to ramp up on this, we were taking about 20 of it down befo people had to report it to us. There is still more progress to make. Were very invested in this, and you have my commitmenthat we view this as an issue of the highest severity and one that we are very focused on. Thank you very much. Senator sasse. Thank you, mr. Chairman, d thank you for sting this hearing. Clearly impoant topics around content moderation. Im a skeptic of the contract moderation polies that exist because i nt think its trance apparent and i dont thinkhe execution is consistent. m skepcal like my colleagues on both sides of the aisle that there is a regulatory fix tt will make it better instead of orse. I also i think its odd that so many in my party are zealous to do this right no when you would have ane coming admistration of the other party that would be writing the rul and regulations aboutit, and i think its telling tt a number of fos on the other side of the dia and i think of senator blumenthal, a guy i like, but who seems almost giddy about the prospect of a new Government Authority toreat online speech, and i think some of us take pause that somebody on the other side of the aisle gets to write these rules and regulations. But to the bader question, first, just to get kind of a level set, and i want to thank both witnessesor being here toy, but when senator lee lays out some of e issues he did abou just every Human Community is going to be situated in a differen place, about policy commitments and priorities and beliefs, but when senator lee said that 93 of facebo employees who contributeo politics doo on the left and , i think it was, of twitter employees contribute on the left, i wou just be interested to see if either of the two of you thinkhat has implications in the shepherdingf your organization again, i recognize fully that youre private organizations, so again, im mor skeptical of a governmental fix for a lot of the problems were lking about here today, but im curious as to whether or not mr. Zuerberg and mr. Dorsey, and i guess well start wth facebook, im curious as to whether you think its likely there is systic bias inside the organization in the execution of morate policies given your Employment Base is so representati of america in general. Senator, think its a good question, and certainly i think it means we have to be more intentional about what we it do, and thoughtful. Our principle and goal is to give everyone a voice a to be a platform for all ideas. As y mentioned, i do think its undisput that our Employee Base, at least the fulltime folks, politically would be somewhat or maybe more thanust a little somewh to the left of where our verall community is, where the community bically spends almost a wide variet of people ross the society. So i think that means we need to be careful and intentnal internally to make sure that bias doesnt seep in to decisions that we make. Although i would point out a couple of things. One is that, you know, people have a lot of different views outside of work, and we expect, and ihink generally see that people conduct themselves professionally. And, second, the folks who are dointhe content review work, we have abt 35,000 people doing content reew, are typically not bsed in silicon valley, theyre based i places all over the country and all over the world, because we serve people in countries all over the world, so i think that the geographic diversity of that is more representative of the community that we serve than just the fulltime Employee Base our headquarters in the bay area. Thanks, mr. Zuckerberg. Mr. Dorsey . You know, this is obviously not something we interview for and even have an understanding of when people are in the company. And wi that underanding, we intend to make su our enrcement and our policy is objective. I realize it looks rather oque, and certainly the outcomes might not alws match up with at intention, with our intention, and the perception of those outcomes may not match up, but thats why i think its so important at were notust transparent around our policies, but the actual operations of our content moderation. If people dont trust our intent, if people are questioning that, thats a failure, and that is something that we need to fix and intend to fix. And i think it would benefit the industry as well. But i do, again, point back to something i said earlier in the testimony, which is a lot of these decisions are not being made by humans anymore, theyre being made by algorithms. Thats around enforcement decisions but also to the things you see and dont see. That is the conversation we should be focused on because that is the enduring use case for everyone in these circumstances. Thank you. I wish it was true that these were all objective, easy questions. A question where if someone said, is the sky green, thats an objective question that the sky is blue and white, not green, but most of the things were talking about here and the places where youre applying content moderation labels are not really simply objective questions, theyre mostly subjective questions. If we talked about medicare for all being easily paid for inside a budget window on assumptions x, y and z that dont raise taxes, thats not true. There isnt