Material on digital platforms be entered into the record. B thank you both for being here today. Without objection. Senator cornyn. Thank you, mr. Chairman. Mr. Dorsey, in recent hearings beforee the Senate Commerce thi committee, you said, i think, that twitter has a policy olicy against this information relating to civic integrity. Gri. President trump and his allies have tweeted hundreds of false al claims about the 2020 election. Trump has falsely claimed victory and alleged widespread voter fraud. So, heres the question. Does misinformation about the results of an election and voter fraud relate to civic integrityr why or why not . Yes, it does. And we label those tweets. When the election has not been t called yet or multiple sources have called it differently. Calld im sorry. I didnt understand that. I di did you say you have been able y to tweet . No. Tweet . We have labeled the tweets that would indicate a different result in the election, called e by multiple sources. In the i see. At what point was that done . The throughout the period. Th up i october 11th up till today. So, when the tweet was initially came in, how long wasn it before you, quote, updated o. It . We didnt update it. Re we put a label on it pointing te the broader conversation. Our goal is to connect people with more information around whats happening with the election. Ckly and that occurred anywhere from five minutes to 30 minutes. Nk but as quickly as as quicklys as we can. Thank you. . Is 30 minutes the maximum time . I dont know. Res we can get you that informationt would you . Im interested in this. Does misinformation about the results of an election and vote fraud relate to civic integrity . Why or why not . Yes, it does. And we also label those tweets that would that would connect indicate whether fraud was happening. Larg again, we connect those to er larger conversations on the we platform. Pro so, we want to provide context here. That is our goal. Providing more context, providing more information. Text goal okay. , wh now a specific question, and i not sure, actually, what the on answer to this should be. But on november 7, President Trump tweeted, and i quote, i qe won this election by a lot, end quote. E. En obviously thats not true. President trump lost the election. The warning label that twitter has applied to the tweet says, and i quote, official sources cs may not have called the race when this was tweeted, end had, quote. Do you believe that label goes far enough to prevent the tweets harms when the tweet is still visible and not accurate . I do because its not just the surface level label. It points to a collection of news articles of information ang conversation that gives you an e expansion on whats happening ei with the election. I guess, you see, my concernt are that these tweets arouse pel people, and it seems to me that the entity that runs this operation ought to have an understanding that when there io a major situation, that the coud tweets can play a unique role in either reassuring or stirring people up to unacceptable levels. On could you comment on that . Misled well, i agree in spirit. Our policy is focused on to misleading information around f the election and the sic procesd to provide greater context, to provide additive information so, people can make decisions around whats happening with the d on p election. In its three phases. Its the runup to the election, thats election day, and also the phase were in right now, postelection. Le our policy is in enforcement are focused on providing more trump information, more context to people in those three phases. Let me give you a specific. On november 7, President Trump tweeted this i won this election by a lot, end quote. The warning label that twitter t has applied to the tweet says, and i quote, official sources may not have called the race when this was tweeted, end c quote. Now, heres the question. Does that label do enough to prevent the tweets harms when the tweet is still visible and is not accurate . I believe its really important we show people a broader context. N thats the intention of the onnt label. It is not just text below a tweet. Its a link to connect to a much larger conversation and news articles across the ion spectrum. Well, give me an example of what would have to happen befor, the situation would warrant a stronger response . Well, we did have stronger responses during election day p and the week after where we didh put you had to click througho to see the content of the tweetd and it limited spread for any anything that went against our civic integrity policy, including premature calls to ol, the to election results. Well, let me give you one more. On november 12th, president prd trump tweeted a Conspiracy Theory that 2. 