YT YouTube2Text

YouTube2Text → Transcripts → Christopher Wylie Testimony on Cambridge Analytica U. S. Senate Committee 5 -16 -18

Christopher Wylie Testimony on Cambridge Analytica U. S. Senate Committee 5 -16 -18 — Transcript

by A Sees · 22,656 words · 3,464 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00to everyone for coming the Facebook
0:08matter involving Alexander Colgan and
0:12Cambridge analytic I shed a bright light
0:16on the data practices of some of our
0:19largest technology companies although
0:22advertisers and political campaigns have
0:25collected and used data for years the
0:28public seemed generally unaware this
0:32story has forced both the public and
0:34lawmakers to confront serious issues
0:37that need to be addressed including what
0:41role Congress should play in promoting
0:43transparency for consumers regarding
0:46data collection and use while ensuring a
0:50well-functioning marketplace for our
0:53data dependent technologies to drive
0:57further innovation we started that
1:00conversation with mr. Zuckerberg last
1:03month I hope that today we can continue
1:07a productive and meaningful debate about
1:11these serious policy issues
1:14unfortunately events like these more
1:17often than not seem to get muddled by
1:20partisanship and efforts to score a
1:23quick sound bite the Facebook story
1:27first broke December 2015 because the
1:31Guardian identified that dr. Kogan had
1:35allegedly transferred Facebook data to
1:38gain Cambridge analytic in violation of
1:41Facebook's data policies according to
1:46Cambridge press releases and a recent
1:49internal report in July 2016
1:52Facebook requested Cambridge and its
1:54affiliates to remove any data received
1:57from dr. Cogan Cambridge said that they
2:01removed the data and filed legal
2:04certification to Facebook saying as much
2:08I had requested that Cambridge analytic
2:12appear at this hearing to
2:14explain these facts and tell their side
2:16of the story
2:17Cambridge however recently commenced
2:20insolvency proceedings and therefore
2:23determined it could not participate in
2:25this area the underlying story has not
2:29changed much since 2015 except for two
2:32important events first Cambridge began
2:36doing work for the Trump campaign and
2:39secondly the Trump President Trump won
2:45the 2016 election these two facts
2:49sounded an alarm that revived the
2:54Cambridge story this does not diminish
2:57the importance of this discussion but
3:01only highlights the extreme partisanship
3:03at play and more importantly that this
3:07conversation could have easily taken
3:09place in 2015 in fact this conversation
3:15could have taken place much earlier
3:18advertising agencies and political
3:20campaigns have utilized data analytic
3:24tools for many years campaigns including
3:28those the presidential candidates in
3:30every election year since at least the
3:331990s use data to micro-target during
3:38the past three presidential elections
3:40these strategies have expanded to social
3:43media platforms specifically Facebook
3:47President Obama's campaign developed an
3:50apt utilizing the same Facebook feature
3:53the Cambridge used to capture the
3:55information of not just the apps users
3:58but also millions of their friends
4:01President Obama's apt potentially pulled
4:04even more information than Cambridge his
4:08app a former Obama campaign official
4:11Carol Davison recently wrote quote
4:14Facebook was surprised that we were able
4:17to suck out the whole social graft end
4:21quote in the 2012 election
4:25we also we could also be talking about
4:28more recent events like BuzzFeed's
4:32partnering with multiple Democratic and
4:35anti-trump super PACs in 2016 in a 2016
4:40interview BuzzFeed's vice president of
4:42politics and advocacy said that one of
4:46the problems BuzzFeed was working with
4:49other partners to solve was quote how
4:53are we going to get women who do not
4:56like Hillary Clinton to vote for her end
4:59of quote
5:00that type of order outreach is not
5:03surprising to many that's because it
5:05happens all the time
5:08similarly it shouldn't be surprising
5:10that President Trump's campaign used
5:12consultants to help reach voters as well
5:16regardless of these events and whether
5:19such tactics are actually effective it
5:22is clear that the use of data across the
5:25political spectrum is only increasing
5:27and so instead of just treating this as
5:31a partisan issue to score political
5:34points the important policy discussion
5:37should really have is whether tech
5:41consumer and Congress where we should go
5:44from here our tech companies have access
5:48to some of our most sensitive data are
5:51these companies doing enough to properly
5:54disclose their data policies and protect
5:59their user date many of the services
6:05offered by such tech companies provide
6:08huge benefits to consumers at little to
6:12no cost our consumers blissfully unaware
6:15or are they making informed choices with
6:20respect to how their data is collected
6:22and used in 2015 the US consumer
6:27technology sector directly provided for
6:30and 7/10 million jobs and generated one
6:34a 9/10 trillion in output 435 billion in
6:38late
6:39income and 192 billion in tax payments
6:43how do we ensure the proper amount of
6:46regulation to protect consumers without
6:48damaging an industry that hasn't been
6:51vital to the economy these are the
6:54questions we should be asking I hope
6:56today's hearing will allow us to
6:59continue that discussion senator
7:00Feinstein thanks very much mr. chairman
7:05for holding the hearing in March of this
7:09year a series of articles and videos who
7:12are published online regarding Cambridge
7:15analytic and its efforts to use personal
7:18Facebook data of millions of Americans
7:21to influence the United States elections
7:24to date numerous governments have
7:26launched formal investigations into the
7:29company including the United Kingdom
7:32Australia Canada Nigeria Kenya and India
7:37there is much we do not know about
7:39Cambridge analytical but there are
7:42significant facts already in the public
7:45record we know that Cambridge analytical
7:48was established by Robert and Rebecca
7:50Mercer in 2013 at the urging of former
7:54White House chief strategists Steve
7:57Bannon as an American subsidiary of a
8:00london-based firm SCL group it has been
8:05reported that the intent of creating an
8:08American shell was to give the
8:10appearance of compliance with the United
8:13States election law that prohibits
8:15foreigners from working on United States
8:18elections according to CEO Alexander
8:21Nicks Cambridge analytic I worked for
8:25candidates in 44 United State elections
8:29in 2014 during the 2016 election cycle
8:34mr. Nick stated that Cambridge analytic
8:38and I quote did all the research all the
8:41data all the analytics all the targeting
8:45and quote we ran all of the digital
8:49campaign the television campaign
8:52and our data informed all the strategy
8:55for the Trump campaign in addition
8:59Cambridge analytic obtained detailed
9:02personal information on approximately 87
9:06million people from Facebook without
9:09their knowledge the massive data set
9:12which reportedly included approximately
9:164,000 data points on each individual was
9:20used by Cambridge analytic and SCL at
9:24Canada to develop a comprehensive voter
9:27voter targeting an online behavioural
9:31influence tool called Project rip on
9:35reportedly project rip on was a software
9:38program that used sophisticated
9:41algorithms to allow campaigns to segment
9:44voters into groups based on
9:47psychological characteristics such as
9:50neurotic or introverted once individuals
9:55were identified and grouped the platform
9:58then provided pre-selected and focused
10:01group tested images and keywords that
10:05were most likely to alter the behavior
10:08of those individuals examples of the
10:11messages developed and used by Cambridge
10:14analytical included keywords such as
10:18drain the swamp and deep state as well
10:21as images of border walls in an
10:25undercover video Cambridge analytic of
10:28managing director mark Turnbull
10:31explained that Cambridge analytic also
10:34created the brand defeat crooked Hillary
10:38the company then created hundreds of
10:41different online advertisements for that
10:44brand including online videos that were
10:48viewed 30 million times through Project
10:52rip on selected images were then sent to
10:55the relevant individuals through online
10:58advertising services like Google and
11:01Facebook
11:04these websites provided feedback on an
11:07individual's reactions to those
11:10advertisements which were then fed
11:12automatically back into the targeting
11:15program this is what we've learned in
11:18the past several months however
11:20significant questions remain and there's
11:23much we still do not know about
11:25Cambridge analytical we do not know the
11:29extent to which it worked with hackers
11:32to illegally obtain damaging information
11:36on candidates including the United
11:39States it was reported in 2015 that
11:43Cambridge analytic as parent company
11:46facilitated the hacking and theft of
11:49sensitive medical records
11:51from a Nigerian presidential candidate
11:54and published them online we do not know
11:58whether Cambridge analytic could use
12:00these tactics in the United States but
12:03this pattern of activity was certainly
12:05used by Russian intelligence during the
12:082016 election we do not know the extent
12:13of Cambridge analytics connections to
12:16weaken WikiLeaks and other Russian
12:19interests it has been reported that
12:21Alexander Nicks contacted WikiLeaks in
12:25June of 2016
12:28mr. Nix has said this was his only
12:30contact with WikiLeaks however his
12:34former partner mr. Nigel Oakes has
12:37suggested that Cambridge analytic as
12:40first contact with Julian Assange was
12:44between 12 and 18 months prior it has
12:48also been reported that other employees
12:51at Cambridge analytic had direct
12:54connections to mr. Assange including
12:58through Assange as former attorney in
13:032016 Alexander Nicks also provided white
13:07papers and briefings to executives from
13:11lukoil Russia's second largest oil firm
13:14about Cambridge analytic as
13:17activities in the united states Lukoil
13:21is currently under united states
13:23sanctions related to the Russian
13:26government's activities in the Ukraine
13:29and in March 2017 Lukoil revealed its
13:34formal information sharing partnership
13:36with the Russian Federal Security
13:38Service the FSB the successor to the KGB
13:43finally we still do not know whether the
13:47data obtained by Cambridge analytic I
13:49was ever shared with or obtained by a
13:53third party the data was originally
13:55obtained through a Facebook application
13:58developed by a Russian born professor
14:01named Alexander Cogan professor Cogan
14:05maintains a teaching position at st.