any math by which medicare for all pays for itself in some shortterm window, but i dont think any of us really think youre going to slap a label on that saying this is disputed math or policy projections. Really whats happening is there is a policy organization grid that people go through when they build the algorithms, even those not built by humans, and theyre run by policy individuals. I would suspect your Employment Base is not 94 left of center, its probably more than that, and i would speculate that part of the reason less than 1 of your employees give money to candidates on the right is there is a social stigma attached to having conservative views inside your organization, and i would guess those same sort of internal, cultural biases inform the subjectivity of which issues end up labeled. So, again, this is sort of an odd place to be in that i am skeptical that the content moderation policies are thought out well, theyre not transparent enough for us to really know, but im definitely skeptical that theyre consistently applied, and yet im not really on the side of thinking there is some easy governmental fix here. There is a lot about section 230 we could debate. I think some of the things senator durbin said about how, in the era of telephones, nobody blamed the phone company for other people having spread misinformation by the phone. Exactly, thats what would be the case if section 230 were actually neutral. But youre applying content moderation policies and seemingly in a way thats not objective. I know im nearly out of time, but i think it would be useful for us to hear from both of you to give a sort of three or fiveyear window into the future. If there isnt new legislation, what is changing besides just saying were moving from humans to more ai . What kwaqualitatively is changi short of the regulations of a new regulation scheme . Can you te me what problems youre ting to solve . R. Zuckerbergfirst, please. Senator, one thing were trying to improve on is transrency both in the policy and the results. Every quarter we issue a standards report that basically details the prevalence ofeach category of harmful content and how effective we are at addressing it before people he to eveneport it to us. Over te, we would like to fill that out and have more detail on that and make it more robust. Weve already committed to an external audit of those metrics that people can trust them even more. People have all kinds of different requests about where we might go in the future, whether thatsbreaking down the stats by country or languag or into more granular buckets, adding more data around precision. But i thin that that would all be very helpful so people can see and holdus accountable for how were doing. For what its woth, i think that that would be part of a regular together framework tat would not feel particularly overreaching to me, is something that could be put in law that wouldreate an apples to apples framework that all companies in this space would have to report on the outcomes and effectiveness of the programs so that way at least we can s how everyone is doing. That seems like a sensible step to me. Thank you, mr. Dorsey. Sorry, mr. Dorsey, same question and then ill give it back. Its a junior acting chairman. Mr. Dorsey . Ank you. If wereonsidering three to five years out, i think the realization that a centralized global content moderation does not scale, and we need to think how we operate these services. And i would point to we certainly need transparency around any process that we have, and around the practice and the outcomes of those moderations. But i think having more control so that individuals can moderate themselves, you know, pushing the power of moderation to the edges and to our customers and to the individuals using the service is something well see more of, and i also believe having more choice around how algorithms is altering my experience and creating my experience is important. So being able to turn off algorithms, being able to choose diffential gorithms that are found and written by third parties in the marketplace, i think is important and a future that would excite and energize us. Thank you. I appreciated my interaction with both your companies in the runup to this, and ihink both of you said some meaty things therebout ways we can move toward greater transpancy, so ill follow up again. Thank you. Mr. Chairman. Senator whitehouse. Thank you, mr. Chairman. Gentleen, let me start with just a moments history to give some context to my questios. When theobacco industry discovered that i product was deadly, it responded to that news with a systematized program of denyin that set of facts. The shot for the Tobacco Industry was not great. It was found in federalourt to have been engaged in massi fraud and was put under court order to cease its fraudulent behavior. At around the same time, the fossil fuel industry began to have a similar problem rarding its product and it picked up the Tobacco Industrys scheme kind of where it left off, including using some of the same individuals, some of th same sbit tis entities, many of the same methods as the Tobacco Industrys participation. These are persistent, hily motivated, fded operations not