7 million votes for him were deleted. The warning label that twitter b has applied to that tweet said, quote, this claim about Election Fraud is disputed. Is now, heres the question, and i think it is a tough issue. Do you believe this label does e enough to prevent the tweets harms, when the tweet is still visible, its a highly emotionat situation, but the tweet has nos factual basis . But the tweet has a link to o more information, the more rmatn conversation, and more context that informs the situation cont whats happening. So, i do believe that connecting people to the larger conversation, giving them more e context is the right path here. But they have to move to mov solicit that contact, right . Its not contained as an addendum to the original tweet. S that the label is an addendum to the original tweet. Ou tap if you tap on it or click it, you will go to an expansion of the information. I see. Can i ask a question of mr. Zuckerberg . Yes, you may. Mr. If you want more, that would be okay. Uld mr. Zuckerberg, at the recent hearing before the senate erg, commerce committee, you said ok that facebook has a, quote, policy in place that prevents any candidate, end quote, from, quote, trying to deal legitimize the result of the election, end quote. But the hashtags stealthevote and voterfraud garnered more than 300,000 interactions on your platform in the hours after President Trump declared victory. Heres the question. Do you believe facebook did you enough to prevent trumps enoug efforts to delegitimize the election result . Electi if so, why have you reached thau conclusion . Senator, i believe that we have we have taken some very significant steps in this area. Not just the adding additional context to specific posts and making it so that when people search for different hashtags wt show additional information, but we also took the unprecedented step of putting the Voter Information Center at the top of facebook and instagram for able everyone in the u. S. That showed them reliable information aboutg the election, including par partnering with organizations d like reuters and the national w election pool to show them te in Accurate Information about the h results of the election. So all taken together, i think , that we really went quite far is terms of helping to distribute h reliable and Accurate Information about what was going on during this election. Okay. Cal let me give you one more along this line. More after President Trump falsely claimed that the election was being stolen, a group called stop the steal was started on facebook. It grew to more than 300,000 users in less than a day, making it one of the Fastest Growing groups, i understand, in ry. Facebook history. B you shut the group down, but gel substantial damage already had been done. Been trump supporters, some of them armed with assault weapons, held stop the steal rallies outside Election Offices. In philadelphia, two armed supporters who had traveled froa virginia were arrested on their way to the citys vote counting center. The heres the question. And this is a tough one. The what are your concerns about the spread of misinformation, no matter how innocent it is, or is not innocent, like trumps claims about the election, that they may incite violence . Senator, im very worried about this, especially any misinformation or content that n could incite violence and during such a volatile period like m this, one of our Top Priorities is making sure that people dont use our platform to organize any violence or civil unrest. Wh that was the basis under which g we took that group, because there were a number of members a who were posting potentially violent or encouraging violent n comments that violated our fo ev policies. We also have broader policies in place around trying to slow the spread of misinformation even more broadly, even when its not going to lead to some kind of e violence or imminent harm and ee thats why we have created this independent factchecking program where we work with more than 80 partners around the world to help do factchecking because people in our communityt have told us they dont want to see misinformation, but they also dont want us deciding whad is true and false, so we weve taken the step of buildint this program, which i believe is more sophisticated than what anyone else in our industry hase so, im very focused on these issues. Well, im happy to hear that because im really struck by it, that people armed with assault weapons as the product of a ide tweet could rally outside an it Election Office and i think it really a serious issue that sin needs to be considered and there need to be once you signal , that and people respond to it, it has to be, in some way, abated or in some way pointed out or restructured on the he internet itself. Now, can you respond to that . Senator feinstein, im can y sorry were almost double tre time. Im sorry. Th thats okay. Te were going to have a vote come up. Senator lee. Thank you, mr. Chairman. I would like to note, as far as the president s argument about i the election and how they turned out, inciting violence, i would like to point out here that thee only violence im aware of has occurred in connection with protump peaceful rally attenders. Uite u i dont quite understand that. Maybe well have a chance to bee dwell on that more in a minute. First i want to talk a little rt about federal law. A li existing federal law and what it requires. Ederex section 5 of the federal trade e Commission Act prohibits businesses from engaging in unf unfair or deceptive trade practices. Es compliance with this particulara law requires there be some consistency between what a Company Represents as their practices and their products and what they actually are. In other words, you cant sell one thing and provide another under the guise of