14:08Petersburg University in Russia a
14:10state-funded institution and has
14:13travelled frequently back and forth to
14:16Russia
14:17these are concerning questions not only
14:20for the United States but for all
14:23democracies around the world based on
14:26what we have already learned there is no
14:29question that the future of data privacy
14:31will have a significant impact on every
14:35aspect of our lives including our basic
14:39constitutional rights today we're going
14:42to hear testimony from Christopher Wiley
14:45who served as a research director at
14:48Cambridge analytic ah from June of 2013
14:52until November of 2014 I understand he
14:57will be able to share insight into some
15:00of these significant questions based on
15:03his first-hand experience so I very much
15:06look forward to hearing from mr. Wiley
15:08and I thank you again mr. chairman for
15:11holding this hearing I'm now going to
15:13introduce our witnesses and after I do
15:19that I would ask you to stand and be
15:21sworn also it was our intention senator
15:26Feinstein and I to come down and shake
15:28hands with you
15:29and welcome you I'm sorry that
15:31everything was talked here so we didn't
15:33want to mess everything up but we want
15:36to both of us want to thank you for
15:37participating and particularly you mr.
15:40Wylie coming as far as you have to help
15:44help us with proper testimony our first
15:48witness is dr. Aden Hirsch as dr. Hirsch
15:53is an associate professor at Tufts
15:55University foking focuses in on American
15:59politics he spent seven years as an
16:02assistant professor at Yale University
16:04is a nationally published author he
16:08earned a PhD degree and master's degree
16:11in political science from Harvard
16:12University and a back and a bachelor's
16:15degree from Tufts next person besides
16:22thanking you we welcome you to this
16:24hearing and hopefully you feel
16:26comfortable and everything I know you've
16:29testified elsewhere
16:31Christopher Wylie is a former director
16:34of research at Cambridge and Latika mr.
16:37Wylie worked with President Obama's
16:39campaign for the Canadian Liberal Party
16:41and for the United Kingdom's Liberal
16:44Democratic Party from 2013 to 2014 he
16:48worked for Cambridge and Lathika while
16:50also studying for a PhD and fashioned
16:53trending forecasting he has an
16:55undergraduate degree in law from the
16:58London School of Economics
16:59lastly dr. Mark Jameson is a visiting
17:04scholar at American Enterprise Institute
17:06he's also the director and günther
17:09professor of public utility Research
17:12Center University of Florida's
17:14Warrington's College of Business dr.
17:17Jameson earned a PhD in economics from
17:21Warrington College of Business at the
17:24University of Florida and a master's in
17:26bachelor's degree from Kansas State
17:28University so would you three stand now
17:32please
17:35do you affirm that the testimony you're
17:39about to give before the committee will
17:41be the truth the whole truth and nothing
17:43but the truth so help you God
17:45all have affirmed and so we'll start
17:49with Professor Hirsch and then mr. Wiley
17:52and then dr. Jamison lastly Feinstein
17:56distinguished members of the committee
17:57thank you for inviting me here today
18:00Bob just a minute we may have to do
18:03something with your microphone try talk
18:09again and see if it's a Sammy is it ok
18:20is there anything we hear we have
18:25another person coming with another
18:27microphone we've had this happen before
18:30so it's not it's not you
18:45and kind of pull it as close as you can
18:48go ahead could you okay we have to turn
19:10it on here is what I'm told
19:22Thanks
19:23okay good thank you the controversy
19:28surrounding Cambridge analytic and
19:30Facebook has raised a number of serious
19:32concerns foreign interference in US
19:34elections the personal privacy of
19:36Facebook users and Cambridge analytic
19:38--is voter targeting strategies my
19:41expertise is on civic engagement
19:42personal data and voter targeting and I
19:45hope to be able to answer questions you
19:46have in these areas let me briefly
19:48summarize a few points in my written
19:50testimony every election brings
19:53exaggerated crane claims about the
19:54effects of the latest technologies after
19:57an election there is always a demand to
19:59figure out why the winning campaign won
20:01and the latest technology used by the
20:03winning campaign is often a good
20:05storyline even if it's false campaign
20:07consultants also have a business
20:09interest in appearing to offer special
20:11skills and products and so they often
20:13embellish their role to the media
20:15Cambridge analytic Oh was relatively new
20:17to the scene in 2016 and it promoted new
20:20strategies of psychographic targeting to
20:22persuade voters from everything I've
20:24seen publicly disclosed about Cambridge
20:26analytic a I am skeptical of the idea
20:28that its strategies of voter persuasion
20:30were unusually effective or contributed
20:32meaningfully to the election outcome to
20:35understand why it's useful to divide a
20:37campaign strategy into mobilization and
20:40persuasion mobilization entails finding
20:43supporters and encouraging them to vote
20:45persuasion entails encouraging voters to
20:47support your candidate compared to
20:50mobilization persuasion is very very
20:52hard and Facebook data or data from
20:54other sources doesn't help very much let
20:56me explain why
20:58first persuasion effects decay rapidly
21:01research has shown that an ad may change
21:03a voters mind for a fleeting moment but
21:06the effect goes away amidst all the
21:08other ads news posts and stimuli that
21:10accompany an election season second
21:13persuasion is hard because it is
21:15actually very hard even with top-notch
21:17data to figure out which voters are
21:19persuadable the reason for this is no
21:23person is persuadable all the time
21:25persuade ability is in a stable
21:27disposition you might be persuadable now
21:29but not tomorrow by one kind of message
21:31or messenger but not another third
21:34persuasions hard because when you try to
21:37figure out who is persuadable your
21:39predictions of who is persuadable are
21:41often wrong they're imperfect and you
21:43end up sending targeted messages to the
21:45wrong population for example campaigns
21:49often want to predict the racial
21:50identity of a voter predictions of which
21:53voters are black or Hispanic are wrong
21:54about 25% of the time so when a campaign
21:58sends a message targeted to these voters
22:00a quarter of the people receiving the
22:01message will be missed targeted research
22:04suggests that voters penalize candidates
22:07who missed target them and if campaigns
22:09get race wrong a quarter of the time
22:11they're going to have much much more
22:13error in estimating something nuanced
22:15like a personality how extroverted or
22:17neurotic you are based on your Facebook
22:19lights which is what Cambridge analytic
22:21acclaimed to be doing the idea that
22:24Cambridge analytic oculd use Facebook
22:25Likes to predict personalities then use
22:27those predictions to effectively target
22:28ads strikes me as implausible given what
22:31we know about the stet the challenges of
22:32persuasion and campaigns no evidence has
22:35been produced publicly about the firm's
22:36profiling or targeting to suggest that
22:38its efforts were effective or that it
22:40overcame the difficulties in persuasion
22:42that I just articulated in spite of my
22:45skepticism there is a lot we do not know
22:47about how campaigns are using social
22:49media to target voters this controversy
22:52gives us some anxiety in part because we
22:54don't really know where the line is
22:56between ads attempting to persuade
22:58voters and ads attempting to manipulate
23:00or deceive voters this anxiety is all
23:04the more understandable because Facebook
23:06hasn't really taken seriously its solemn
23:08civic role as a facilitator of news and
23:11political communication
23:12in my written testimony I described an
23:15initiative currently underway by which
23:17independent researchers can measure the
23:19effects of facebook ad targeting and
23:20news sharing the success of this program
23:23depends on a serious commitment by
23:25Facebook to share its data even and
23:27especially in cases that will bring
23:29negative press to the company in raising
23:32skepticism about Cambridge analytics I
23:34do not at all mean to suggest that there
23:36aren't serious concerns stemming from
23:37this controversy related to privacy
23:39foreign interference and a troubling