unlike a hoile Intelligence Service would n, and they are quite secretive and were now seei a new form, i guessou would call it election nial, happening around our cuntry right now. So thats the background that i com at this with seeing, and i wonder if each of you see a difference in individual error and basically mass disinformation. Is there a difference between odd people with fringe views who offer personal opinio and an orchestrated plan of deliberate misinformationor disinformation that is driven by motivated interests, whether foreign or domestic . Senator, i absolutely think that there is a difference, and you can see it on the platforms. Under our policies and operations, we view these coordinated and authentic behavior operations, networks of fake and sometimes combining with real accounts to push out a message but make it seem like its coming from a different place that it is or might be more popular than it is. This is what we saw the Internet Research agency out of russia do in 2016, and since then a number of other governments and private organizations including some Companies Like what youve mentioned have engaged in this behavior. Now, the good news is that i think that the industry has generally gotten its systems to be a lot more sophisticated to defend against that in the last several years. Its a combination of ai systems that we built to find networks of accounts that arent really behaving quite the way a normal person would, coupled with large numbers of content reviewers, sometimes with expertise in counterterrorism or counterintelligence, and then some signal sharing, whether its with the intelligence community, law enforcement, different groups that have expertise in different areas and with other tech platforms. But this is a big effort on, i think, all of ou sides to make sure that we can defend against this kind of interference, and i think ware Getting Better and better at it. Well, l me encourage you to persist. Ayou know, the last tim you were here, yowere asked abou advertising paid for on fabook denominated in rubles, which was not acheme easy to penetrate, but your organizationas able to penetrate it, a your original setup was simply to allow a slled corporation to act between the real reader and not, so i ask you to continue to make sure real voices are what are heard on facebk. Mr. Dorsey, lete turn to you anask the same question i the context of bots. Brown universit recently did a study that showed about 25 of all tweets about Climate Change are generated by bots. Most of tem obviously push out climate denial as i descrbe that operation. How twitters capity to identify abot as opposed to a real customer . To build off your previous question, i do think there is a difference, as mark said, and i do think there are many coordinated campaigns to manipulate the public conversation, to divide people all around the world, to confuse and generally to distract, and we do have policies enforcement to prevent as much of this as possible. Its a growing threat and it shows no signs of slowing down. One way that entities do this, sometimes it may look like a bot, its actually a human that is organized with other humans for a particular agenda. So it is challenging. We are doing work right now to better identify bots on our service. Let me just interject, mr. Dorsey, real quick. As a baseline proposition, do you agree that a bot does not deserve a voice on your platform, that it should be actual people and organizations . I dont believe that as a highlight, i think we should be labeling bots so people know what theyre interacting with. There are plenty of bots on our service that provide a valuable function and i wouldnt want to take that away. Let me ask both of you, and maybe you can supplement this with an answer in writing for the record, because my time is getting short and this is a complicated question. But the question is, when does it matter to twitter, and when kno who the actual entity is who is using your platform . Well start with you, mr. Dorsey, since mr. Zuckerberg we first last time. U can defer to it in answer if you like, since my time is running short. Well addto this cversation with written answer. But i do believe that pseudonymity is important. We have seen its usefulness with activists. And i think thats critical. But certainly, there are times, and its judged by severity of potential outcomes where we need to dig into identity and take acons. Well follow up with that. And let me just ask u, since my time has expired,. Zuckerberg, to respond or have yourrganization respond in writing. Thank you. Thank you. Before we senator whitehouse brought up something very important. Im going to ask this as directly as i can. To facebook and twitter, do you have anyinternal research or evidence to suggest that your platforms can be addictive . Mr. Zuckerberg. Senator, i think we can followup with a summary of research that we have. But from what ive seen so far, its inclusive, and most of the Research Suggests that the vast majority of people do not perceive or experience these services asddictive or have issues. But i do think that should be controls given to people, to help them manage thr experience there. And its being focused