providing Something Different than what ia sold. Both twitter and facebook represent and have represented y for years to their users, theirw customers, that they take a i neutral approach to election content moderation. However, as weve heard today and as we will continue to hear today and into the foreseeable s future, there are instances in d which your platforms are takings a very distinctively partisan approach, and not a neutral one, to electionrelated content moderation. For example, just days before the election, twitter suspended the account of mark morgan. Mark morgan is the commissioner of the u. S. Customs and Border Protection office. They suspended commissioner su morgans twitter account specifically for a tweet a celebrating the success of the u. S. Southern border wall. Apparently commissioner morganr tweet, his comments about the ea border wall, violated twitter platform rules governing what ii calls hateful conduct. Now, ive read the offending post, and the offending post from commissioner mark morgan reads as follows at cbp and the army corps of engineers, erg continue to build new wall every day. Every mile helps us stop gang murderers, sexual predators and drugs from entering our countrye its a fact, walls work, period, close quote. Mr. Dorsey, can you tell me in one sentence what exactly is hateful will commissioner morgans tweet that i just read . Well, we evaluated his tweet again and we found we were wrong. That was a mistake and that was due to the fact that we had heightened awareness around government accounts during this time. K to so, there was a mistake, we m reverted it. Thank you. I appreciate that. T we well get back to more of that in a minute and get back to the fact that i understand mistakes happen, but what were going top see today is that mistakes m happen a whole lot more. On almost entirely on one side of f the political aisle rather thann the other. Now, commissioner morgans statement in that tweet, as was initially taken down, is factual. Theres nothing remotely hatefud about it and yet it was taken down. On october 15th facebook relied on a thirdpartys factbased assessment to ban two ban advertisers from facebook for, quote, partially false information. Nformati both were factual. Ealed they revealed joe biden and n Kamala Harris views on lateterm abortion. Joe biden has stated that he wont accept any restrictions on abortion, and senator harris nt views are such that she voted against requiring care for a child born alive during a botched abortion. The very next day the thirdparty factchecker issued a statement retracting the e assessment and retracting it as erroneous. However, it stunningly took facebook almost two more weeks, until october 29th, when votinge had already started in many de jurisdictions, to lift the ban e on these legitimate ads. Ads the factchecker had alreade declared a couple of weeks earlier were erroneously taken down. T mr. Zuckerberg, why on earth di, it take facebook two weeks to correct this error . Senator, im not familiar with the details of us reenabling that ad. I can follow up with you after. Its possible that this was a r just a mistake or delay, and, unfortunately, when we handle ml millions or billions of pieces of content a day, you know, while we strive to do as well ae possible and be as precise as possible, we will make some mistakes. Ill m thank you. K you. I appreciate your acknowledgmen, there are mistakes, as noted previously. Of those mistakes sure happen a tn whole lot more on one side of e the political spectrum than the other. Now, this is understandable. H yr were humans. Its also understandable why this might occur. , maybe some of it has to deal with your employees. 92. 38 of facebook employees who donate to federal candidates se gave to democrats. At twitter its even more starke than that, but 99. 3 of twitter employees who donated to federal candidates gave to democrats. Thm so, these mistakes may be at mistakes but theyre mistakes that rhyme. Hyme. They may not repeat themselves,t but they rhyme. And the consistent theme happenc to be republicans, conservatives and prolife activists. Now, i would like to ask both of you, a list of every user or ist every content creator who has ey been deplatformed or had their contents altered or had some other adverse action taken by either facebook or twitter altered. Alt is there some list that identifies each user for which that has happened . Mr. Zuckerberg, lets hear from you. You. Just a yes or no question. Does such a list exist . Does senator, im not aware of anything like that existing. Mr. Dorsey, how about you . Im not exactly sure what youre asking but we do you have a list of users or content creators who had somk adverse actions taken from them, having their ad altered becauset of the content of their post . Nt certainly whenever we take an action, its recorded somewheree in a database, but im not sure if from thats helpful. What im hearing from both of you is while there may not be an actual list, and i would like ac you both to look to see if such a list exist, but even if theret isnt there is a de facto list. I would like to ask both of you, if such a list exists, please send it to me. Id like to see it. If such a list does not exist, you certainly do have the data necessary in or