23:42role of social media in transmitting
23:44news information and fake news and fake
23:47information rather I hope we can focus
23:50on the more important of these issues as
23:51I discuss in my written testimony thank
23:54you again for your attention to the
23:55subject matter and I welcome your
23:56questions talk all right how's that good
24:09mr. chairman senators thank you for the
24:12opportunity to invite me to speak today
24:15Cambridge analytics is the canary in the
24:18coal mine we must address the digital
24:20echo chambers that are being exploited
24:22to algorithmically segregate American
24:25society online communities should unite
24:28us and not divide us data is the new
24:31electricity of our date of our digital
24:33economy and just like electricity we
24:36cannot escape data online platforms
24:39terms and conditions present users with
24:41a false choice because using the
24:43Internet is no longer a choice Americans
24:46cannot opt out of the 21st century all
24:49revolutions throw up new power
24:51structures the American Revolution
24:53required the Constitution to ensure that
24:55citizens of the young Republic were
24:57protected from the excesses of arbitrary
24:59government the Industrial Revolution
25:01required protections for workers against
25:03hazardous conditions in the workplace
25:05and in the environment so too with the
25:08digital revolution unless we realize
25:10that there are there is a new game being
25:12played and that the rights to life
25:14liberty and the pursuit of happiness
25:16need to be defended from those who might
25:19undermine them using new technologies
25:20whether they are corporations whether
25:23they are Nations or whether they're
25:24non-state actors
25:25I've come here today voluntarily as a
25:28witness and as a whistleblower I've
25:31already reported these matters to the
25:33authorities and it should be made clear
25:34that I am NOT a target of these
25:36investigations I was the director of
25:38research at SCL and Cambridge analytic
25:41Oh from mid 2013 to late 2014
25:44SDL group was a British military
25:46contractor specializing in information
25:49operations Cambridge analytic a create
25:52was created to allow SCL to work in the
25:54United States
25:55Cambridge analytic I did not have any
25:57employed staff and all of its clients
25:59were handled by SCL although lawyers
26:02warned about using foreign citizens the
26:04firm installed a non-us citizen as its
26:07CEO and embedded non-us citizens in
26:10American campaigns Cambridge analytic I
26:13offered voter disengagement services in
26:15the United States and there are internal
26:17documents that I have seen that make
26:19reference to this tactic my
26:21understanding was that this was targeted
26:23at African American voters when I was at
26:26Cambridge analytic app I was also made
26:28aware of the firm's
26:29black ops capacity I have seen documents
26:32relating to instances where the firm
26:34sought to procure hacked material some
26:37of these targets of some of the targets
26:39of these special intelligence services
26:41are currently heads of state in various
26:43countries I've also seen internal
26:45documents that make reference to the use
26:47of specialised and sir intelligence
26:49services from former members of Israeli
26:52and Russian state security services a
26:55further concern is Cambridge analytic
26:58--is employment of people closely
26:59associated with WikiLeaks and Julian
27:01Assange in addition to making approaches
27:04to WikiLeaks the firm hired two senior
27:06staff both of whom were previously aides
27:08to the British lawyer representing
27:09Julian Assange Cambridge analytic has
27:12sought to identify mental
27:13vulnerabilities in voters and worked to
27:15exploit them by targeting information
27:18designed to activate some of the worst
27:20characteristics in people such as
27:22neuroticism paranoia and racial biases
27:24to be clear the work of Cambridge
27:27analytic is not equivalent to addition
27:29to traditional marketing Cambridge
27:31analytic has specialized in
27:32disinformation spreading rumours comprar
27:35Matt and propaganda
27:36for those who claim that profiling does
27:39not work this contradicts copious
27:40amounts of peer-reviewed literature in
27:42top scientific journals even Facebook
27:44applied for a patent on quote
27:46determining user personality
27:48characteristics from social networking
27:49systems dr. Alexander Cogan was selected
27:52by Cambridge analytic to lead the data
27:54harvesting operation and over 80 million
27:57data subjects had their personal data at
27:59misappropriated given this scale this
28:01could be one of the largest breaches of
28:03Facebook's data Cambridge analytic akan
28:05tractors had also worked on pro-russian
28:08political operations in Eastern Europe
28:09including with suspected Russian
28:11intelligence agents Cambridge analytic
28:14has set up focus groups and polling on
28:16Americans views on the leadership of
28:18Vladimir Putin and Russian expansionism
28:20in Eastern Europe dr. Kogan was also
28:24working on Russian state funded research
28:26projects the Russian project at st.
28:28Petersburg was also building algorithms
28:30using Facebook data for psychological
28:32profiling and research social media
28:34trolling in fact Cambridge analytic a
28:37pitched quote the interesting work Alex
28:39Cogan has been doing for the Russians
28:41end quote to his other clients Cambridge
28:44analytical also was in close contact
28:45with senior executives at lukoil one of
28:48Russia's largest oil companies Cambridge
28:49analytic a presented lukoil with
28:51documents outlining his experience with
28:53foreign gist information rumor campaigns
28:55and it's American data assets Facebook
28:58knew about Cambridge analytic his
29:00schemes since 2015 before the story
29:03broke Facebook threatened to sue The
29:05Guardian and then banned me for
29:06whistleblowing responding on behalf of
29:09the British government the UK Secretary
29:11of State for culture called my ban
29:13outrageous because it reveals the
29:15unrestrained power technology companies
29:17can exercise over ordinary citizens when
29:20a person's entire online presence can be
29:22so quickly and so thoroughly eliminated
29:24from existence there is no check on this
29:27power and it raises a series question
29:29for Republicans and Democrats alike what
29:31happens to our democracy when these
29:33country when these companies can delete
29:35people at will when they speak out mark
29:38zuckerberg continual refusal to
29:40cooperate with the British inquiry
29:41reveals the challenge that other
29:43countries whole face to hold companies
29:45like Facebook to account the British
29:47Parliament is now considering a standing
29:49summons on mr. Zuckerberg
29:50after Facebook refused and failed to
29:53answer 40 of their questions
29:54the Cambridge analytic ask Angelou has
29:56exposed that social platforms are no
29:58longer safe for users these platforms
30:01are critical part of American cyberspace
30:03and are in desperate need a protection
30:05and oversight I'm still optimistic about
30:08the future of technology but we should
30:10not walk into the future blind and is
30:12the job of lawmakers to ensure that
30:14technology serves citizens and not the
30:16other way around
30:16Thank You chairman Grassley ranking
30:23member Feinstein and members of the
30:24committee thank you for the opportunity
30:26to appear before you today I can
30:29summarize my testimony in three
30:30sentences first using Facebook and other
30:33social media data in ways that are not
30:35transparent to users it's not unusual in
30:38modern political activity second
30:41Facebook's problems appear to result
30:42from a rapidly changing company allowing
30:45itself to drift from forming communities
30:47to serving advertisers and developers
30:49not from a lack of regulation third new
30:53regulations aimed at Facebook's errors
30:55are more likely to protect the business
30:57from competition than benefit consumers
30:59let me summarize my thoughts on each the
31:02best known are probably most effective
31:04political use of Facebook and other data
31:06analytics were the national campaigns of
31:09President Barack Obama so I'll focus
31:10there these campaigns were uniquely
31:13capable at leveraging Facebook Facebook
31:16co-founder Chris Hughes put together the
31:17new media strategy for Obama's US Senate
31:20and 2008 presidential campaigns and a
31:232012 campaign took things to a new level
31:26the campaign's obtained Facebook data of
31:29supporters and their Facebook friends