on. Mr. Dorsey. Im not aware of internal research, but we can followup. But i do think, like anything else, these tls can be addictive. And we should be aware of it, acknowledge it a make sure that were making our customers aware of habits of usage. The more information, the better. Thk you. Thank you,r. Chairman. In the late 19. Caller 19thcenturies, the heads of the big corporations, the robrt barons got together, they set rates, got together, determed how to control cord flow and competition. Ill be darned if were not back there again except this time, you the robert barrons, your companies are the most powerful companies in the world. In recent, my ofice was contacted by aformer whistleblower with content of the cpany. I want to start with an internal platform calls tasks that facebook ues to coordinate projects inclueing censorship. Thetask platform allows facebook eloyees to communicate aboutrojects theyre working on together, that begins facebook censor tes including the socalled Community Wellbeing team, the Integrity Team and e hate speech engineering team. Who a use the test platform to discuss which individuals or hash tags or websites to ban. Mr. Zuckerberg, youre familiar with the test platform, arent you . Senator, we use the task system for i think as you say, for people coordinating all kinds of work across the company, although im not sure if id agree with the characterizations specifically around content moderation that you gave. Well, lets get into that. And let me see if we can refresh your memory and provide folks at home watching with an example. Here over my shoulder is an exmple, a screen shot, of the task platform in use. Youll notice that the cameras zoom in, several references to Election Integrity throughout, on these lists of tasks. Again, this is shared across facebook sites, company locations, by working groups. What particularly intrigued me is that the platform reflects censorship input from google and twitter as well. So, facebook, as i understand it, facebook censorship teams communicate with their counterparts at twitter and google. And then enter those companys suggestions for censorship on to the task platform, so that facebook can then follow up with them. And effectively coordinate their censorship efforts. Mr. Zuckerberg, let me jus ask you directly, under oath now, does facebook coordinate its content moderation policies in any way with google or twitter . Senator, let me be clear about this, we we do coordinate and share signals on securityrelated topics. So, for example, if there is a signal around a terrorist attack. Or around Child Exploitation imagery, or around a Foreign Government creating an influence operation, that is an area where the companies do share signals about what they see. But i think its important to be very clear that is distinct from the content and moderation policies that we or the other companies have, where once we share intelligence or signals between the companies, each Company Makes its own assessment of the right way to address and deal with thatinformation. Well, m talking about content moderaon, im talking aboutindividuals, websites, hash tags, phrases to ba is it your testimony that you do not communicate with twitter or google, about content moderation . About individuals, websites, phrases, hash tags to ban . Just yes or no . Do you communicate with twitter or google about coordinating your policies in this way . Senator, we do not coordinate our policies. Do your Facebook Content moderation teams communicate with their counterparts at twitter or google . Senator, im not aware of anything specifi buti think it would be probably pretty normal for people to talk to their peers and colleagues in the industry. It would be normal, and you dont do it . No, im saying that im not aware of any particular conversation. But i would expect at some level of communication probably happens. Ah. But its different from coordinating what our policies are or our responses in specific instances. Ll, fortunately, i derstand that the task platform is searchable. So, will yo provide a list of every mention of google or twitter from the task platform to this committee . Senator, thats something that i can followup with you and your team after on. Yes orno . Im sure you can followup with the list. Why dont you commit while ive got you here under oath. Its so much better to do this under oath. Will you commit now of providing a list to the task platform of every mention of google or twitter . Senator, respectfully, without having looked into this, im not aware of any sensitivity that might exist around that. So, i dont think it would be wise for me to commit to that right now. How many items on the task platform reflect that facebook, twitter and google are sharing information about websites or hash tags or platforms that tey want to suppress . Senator, i doot know. Will you provide a list of every webte and hash tag that Facebook Content moderation teams have discussed, banning on the task platform . Senator, again, i would be happy to follow up with you or your team, to discuss further how we might move forward on that. Will you commit sir senator cruz and senator lee both asks you for lists, website that have been