31:31through Facebook apps and by having
31:33supporters log in to my Barack Obama
31:35comm using their Facebook accounts the
31:382012 campaign's tech team was reputed to
31:41be able to exploit Facebook capabilities
31:44before even Facebook knew those
31:46capabilities existed one Facebook
31:48outreach program lets support us know if
31:50the close friends had not voted during
31:52an election and encouraged the
31:54supporters to contact those friends this
31:57required intimate knowledge of
31:58supporters Facebook friends and the
32:01nature of their relationships
32:03regarding Facebook's failings the
32:05company's primary failure is not being
32:07clear and candid with its users I
32:09distinguished between users who are the
32:11subscribers and the customers who are
32:14those entities that buy ads and other
32:16services for markets to perform well
32:18users should have complete and
32:21understandable information on the nature
32:23of the services they're using even those
32:26that have zero monetary price as in the
32:28case of Facebook this isn't happening
32:30the company has changed how it serves
32:32and uses subscribers without ensuring
32:35that they fully understand Facebook has
32:38morphed from a connector of communities
32:40to someone that investigates people's
32:42lives and filters their Facebook
32:44communications the pivotal moment
32:46appears to be in 2007 when Facebook's
32:49growth stalled from Facebook's beginning
32:51adding users was a primary goal the
32:54stall triggered the company to hire a
32:56team that according to some of its
32:58former members attempted to use
33:00psychological manipulation to increase
33:02users time on Facebook since then it
33:06seems fair to describe the company's
33:08business as gathering people into a
33:10context where they reveal information
33:12about themselves so that others can
33:14market products and ideas in a sense
33:17Facebook's users are its products one of
33:20Facebook's methods for expanding its
33:22reach was a development of newsfeed
33:24newsfeed changed Facebook yet again
33:27making it into a discussion monitor
33:29determining who's a loud voice and who
33:32hears what voices on the platform
33:35Facebook's shifts appear to be the case
33:37of a company allowing itself to drift
33:40each step over the years probably made
33:42sense by itself but taken as a whole
33:45they constitute a change in who the
33:48company is many businesses have gone
33:50down this road when things go too far
33:53competition steps in and now turn my
33:56attention to my last point new
33:58regulations are likely to harm Facebook
34:00users there are two reasons one is that
34:03the problems are business problems that
34:06commonly occur so new regulations are
34:08unlikely to help they won't be focused
34:10on the main issue the second reason is
34:13that new regulations would likely serve
34:15to protect Facebook
34:17and other large internet companies from
34:19competition the European Union's new
34:21general data protection regulation is a
34:24case in point the new regulations are
34:27driving some tech companies especially
34:28small ones out of Europe such
34:31regulations impede freedom stifle
34:33innovation and reduce competition I'd be
34:36glad to answer any questions thanks to
34:40everybody for your testimony and we'll
34:43go to questions now and we'll have five
34:45minute rounds for each individual I'm
34:49gonna start with mr. Whitely I think
34:52these will be easy questions for you but
34:54listen while I lead up to them you
34:57joined SCL group Cambridge analytic
35:01group of companies in June of 2013 in
35:05May of 2014 the group received Facebook
35:10data from Alexander Colgan you stopped
35:14working for the group in July 2014 in
35:18early 2016 Facebook contacted the group
35:22and yourself and asked that Facebook
35:24data received from mr. Cogan be deleted
35:27in March of 2017 following an internal
35:31audit the group certified it had purged
35:34all of mr. Cullen's Facebook data from
35:37its servers the group was retained by
35:41the Trump campaign in the summer of 2016
35:45so these three questions that I think
35:49will be easy for you to answer
35:50so SCL Cambridge had not been retained
35:55by the Trump campaign during your
35:56employment I when I worked at Cambridge
36:02analytic at the Trump campaign was not a
36:04client no - and you did not work for the
36:08Trump campaign while at SCL Cambridge I
36:12have never worked for the Trump campaign
36:14and you weren't there when the company
36:16made its certification to Facebook I was
36:20not in the company or or engaged of the
36:23company when they when they had that
36:24dealing with Facebook now thank you for
36:26that now mister dr. Hirsch
36:29there's been a lot of media attention
36:31around allegations that Cambridge
36:34analytic helped president Trump
36:36improperly influence the 2016 election
36:39by utilizing data received from Facebook
36:42Cambridge claims that the Facebook data
36:44was ineffective was deleted upon
36:47requests from Facebook and was not used
36:50in their work for the Trump campaign in
36:52if Cambridge
36:54so my question if Cambridge had kept and
36:56utilized the Facebook data what impact
37:00do you think the organisation like
37:01Cambridge analytic and strategies which
37:04mr. Wylie described quote as military
37:09style information warfare in the quote
37:11can have on influencing the outcome of
37:14an election given the kind of data
37:16received from Facebook thank you the
37:19question you know as I said in my
37:20testimony it is hard to move people it's
37:23easier to mobilize or potentially D
37:25mobilize people in to persuade people
37:27but as I said there's has been any
37:29evidence presented from Facebook or
37:31Cambridge analytics data that probably
37:33does exist that could answer this
37:35question in other words when Facebook
37:37ads are run or when companies like
37:39Cambridge analytical run as they do it
37:41so with experimentation they have a
37:42control group in a treatment group and
37:44they could know the answer to this
37:45question so if these things were
37:47actually effective I think Cambridge
37:48elekid would know and Facebook would
37:50know based on what we've seen from
37:52public reports and from this history of
37:54targeting about why it's hard to move
37:55people I you know I'm skeptical of of
37:59this this data in these ads moving
38:02people in a substantial way the media
38:08has portrayed the Trump campaign use of
38:10data and firms like Cambridge analytical
38:13as nefarious actions to manipulate the
38:17public quote or question are these
38:20strategies and use of data is something
38:22new in the political world and how about
38:25advertising generally now thank you mr.
38:28mr. chairman know this this is not new
38:31in the political world it's been around
38:32for a long time it feels new to people
38:34because it's never gotten into the
38:36public press before at least not in this
38:38kind of a volume so it's it's
38:40understandable that people feel that
38:42this is violate
38:43gnorm but it's been a norm for a long
38:45time I'll reserve my time and go to
38:49senator Feinstein thanks mr. chairman
38:54mr. Wiley in February 2018 special
38:59counsel Muller indicted 13 Russian
39:02nationals and three companies for their
39:05part in a well-funded coordinated
39:07campaign of information warfare using
39:11social media this information warfare
39:14campaign spearheaded through the
39:16russian-backed internet research agency
39:19started as early as 2014 what can you
39:24tell us about possible connections
39:26between the SCL group or Cambridge
39:31analytic
39:32and Russia thanks for your questions so
39:37one of my concerns is the the level of
39:41engagement that the company had with the
39:44company being Cambridge analytic out
39:46with Luke Oil and executives from Luke
39:48oil which is Russia's second largest oil
39:49company the the firm Cambridge analytic
39:54I made presentations and sent documents
39:57to Luke oil that made reference to its
40:00experience in disinformation that made
40:03reference to its experience in rumor
40:05campaigns attitudinal inoculation
40:09Alexander Nicks emailed me to say that
40:12he passed on a white paper to the CEO of
40:16Luke oil that white paper discussed the
40:21data assets that the company had in the
40:23United States and the strategies that it
40:25was employing the lead researcher that
40:29Cambridge analytical used to harvest the
40:32Facebook data dr. Cogan was also working
40:34on projects in Russia on psychological
40:37profiling at the University of st.
40:39Petersburg the the company had also
40:43engaged contractors who had previously
40:50worked in Eastern Europe for pro-russian
40:53parties
40:55and and indeed the company decided to
40:59test Americans views on the leadership
41:02style of Vladimir Putin and Americans
41:05views on Eastern European issues
41:09relating to Russian expansionism in the
41:11region to be clear the only foreign
41:14leader that was tested when I was there
41:16was Vladimir Putin and and the the bulk
41:20of the foreign issues that were being
41:21tested focused on on Russian
41:23expansionism in 2014 one of the concerns
41:27that I have is that there was a lot of
41:31contact with Russia both Russian
41:35companies and then also via dr. Cook ins
41:37research presentations that were being
41:38made in Russia that made that made it
41:40known that this research was being done
41:42so what I can't I can't say definitively
41:44that this had any relation to the
41:46internet research agency for example but
41:49what I can say is that a lot of noise
41:51was being made to to companies and
41:53individuals who are connected to the
41:56Russian government and and for me that
41:58is of substantial concern do you think
42:04it's possible or even likely that the
42:07Facebook data harvested by Cambridge
42:10analytic ended up in Russia what I can
42:14say is that the lead researcher dr.
42:17Kogan who was managing the Facebook
42:19harvesting project for Cambridge
42:22analytic at was at the time working on
42:26projects that related to psychological
42:27profiling in Russia with a Russian team
42:30as that was going on I also know that he
42:32was traveling to Russia I also know
42:34based on conversations that I had with
42:36him at the time that he was making it
42:38known to his colleagues in Russia about
42:40the project and so I can't I can't say
42:44definitively one way or the other if if
42:47these data sets did end up in Russia but
42:49what I can say is that it would have
42:51been very easy to facilitate that you
42:53told the UK House of Commons that
42:57Cambridge analytic a pitched the Russian
42:59oil company lukoil in its services you
43:03have also said that Alexander Nicks gave
43:06Luke oil a white paper you
43:08that explained Cambridge analytic as
43:10data collection and online targeting of
43:13Americans
43:14lukoil is on the united states sanctions
43:17list and is said to be tied to the
43:20Russian Federal Security Service they
43:24also known as the former KGB when and
43:27where were these meetings taking place
43:29in London in the United Kingdom and also
43:34via the via the phone
43:36what did Cambridge analytic Attell
43:39Lukoil
43:40about its data on Americans did it share
43:43any of that data or is it possible
43:46Russia acquired any of its data on
43:49Americans in sending in sending the
43:53white paper and discussions that I had
43:56with Alexander Knicks about what he was
43:58speaking to the company about I know
44:00that the scale the scale of the of the
44:04data and the location of the data was
44:07made known and also that dr. Kogan was
44:10involved in in that in that data
44:14collection project which and the concern
44:16that I have is that if if you were
44:20intending on acquiring the data even if
44:22you were not intending to acquire the
44:24data with the willing participation of
44:26Cambridge analytics
44:27what was made known was that this kind
44:29of this data could have been easily
44:30acquired by something as simple as a key
44:32logger on dr. Cohen's computer when he
44:34was visiting Russia Canada Thank You mr.
44:40chairman
44:42professor Hersh I don't really have a
44:45question but if what you I'll come back
44:48to you may have time if what you're
44:49saying is that people in America are not
44:54persuadable or persuaded by advertising
44:57I think that's rubbish I think some
45:02really smart people spent 200 billion
45:05dollars
45:05206 million dollars last year on
45:08advertising and I hear kids all the time
45:11walk around saying dilly dilly they
45:14didn't just dream that up okay
45:17but mr. wildly let me ask you a couple
45:21questions I'm not interested in innuendo
45:24or speculation or rumor other than
45:29Facebook lists for me the sources of all
45:32of the data that you know Cambridge
45:35analytically used while you work for
45:39there was several different consumer
45:43data vendors that were used can you name
45:46them please be specific I believe I
45:50believe it
45:53Experion data was used that Believix
45:55axiom data was used I'm not sure but I
46:00can get back to you on whether contracts
46:03were directly with axiom and expiry
46:05seller the state state voter rolls and
46:10registration
46:11who else there were smaller firms which
46:14I can't give you just off the top of my
46:16head who had more specialize in each
46:19data and then in terms of online data
46:22there were experiments being done on
46:25collecting other social media data so
46:27for example Twitter that was also used
46:30but the the basis of the modeling that
46:34was being done at the time that I was
46:35there was primarily using Facebook data
46:38okay
46:39did Cambridge analytic put Facebook
46:42aside for a second the Justice
46:45Department FBI get to the bottom of that
46:48did did Cambridge on Evonik obtained any
46:52of this information and lawfully while
46:54you were there I'm not a lawyer in the
46:59United States so I couldn't comment on
47:01whether it was lawful or unlawful well
47:04did they do it improperly I mean you've
47:07been making normative judgments quite
47:10often I you know don't get religion now
47:13on me so so the the faith the Facebook
47:17data I believed was other than Facebook
47:21other than Facebook the the consumer
47:25data lists that were used were acquired
47:29via contract
47:30that were signed and painted it did
47:32Cambridge analytic ahac anybody while
47:37you were there did you know of the III
47:39have seen documents that make reference
47:42to special intelligence services and
47:45information gathering network that's do
47:48you have copies of those documents not
47:49with me now okay
47:51did Cambridge analytic could get money
47:53from WikiLeaks or get to information
47:55from weakened leaks not when I was there
47:57now okay do you know if they got it
48:00after you were there in terms of money I
48:03don't know you know that money I
48:05misspoke a data I'm still worked up over
48:08Delhi Delhi I'm I I wasn't there when
48:14the when the request was made while you
48:17were there excuse me for interrupting
48:18I'm not trying to be rude but we have a
48:20limited amount of time sure while you
48:22were in Cambridge analytical listen even
48:24give me the names of all its clients
48:26whether it was issue or candidate so at
48:32risk of speaking I am happy to give you
48:36a complete list of the clients that that
48:39were being used at the time okay well
48:44there were I know that there were PACs
48:46and various candidates that were being
48:48like who I mean what can't what issues
48:51in Canada ch so there was a network of
48:54PACs that I believe were primarily
48:58financed by Robert Mercer that were and
49:03then several senatorial and
49:05congressional candidates I believe any
49:07issues there were issue campaigns so for
49:11example John Bolton's PAC was if you
49:14consider that an issue based campaign
49:15then that was an issue based campaign
49:17did Cambridge analytical work for Russia
49:19or anybody connected with Russia while
49:21you were there at work
49:24the we did not have a Russian client at
49:26the time that I was there okay I'll
49:32yield back my 9 seconds
49:34Thanks Thank You mr. Wylie senator leahy
49:37and looking back to the early 2014 under
49:49the leadership of Steve Bannon Cambridge
49:53and literature reportedly began testing
49:55slogans like build the wall deep state
49:59drain the swamp results of message
50:03testing images and policies of Russian
50:07President Vladimir Putin now then later
50:13on of course heard built the wall deep
50:15state drain the swamp in the Trump
50:17campaign but they were taking this they
50:21said before there was any Trump campaign
50:23is that correct yes so the the company
50:28was testing as you as you said slogans
50:32like during the swamp images of walls
50:34paranoia about the deep state in 2014
50:38before the Trump campaign was in
50:40existence the the company learned that
50:47there were segments of the population
50:50that responded to messages like drain
50:55the swamp or or images of walls or
50:58indeed paranoia about the deep state
51:01that weren't necessarily always
51:03reflected in mainstream polling or
51:06mainstream political discourse that that
51:08Steve Bannon was interested in in using
51:11to build his movement you've noted that
51:15Cambridge analytical not actually have
51:18any employed staff it was a front group
51:21of the uk-based SCL group they had
51:25Rebecca Mercer Steve ban among others
51:29other board of directors client work was