subject to content moderation. You were in doubt whether any exists. But youve acknowledged that the task platform exists, tat it is searchable. So, will you commit to providing the information you have logged on the task website about content moderation that your company has undertaken . Yes or no . Senator, i think it would be better to follow up, once ive had a chance to discuss with my team what a sense of that would be. That might prevent the kind of sharing that youre talking about. But on ive done that, ill be happy to follow up. You wont commit to dit here. We could, of course, subpoena this iormation but id much more want to get it vluntarily rom you. Let everybody take note that mr. Zuckberg has repeatedly refused to provide information that he knows that he has. Let me switch to a different topic, mr. Zuckerberg, tell me absenout sentra. Tell about the tool named sentra. Im not aware of any tool. Theres a demonstrative be iviv shoulder. Sentra tracks different profiles that a user visits. The message recipients. Their linked accounts, the pages they visit around the web that have facebook buttons. Central also uses behavorial data to monitor User Accounts even if accounts are under a different name. You can see a screen shot provided on the sentra platform. We blocked out the individuals name. Although you see their birth date and age when they first started using facebook, last login, as well as any manner of trackings how many different accountsre associated with their name what accounts have they visited. What photos have they tagged on and on and on. Mr. Zuckerberg, how many sites in the United States have been shut down and subject to review by sentra . Senator, iont actually know, becaus im not familiar with that tool. Im sure we have tools but im not familiar with that name. Do you have a tool that does exactly as ive described and that you can see here over my shoulder . Or are you saying that doesnt exist . Senator, im saying that im not familiar with it. And id be happy to follow up. And get you and your team the information you that would like on this. But im limited in what i can what im familiar with and can share today. Almost amazing to me, mr. Chairman, how many people before this committee suddenly develop amnesia, maybe its about something in the air in the room. When a Facebook User accesses a private information like personal preferences or data, is a record made of that, mr. Zuckerberg . Sorry, senator, can you repeat that . Is a record made anytime a facebook employee accesses a Facebook Users messages . Is there a record made anytime a facebook employee does that . Senator, i believe so. Does it trigger an audit . Senator, i think sometimes it may. How many audits have been conducted . Senator, io not know the exact number of you can give me a list . Senator, we can follow up on that, to see what would be used here. Im almost finished, mr. Chairman. Will you commit to giving us a list of how many Times Facebook employees have accessed users personal information without their knowledge, yes or no . Senator, we should follow up on what would be useful here, of course in the operations of the compan if somebody reports sometimes, sometimes, its necessary for people at the company go review and understand the conxt around what is happing when someone reports something. So, this is fairly frequent. And as a matter of course, we do have secury systems that can detect and now most patterns to flag. But we should follow up on more detail onwhat youre interested in. Mr. Chairman, id just sayin closing what we have is clear evidence of coordination betwe twitter, google and facebook. Mr. Zuckerberg knows he has the tools t track this, yet, he doesnt remember or won commit toetting us see it. We have evidence offacebook tracking users acss the web. Mr. Zuckerberg wont answer questions about it, cant remember the me, isnt sure of tools deployed inthis way and wont give us basic information. I submit to you this is totally unacceptable and totally predictable. Because this is exactly what the Tech Companies have done to the American People and congress for years now. Which is why it is time we took action against these modern day barrons. Senator klobuchar. And a you know, mr. Chair, im the chair on the subcommittee. Im goi to take a little different approach when it comes to policy because i derstand why they might be coordinating en itcomes to security. What want to focus on is what i think were seeing all over this country, not just in tech. Were seeing a startup slump. Were seeing more and more coolidation, and throughout history, were seen thats not good for sll businesses. Its not good for consumers, and its not good for capitalism in the end. Even successful companies, even polar companies, and even innovati companies are subject to the antitrust laws of this country. When i asked mr. Pichai about this at the Commerce Committee hearing a few weeksago, he told me google was happy to take feedck. And my response was that the justice departmen already provided feedback in the form of antitrust complaint. And inow there is investigation reportedly going on out ofthe ftc right now regarding your company, mr. Zuckerberg. So, i want to start with