51:33handled by a CL am I correct on that yes
51:37now our laws prohibit not Americans from
51:42working on US campaigns
51:44to protect our elections from foreign
51:47interference a 2014 memo from Cambridge
51:53analytical's outside legal counsel Beck
51:56emergency Bennett made these
51:59restrictions very clear did Cambridge
52:04analytic up all of these legal
52:06requirements that his work on American
52:08campaigns yeah so so I'm the the source
52:11that provided the media with thought
52:13that memo which I saw at the tail ends
52:15of my engagement at Cambridge analytic
52:17are the to my understanding that memo
52:21was disregarded because Alexander nicks
52:23continued to be the CEO and they
52:26continued to send people to the United
52:27States who weren't American citizens one
52:30thing that I would just say is that many
52:31of the people who were indeed sent to
52:33the United States who weren't American
52:35citizens were not privy to that memo and
52:37were not made aware that that
52:39potentially there were violations of US
52:41law discusses be used an election
52:50advertising for sometimes they know
52:53somebody lives in particular state they
52:55could target ads that reflect the
52:59interests of that state not all
53:03Cambridge were Cambridge analytical and
53:08correct if I'm wrong in this obtained
53:11the unauthorized Facebook data of 87
53:14million people and then target them was
53:18manipulative disinformation that is that
53:21a quick statement that wasn't everything
53:24that they did but yes that is something
53:26that they did do
53:30how does traditional online marketing
53:33compare with how Cambridge and lytic
53:36obtained individuals information how
53:40they used it so when you are looking at
53:46traditional marketing traditional
53:47marketing first of all doesn't
53:50misappropriate tens of millions of
53:52people's data if they're performing
53:54their duties legally and it is not it is
53:59not or should not be targeted at
54:01people's mental vulnerabilities such as
54:04neuroticism or paranoia or indeed racial
54:07biases traditional marketing does not
54:09exacerbate people's innate prejudices
54:13and and and and and coerce them and make
54:18them believe things that aren't
54:20necessarily true whole service
54:24propaganda machine the discouraged
54:26voters who are more prone to voting for
54:29democratic or liberal candidates and
54:32some of their us clients prevailed
54:34extremely close races
54:38why did Cambridge analytic as American
54:41clients investors like the Mercer family
54:44invested twenty million dollars believe
54:46these tactics might be helpful so Steve
54:51Bannon is a follower of something called
54:53the Breitbart doctrine which which
54:55posits that politics is downstream from
54:58culture so if you want to have any
55:00lasting rendering changes in politics
55:02you have to focus on the culture and
55:05when when Steve banning uses the term
55:08culture war he uses that term pointedly
55:11and they were seeking out companies that
55:16could build an arsenal of informational
55:18weapons to fight that war which is why
55:20they went to a British military
55:22contractor which specialized in
55:24information operations especially in
55:33Russian state security services now let
55:36me announce to everybody that record
55:39stays open for a week and then a few
55:43folks get written questions for
55:44people that are here like Senator Leahy
55:46or people that don't come we'd
55:48appreciate if you'd answer them and get
55:50them back to us as soon as you can
55:53senator Lee Thank You mr. chairman mr.
55:56Wiley I'd like to start with you if
55:57that's okay
55:59mr. Wiley while you fault Cambridge
56:02analytic Oh for using its data that it
56:06obtained you say without authorization
56:08its vote or targeting efforts at the
56:10same time you took that same data with
56:14you upon leaving the company isn't that
56:16right so so because there were no staff
56:21at Cambridge analytic huh
56:23most people were contractors or indeed
56:25had companies so I received a copy of
56:29the data okay so you had that data and
56:32then you started your own company
56:33no I my company was in existence before
56:36I left Cambridge and okay okay because I
56:39was a contractor but after leaving the
56:41company you you had a meeting or perhaps
56:45a series of meetings with a major
56:47campaign to discuss some micro targeting
56:49techniques in short yeah that's no
56:52that's not true I didn't I didn't meet
56:54another campaign to discuss that if
56:59you're in reference to that data set
57:00well it appears that you tried to use
57:03some of the same market as your former
57:07company you were gonna use that data for
57:08something right
57:09you were just going to leave it idle
57:11well to be clear the data was was never
57:15used on any commercial so why did you
57:18take it with you when it's not that I
57:21took the data with me it's just that it
57:24was still in existence at the time that
57:25I left okay what I didn't to be clear I
57:28didn't take any data from Cambridge
57:30analytic oh you didn't take it in the
57:33sense that it was already with you yes
57:36okay what type of work did you
57:39anticipate that your company would
57:40perform after you left Cambridge
57:41analytic oh so I I work mostly in data
57:46analytics looking at different kinds of
57:49social trends so after after I left
57:53Cambridge analytic I
57:55you know I continued working on
57:57independent projects but to be clear I
58:01didn't use that data on any commercial
58:06contract it couldn't that data have
58:08proven useful in some of your work it
58:11could have but I didn't I didn't go on
58:14and sell it right but you could have and
58:16had you been successful with that and
58:18had you gone on and your business become
58:20successful couldn't you have now been at
58:22the receiving end of some of the same
58:23questions now going to Cambridge
58:24analytic a yes but it didn't happen
58:26understood want to go over some
58:30statements that you've made in paragraph
58:3315 of your testimony you say and I quote
58:36when I was at s CL and CA I made it was
58:41made aware of the firm's black ops
58:43capacity which I understood to include
58:45using hackers to break into computer
58:47systems to acquire comprar mat or other
58:50intelligence for its clients the firm
58:52referred to these operations as special
58:53intelligence services or special IT
58:55services how did you learn about these
58:58blackops capacities um Alexander Nix
59:01told me and who was a practical matter
59:04was involved in these black ops my
59:09understanding was that in in some of the
59:13projects that SDL group pants in
59:17different parts of the world you know
59:21misappropriated information was used as
59:24as compromised in in elections in
59:28particular against opposition candidates
59:30okay in paragraph 30 if you say quote
59:37the Russian project undertaken by doctor
59:40Corrigan had a particular focus on the
59:43dark triad traits of narcissism
59:45Machiavellianism
59:47and psychopathy the Russian project also
59:50conducted behavior research on online
59:53trolling how did you learn all of this
59:56and can you describe those projects in
59:58more detail so at the time dr. Kogan
1:00:02told me about it some of the research
1:00:04that he was doing he also told the
1:00:06company about some of the research that
1:00:07was
1:00:08being done by that team in Russia so
1:00:12there's email correspondence from the
1:00:14firm that for example referenced the
1:00:17quote work that he was doing for the
1:00:20Russians so it was initially through
1:00:24conversations that I had with dr. Kogan
1:00:26and then later through the investigative
1:00:29reporting that has been done for the
1:00:31past year with the Guardian and others
1:00:34more details have emerged as well okay
1:00:37professor Hersh the use of social media
1:00:40to micro-target is a fairly new practice
1:00:44but it's my understanding that micro
1:00:46targeting itself is not and sadly the
1:00:50use of provocative information to either
1:00:52divide the electorate or to mobilize
1:00:54portions of the electorate has a long
1:00:56history in our country's political
1:00:58campaigns it was the use of social media
1:01:00to micro-target different than what it's
1:01:04been done how it's been done in the past
1:01:06so there's a lot we don't know we don't
1:01:09know about its effectiveness in some
1:01:10ways but it often looks the same and
1:01:12also responding to senator Kennedy you
1:01:14know just because the campaign spend a
1:01:16lot of money on a particular ad kind
1:01:18doesn't mean it works for a long time
1:01:20campaigns are spending money on
1:01:21robocalls
1:01:22countless experiments have shown that
1:01:23robocalls do nothing in an environment
1:01:26where there's lots of stimuli a lot
1:01:28going on in the campaign a lot of
1:01:30campaign ads don't really work probably
1:01:32nobody in this room or nobody that
1:01:34anyone in this room knows changed their
1:01:35mind as a result of any campaign ad in
1:01:38the 2016 election now for someone who's
1:01:40a director of research for Cambridge
1:01:41analytic they should know I'd be
1:01:43shocking if they didn't know the actual
1:01:45effect an estimate of which campaign ads
1:01:49do what part of my skepticism comes
1:01:51because given that there's been
1:01:53whistleblowing there's been no
1:01:54presentation of any such evidence in
1:01:56randomized controlled environments in
1:01:58which someone shows here is the effect
1:02:00this ad targeting neuroticism or
1:02:03whatever had some effect so again it's
1:02:06very hard to sort this out as technology
1:02:08changes from 2000 2004 8 12 16 lots of
1:02:13things are new but one point we often
1:02:15come back to is that in the presidential
1:02:17election particularly when there's so
1:02:18much going on the effect of one AD or
1:02:21one kind of AD or one row
1:02:22we'll call is usually zero that Senator
1:02:25Whitehouse thank you
1:02:27hello mr. Wylie you've said that
1:02:29Cambridge analytic and the SCL group are
1:02:32effectively the same thing and that
1:02:34Cambridge analytic who was the
1:02:36front-facing company for SDL's American
1:02:38operations is that correct yes what is
1:02:41SCL elections so there is a group
1:02:46company in the UK or was called SEL
1:02:50group which had several dif divisions
1:02:53the largest division when I first joined
1:02:55was defense so SEL defence SEL elections
1:02:59was one of the other divisions SEL
1:03:01commercial etc and they all handled
1:03:03different markets for the company so SEL
1:03:06elections handled political what is what
1:03:10are SEL Canada and aggregate IQ those
1:03:15were subcontractors that were set up
1:03:17during the time that I was there to
1:03:21build out a software infrastructure they
1:03:23played a very significant role in
1:03:25building the the actual infrastructure
1:03:27of rip on the the software product rip
1:03:29on are they essentially the same entity
1:03:31SEL Canada and aggregate IQ you could
1:03:34you could think of them like a franchise
1:03:35and the rippin program was the program
1:03:39that developed the software to use the
1:03:41Facebook data yes once you have
1:03:45algorithms and a target a set of targets
1:03:49you need something to then actually
1:03:50connect those targets with an online
1:03:53Display Network so that that's part of
1:03:55the role that Rippon played what is
1:03:56global science research global science
1:03:58research was the company that was set up
1:04:02by dr. Kogan and you've said that it
1:04:04became a company simply to service
1:04:06Cambridge and oolitic got correct it it
1:04:09became a company
1:04:10as I understand it it became a company
1:04:13so that it could sign a contract with
1:04:15Cambridge a melodica or rather
1:04:16technically SEL is it fair to describe
1:04:20the entities that I have all just
1:04:23described as a coordinated network yes
1:04:27and what was the role of Robert Mercer
1:04:29and funding that coordinated network he
1:04:32was the primary funder who put in tens
1:04:34of millions
1:04:35of US dollars into Cambridge analytic
1:04:37which then distributed that money to
1:04:39that network did that Cambridge
1:04:41analytical network including SCL have a
1:04:45recurring contracting relationship with
1:04:47black cube when I was there we did not
1:04:53have a contract with black cube have you
1:04:55since become aware of a connection
1:04:57between a CL group and black cube and
1:05:00working together on projects I've become
1:05:04aware of relationships that the company
1:05:08had with former members of Israeli
1:05:12security services
1:05:13how about ASI data science you've said
1:05:16that that company CTO worked on SCL
1:05:19projects that it was a subcontractor to
1:05:22Cambridge analytic and that there was a
1:05:24revolving cast of data scientists
1:05:26between that organization and Cambridge
1:05:28analytic ah yes there are a frequent
1:05:31contractor when I was there I believed
1:05:35their role was as a supplier of as the
1:05:38company was growing there was a an
1:05:40increased demand for more and more data
1:05:43scientists and they were I believe the
1:05:46contract that they had was to provide
1:05:48data scientists as your company consider
1:05:50them part of the Cambridge analytical
1:05:54Network we described
1:05:55um if if your definition of network is
1:05:58anybody who has an ongoing relationship
1:06:00then sure you said that the company
1:06:03Palantir had staff that were working on
1:06:06the data at Cambridge analytic that they
1:06:09were meetings with Palantir at Palantir
1:06:11offices and that Palantir staff helped
1:06:13build the models for the Rippon program
1:06:16is that all correct yes
1:06:19although found to clarify Palantir said
1:06:21that all of the work that was being done
1:06:23by Palantir staff was done in a personal
1:06:26capacity so let's just take I've got a
1:06:28minute left let's just take a quick look
1:06:30at Briggs in some of the forces behind
1:06:33brexit were vote leave be leave and
1:06:36veterans for Britain correct yes all of
1:06:39them had contracts with aggregate IQ yes
1:06:42is there anything peculiar about their
1:06:44contracts with aggregate IQ it's highly
1:06:47suggestive
1:06:48coordination and data sharing but that
1:06:52is currently being investigated by the
1:06:53Electoral Commission in the UK because a
1:06:55grid' IQ would have been really hard to
1:06:57find at the time it didn't even have a
1:06:58website it didn't it didn't have a
1:07:00website now and do we know that vote
1:07:01leave actually funneled money and to
1:07:03believe and into veterans for Britain
1:07:05that then went to fund aggregate IQ
1:07:07contracts yes we do
1:07:10do we also have connections between Levy
1:07:12you and you Kip and Eldon insurance yes
1:07:17related to a IQs Briggs at efforts our
1:07:22not to a IQs brexit efforts but rather
1:07:25that that side of the leave campaign
1:07:27engaged with Cambridge analytic at
1:07:29Cambridge analytical directly okay well
1:07:31my time has expired
1:07:32I will if there's a second round I'd
1:07:34love to have one senator corner and then
1:07:37then it will be centered clover char I'm
1:07:39going to step out for a few minutes or
1:07:40at the end of his five minutes you take
1:07:42over mister widely let me start with you
1:07:46please the sort of data mining that
1:07:49you've been describing in the targeting
1:07:52of messages has multiple applications
1:07:56correct it could be a commercial
1:07:58application for example and I buy
1:08:00something on Amazon or navy or Netflix
1:08:04they can send me information about
1:08:06something else I might like it could be
1:08:09used to persuade people in a political
1:08:12campaign for against a candidate or for
1:08:14against a particular issue and it's also
1:08:16can be used for covert information
1:08:20operations by governments correct all of
1:08:24that is correct yes data is like any
1:08:27kind of tool you can use it for various
1:08:29various means some very beneficial and
1:08:32legal and others not right so did SCO or
1:08:35Cambridge analytic serve all comers in
1:08:38other words were you open for business
1:08:40to whoever wanted to purchase the
1:08:42services or was that they that well that
1:08:45was the that was the impression that I
1:08:47gaunt Alexander Nix was quite keen on
1:08:50selling contracts however after
1:08:56Robert Mercer put his investment into
1:08:57Cambridge analytic I do know that we
1:08:59weren't the only restriction that we had
1:09:01was to not work with Democrats okay okay
1:09:04well like for example fusion GPS is much
1:09:09in the news and they provided opposition
1:09:12research to the DNC and the against the
1:09:16Trump campaign the the data itself in
1:09:20the the means by which you analyze it
1:09:22and use it is pretty much agnostic
1:09:25correct in other words you can use it
1:09:26for against a political candidate
1:09:29product or for an information campaign
1:09:33like we said yes okay so mr. Zuckerberg
1:09:37when he was here the other day I he kept
1:09:41saying we don't sell data and I
1:09:45responded to him well you clearly rented
1:09:48I don't know whether that's a fair
1:09:50characterization or not how would you
1:09:52characterize the social media platforms
1:09:56like Facebook use of personal data they
1:10:00say they don't sell it how would you
1:10:02characterize it they've created a
1:10:05platform that that encourages the use of
1:10:08data so it's true that you can't go to
1:10:12Facebook and simply buy Facebook's data
1:10:15but they make it readily available to
1:10:18its customers via its network of
1:10:21applications or the fact that the
1:10:25layouts of people's profiles on Facebook
1:10:28makes it very conducive to scraping data
1:10:30for example although Facebook would say
1:10:33that they don't allow that they still
1:10:35create an a set up which which catalyzes
1:10:39its misuse in my view well mr.