excsionary conduct, rerding excluding smaller competitors by limiting intraoperability with the fabook platform. The investigation we saw in the house receny gauge ve us a numbr of Companies Including via and arc. N only damaged the ality of these Small Businesses to compete, but it deprived customers of coenient access. You one the most succsful companies, Biggest Companies in the world, mr. Zuckerberg. Facebook. Do you think this is fair competition or not . With regard to the intraoperability and how youve conducted yourself with these companies . Senator, im generally in favor of intraoperability, and able to access. Thats why we built the facebook platform in 2007. Some of the policies that you mentioned, i think came about, because what we were seeing was not necessarily startups, but larger competitors like google and some of our chinaese arri access the system. At the time that wasnt what we were trying to enable. We may have a nonchinese example here. I just want to know i know that maybe we could hear from mr. Dorsey. And i have concerns about facebooks treatment of twitter subsidiary vine. Its my understanding what facebook recognize the vine as a competitor after twitter acquired it cut off vines ability to operate with facebook so that users couldnt upload their videos to facebook. I think that twitter shut down vine in 2016. Mr. Dorsey, you can tell me about the actual impact of facebooks actions on vines business . On vines ability to compete and your decision to shut down the service . And i know youre not a chinese company. Well, i dont know about the intent on the other side, but i know our own experience was we found it extremely challenging to compete with vine. And ultimaty, it decided that the ball moved past us, and we shut it down. Dont know the specificsand tactics d what was done, but we did find it very, very Challenging Market to enter, even though we existed prior to some of our peers doing the same thing. Okay. Im going to move to Something Else quickly. Instagram and whatsapp and weve had terms that instagram was nascent and if they grow they could be disruptive to us. Later, you write in an email, one of the purposes of facebook to acquire it would be to neutralize a competitor. You wrote those emails referenced in a house report, is that right, mrzuckerberg . Senator, i believe so. Ive always distinguished between two things, thgh. One is that we had some competitn with instagram in the growing space of kind of camera apps and photo sharing apps. But at the time i dont think we or anyonelse viewed instram as a competitor, as a large multipurpose social platform. In fact, people at the time kind of mocked our acquisition, because they thought that we dramatically spent more money than we should have to acquire something that was viewed as primarily as a camera and photosharing app at the time. Heres the view, though, we dont know it would have done. And when we look at your emails, it kind of leads us down this road as well with whatsapp, that part of the purchase of these nascent competitors is to ill use the words of ftc chairman joe simons who just said last week a competitor can squash a competitor by buying it not just by targeting it with anticompetitive activity. So, i know that this is a subject of investigation. Maybe well be hearing something soon. But i think its something that Committee Members better about wear of, not just with facebook. But whats been going on with these deals that have gone through. And how has that led to more and more nsolidation. And how we as t senate, and i just taed to chairman gram about this last week uld actuallyo something about this, by change something of the standards in our laws to make it easier to bring these cases and no just involving tech. So, i want to g to something here at the end. The political ad we had in front of the Commerce Committee. Mr. Zuckerberg, i know you said that facebook had made over 2 billion on political ads over the last few years. You said this was your quote, relatively small part of your revenue. I know that. But its kind of a big part of thelives of politics when tha much money is being spent on ds. This was a bill actually housed with senator graham. Yet, we have seen political a that keep creeping through despte your efforts to police hem on your own. This is why id so badly like to pass the honest ad act. In three battleground states ballots marked for dona trump will be discarded. It stated in three baleground states, paid ad, ballots for donald trump he been discarded. Th pyed between december 29th and october 29th 2020, does this ad violate facebooks policy . Sorry, you can repeat what the ad was . The ad was an america action news ad. Theyve advertised a lot of them on your platform. And it said in three battleground states, ballots marked for donald trump had to have been discarded. This was preelection. Senator, i dont know off the top of my head if that ad violates our policies id be heal to follow up on that. Would you commit to a policy where actuly peoples eyes, people could review these ads, instead of just being hit with algorithm review . Senator, we doave review and verification of political advertisers before they can advertise. Okay. So does every ad go through a human being . Senator, i dont know huh . I think every our policy is that we want to verify the authenticity of anyone whos doing political or social issue advertising. And i think its worth noting that our people reviewers are not in all cases always more accurate than the technical systems. So are you saying people really review every ad, yes or no, or i dont know. Senator, i dont know. Well follow up in the written. You brought this cease and desist order against nyu for publishing a report that facebook is now properly labeled approximately 37 million in political ads. Why would you not support this project . Why would you bring a cease and desist against them . Senator, is that the project that was scraping the data in the way that might havbeen its your definion. With the ftc Consent Decree that we have. The reason its happening is because we havent passed the honest ads act. Theyre trying, theyre n violating privacy. Theyre trying to get the ads so people can see the ads, other campaigns, journalists, everyone. Senator, you know that i support the honest ads act and agree that we should have that passed. And even before that, that weve implemented it across our systems. But i think in the case that youre referring to, that project was scraping data in a way that we agreed in our ftc Consent Decree around privacy that we would not allow. So, we have to follow up on that. And make sure thate take steps totop that violation. Okay. Of the last, mr. Dorsey, do you think the should be more transparency with algorithms . As part of this is n just im off of the ads ow, im on just generically, part this is that pele dont know how this data is gog across the syems and across the platforms. And people basically are buying access, within my impression, so that even if you say, whats the news inhe last 24 hours, old stuff comes up, somethings gone awry from the beginnings of this. Would it be helpful,do you think, if there was more transparency with algorithms . I do think it would be helpful, but its technically very, very challenging to enforce that. I think a better option is providing more choice, to be able to turn off the algorithms or choose a different algorithm, so people can seal how it affects ones experience. Okay, thank you. I ask that both of you look at the bill that senator kennedy and i have, the journalist and conservation act to help the content providers negotiate with digital platforms. Thank. Thank you. Senator tillis. Thank you mr. Chairman, thank yo gentlemen, for joining. Whether cha mr. Chairman, i know youve asked the question a couple times whether or not these platforms can be addictive. I think they probably can b based on what ive read one of two ways, theycould be in personality a engagement of a tool that they can somehow relate to. But i also tnk theres a transactional addiction. A i think you also mentioned social dilemma. I think thats the use of analytics,which i dont critcize among the platforms. But its the use ofnalytics to addict you to go down a certa path to produce a certain outcome. And that could either be an outcome performg an opnion, or an outcome buying something yodidnt even think about 30 minutes before you started going down that path. So, i think there are thin that weve got to look at. I do agree with mr. Zuerberg and mr. Dorsey, its not conclusive, but common sense would tell you its a pblem already and could become a bigger problem. Mr. Zuckerberg, id likeo go ba to the task platform for a minute. When i looked at the screenhot that senator hawley pu up, it look a lot like a Work Management tool. Can you tell me a little bit about that . And how many people are actually engaged as users on that platform at facebook . Senator, yes. Thank you. I was a bit surprised by senator hawleys focus on our task system, because all this is, its a basic internal project management tool. Its exactly what the name sounds like. Its used by companies by people across our company, thousands of times a day torsion assign projects and track them. Its used for all manner of different types of sks. Across diffent people and teams. And d you know, oughly, how many facebook, either contractors or fulltime employees are actually users of the task platform . I think that probably the majority of facebook employees, people that we work with, have some interaction with the task system as part of some part of their work. Its basically just a companywide todo list. The other platform that senator hawley mentioned was the se sentra platform. You said you werent familiar with that one. I think it would be something as a followup to really understand the nature of that platform. I wont press you on it today because you said you werent specifically familiar with the name of the tool. But i would be more interested in how its used. But, mr. Dorsey, does twitter have a platform similar to the task platfm for Work Management communication among staff . Absolutely. Even the smallest compans use these tools. We use a tool called jeera. I was involved in implementing these in my tim in tecology i can see why you have these platforms. Mr. Zuckerberg, you dnt think there was aystemic connection between google and twitter. But you can see how pele in similar profeions may have a discussion, relationship, maybe talk about it over a beer. But could you see how the skeptic could see how these platforms could be used across platforms to fce certain outcomes . Lets say youhad 100 people at facebook, 100 ople at twitter and 100 people at google that all had a political event. They get together. They share notes, and then they go bck and make decisions that could make it appear like its a corporate initiative. But it coulbe an initiative by wellintentioned, but misguided staff. Cow at least see of that as the possible . Senator, i understandthe concern. And i think that coordination, specifically, on writin the policies or enforcement decisions could problematic in the way that youre saying. Which is why i real wanted to make sure it was clear that what we do is share significaals aro potential harms that were seeing. Whether its specific content in the aftermath of that terrorist attack that people are trying to share virally. So that one, if one platform is seeing it, another platform can be prepared. That it will probable see that content son, too. Signals around foreign interference in elections but i think its quite important that each company deals with the signals in a way that is inine with its ownolicy us that in away is very different from saying that companies are sort of coordinating what the policy should be. I underand the policy should be clear on that. With what we do and dont do the. I agree with that. I would find it horribly iesponsible to see that is this a systematic approach across the atforms. But just with the sheer numbers of people that you all employ now, i could see how some of whats been suggested in the hearing could actually occur, with just small groups of people trying to manipulate certin outcom. I dont want to get into details there, except to know, the task platform, if its similar ones that i have experience with, has a lot of logging, has a lot of data, where mybe you could do yourself a service by saying, you knw, i hear whats een suggted here. But in analyzing interactions between groups of people, seeing aberrations and some people more active or geared towas one outco or another. It could actually help aleve some of our concerns with the way the platforms are being manipated. Im not going tohave time to dri down into the specific questions. Im glad youre all open on a regulary outcome. I will tell you, if you listen to my colleagues both sides of the aisle today, i fully expect that congress is going to act in the next congress, that were going to produce an outcome. And some people think that thats not possible because maybe the republicans and democrats are farapart. But if you listen to what theyre askng you, theyre concerned with the kind outcome that they didnt like on social media in equal measure. So, i do believe that you would be well serv to come tohe table as an industry and identi things, mr. Zuckeerg, like what you sa about transparency. And mr. Dorsey, i do thi that the algorithms, when you talk about the sheer scale, are probably the most Sustainable Way togo but were still going to have to have some confidence. I like your conct on choice as well. But were going to have to have more visibili in whats occurred and whats produced certain outcomes. Like a vetans day pt that i did after the election. It was actually after my opponent had conceded. I just posted a picture, thanking veterans. And for a period of time, i think it was suspended and then directing people towards election sults. I would like to think, if that was the result of an algorithmic cision that my oppontho almost certainposted a veterans ad and every other person up for ection got a similar treatmen because if they didnt, it wou seem to me there was some other factor in play if these algorithms are being applied the base. In that case, political commentary from elected officials or candidates. So, i view this hearing as an opportunity to speak your mmitment on two things. One, i mentioned to ou yesterday. Ivgot an intelltual property subcommittee hearing in the middle of december. I would like to have a facebook and twitter representative there. I know that youre very different platforms. But i think you play very prominently in a hearing that senator coons sitting across from me now would like to have you repsented. I think i can speak to senator coons that would be helpful, wed like to get your commitment to have witnesses for that hearing in the middle of cember. Mr. Zuckerberg, can i get that commitment . Senator, yes, we wi make sure that we have the right suect Matter Expert to join your hearing. Thank you. Mr. Dorsey . Well follow up with determining the be course of action. Thank you. And then well be following up on a series of qestions that id like to ask. Let me get my head around some of the analytics information that i think you almost certainly have and hopefully willing to share. But well do that in a collaborative way in my office. Thank you for being here today. Thank you, were going to take a fiveminute break. I think our witnesses have requested a break. And they certainly have earned it. If thats okay with you, senator coons, well come back in about five minutes. Thank you