1:10:44Zuckerberg also said that the Terms of
1:10:46Service that that consumers agreed to
1:10:50when they sign on to Facebook he said
1:10:53people probably don't read it or if they
1:10:57read it they don't really understand it
1:10:58is your impression that most of the
1:11:01public has no idea about what they are
1:11:05sharing with these these companies
1:11:08when you even talk to lawyers who read
1:11:11through the terms and conditions some of
1:11:12its even dense for a lawyer so III think
1:11:16that it's unreasonable to expect a
1:11:19regular ordinary person to to have the
1:11:23burden of understanding dense legal text
1:11:25and the other thing that I would say is
1:11:27that you know social media is not really
1:11:31a choice for most people the internet is
1:11:33not really a choice for most people it
1:11:35is very difficult to be a functioning
1:11:36member of the workforce or society and
1:11:38refused to use the internet I don't know
1:11:40a job that would let you go in and
1:11:42refuse to use Google for example I don't
1:11:44know
1:11:45you know most most hiring requires a
1:11:47LinkedIn profile now so so although
1:11:50although we use this narrative of choice
1:11:52because someone's pushing a button even
1:11:55if they had read the Terms and
1:11:56Conditions and understood it they
1:11:58substantively don't really have a choice
1:11:59because in the modern workforce you have
1:12:02to use social media you have to use the
1:12:03internet don't really have other options
1:12:04I don't know a job that would hire
1:12:06somebody who refuses to use the internet
1:12:08so do you think it's too much to expect
1:12:10that these social media platforms get
1:12:13consumers informed consent I mean there
1:12:16is this idea in the law for example if
1:12:19you're going to consent to a surgical
1:12:21procedure by your doctor that the doctor
1:12:23must that your consent must be informed
1:12:25in other words you have to understand
1:12:27what you're agreeing to do you think
1:12:29that's too much to ask for in this
1:12:31context it's not that it's too much is
1:12:33not too much to ask for people
1:12:35absolutely should have informed consent
1:12:37but the the analogy is not is not
1:12:39equivalent when you when you go and see
1:12:42a doctor and you need a surgery you need
1:12:44something and that that you can sending
1:12:46to that surgery is proportionate to the
1:12:48to the the benefit that you're getting
1:12:50when somebody when somebody consents
1:12:52quote-unquote to something online even
1:12:56if they understand that they're
1:12:57consenting to their data can you know
1:13:00being harvested if if that's the only
1:13:01way that you can get a job it's not
1:13:03really genuinely a fair situation and so
1:13:06the the point that I that I would make
1:13:08to you is that it should take a step
1:13:11back from this narrative of consent and
1:13:12start to look at the fact that people
1:13:14don't have a choice they have to use a
1:13:16lot of these platforms to be functional
1:13:18in society in the workforce well I don't
1:13:19have to use Facebook I can use too
1:13:21I can use but-but-but-but all of these
1:13:24platforms do the same thing right they
1:13:26all they all as soon as you sign up to
1:13:29Twitter as soon as you sign up to Google
1:13:31as soon as you sign up to anything they
1:13:33all will be collecting a hugely
1:13:36disproportionate amount of data compared
1:13:38to the utility that they provide to you
1:13:40and further the other thing that I would
1:13:42say is that when you're looking at
1:13:43technology you know there's a question
1:13:45of not just informed consent in the
1:13:47present but reasonable expectations in
1:13:49the future when I first signed up to
1:13:51Facebook it didn't have facial
1:13:52recognition I posted all of my photos
1:13:54and then Facebook developed facial
1:13:56recognition algorithms that then can go
1:13:58and search the internet and find other
1:14:00things that I'm doing right so so it's
1:14:03not a question of just informed consent
1:14:04it's about is it proportionate to the
1:14:06benefit that the consumer is getting is
1:14:08it reasonably expected in the future if
1:14:10there's future developments and that
1:14:11technology that that the consumer did
1:14:13actually consent to that at the time and
1:14:16more broadly this narrative of consent
1:14:18is slightly problematic in the sense
1:14:20that when people have to use these
1:14:23platforms it doesn't matter whether or
1:14:25not they understand if they say if they
1:14:27have to use it to get a job they will
1:14:28still use it and so we are we are sort
1:14:30of coercing and compelling people to
1:14:32hand over a lot of information which
1:14:34which not only now could be dangerous
1:14:36but you have to imagine in the future
1:14:38also what the developments of technology
1:14:40you know what developments will see
1:14:42moving forward and what kinds of risks
1:14:44will be exposing to people ten years
1:14:47from now when that data about them still
1:14:48exists well my my time is up I would
1:14:50just conclude by saying that companies
1:14:53now can and do market their services and
1:14:56products based upon people's
1:14:58expectations of privacy and so if
1:15:01consumers are fully informed about what
1:15:03they're doing what the consequences are
1:15:05they can make their choices that may
1:15:07create markets for other alternative
1:15:10platforms that where people's data will
1:15:12be more protected okay thank you very
1:15:19much mr. chairman thank you for your
1:15:21succinct answers mr. Wiley and there to
1:15:24the point
1:15:24last month Senator Kennedy and I
1:15:26introduced a bill to protect consumers
1:15:29privacy online and included in that
1:15:33bills giving consumers the right
1:15:35to control their own data by allowing
1:15:37people to opt out of having their data
1:15:39collected and requiring companies to
1:15:41notify consumers of a privacy violation
1:15:44within 72 hours you previously expressed
1:15:47support for allowing users to opt out of
1:15:50all targeting criterion three clicks or
1:15:52less as well as rules to require the
1:15:54data collected by each app be
1:15:56proportional to the apps actual purpose
1:15:59would Cambridge analytic have been able
1:16:01to harvest the data of Facebook users
1:16:03and their friends if the users had opted
1:16:05out of having their data attracted by
1:16:07Facebook it it would well the the the
1:16:12the the the problem was that Facebook
1:16:14had had set up applications that that
1:16:19that physically allowed the collection
1:16:21of friends data so one of the other
1:16:22things that I would say is that in
1:16:24addition to giving consumers rights we
1:16:26should be putting obligations on on
1:16:29companies themselves so a principle of
1:16:31privacy by design which treats privacy
1:16:34as an engineering problem as a safety
1:16:36problem would also be incredibly
1:16:39beneficial because privacy is not just a
1:16:42governance issue or a terms issue or a
1:16:44policy issue it's a physical engineering
1:16:46issue when it comes to software I do you
1:16:48think that they would just do that on
1:16:49their own or do you think you know good
1:16:50to have some thank you that is my
1:16:53response to one of my friends on the
1:16:55other side of the Allen white center
1:16:56Kennedy and I have proposed this bill
1:16:58secondly the honest ads Act a bill that
1:17:00I've done with Senator McCain and
1:17:02Senator Warner
1:17:04I understand you've supported the idea
1:17:06that there should be more transparency
1:17:08of political ads absolutely no there's
1:17:10no requirements in place in your written
1:17:12testimony you stated if a foreign actor
1:17:15drop propaganda leaflets by airplane
1:17:17over Florida or Michigan that would
1:17:19universally be condemned as a hostile
1:17:22act but this is happening online as you
1:17:25know Facebook Twitter Microsoft are now
1:17:28supporting my bill but and they are
1:17:31taking measures to dispose these ads
1:17:34especially Facebook do you think that
1:17:36this patchwork of voluntary measures
1:17:39though will be the answer no I think
1:17:41that just just in the same way that we
1:17:43require safety standards and everything
1:17:45else that we care about we should be
1:17:46requiring safety standards
1:17:48and transparency standards in software
1:17:50and online platforms thank you
1:17:52at last month hearing I asked mr.
1:17:54Zuckerberg if Facebook had determined
1:17:56whether the up to 87 million Facebook
1:17:58users whose data was shared with
1:18:00Cambridge analytics were concentrated in
1:18:03certain states mr. Zuckerberg said that
1:18:06he would follow up with the state by
1:18:07state break down of those users I want
1:18:09to ask if you have any knowledge as to
1:18:11whether those Facebook users were mostly
1:18:13in any particular States I I can't say
1:18:17off the top of my head what the density
1:18:19was in each state but I do know that
1:18:21there were a particular focus there was
1:18:24a particular focus on on states from the
1:18:27company's activities on on swing states
1:18:29and states that were winnable by
1:18:30Republicans states like Wisconsin
1:18:32Michigan yes
1:18:35Pennsylvania and I we again await that
1:18:39information from Facebook mr. Wiley I
1:18:42also asked mr. Zuckerberg when whether
1:18:44any of the roughly a hundred twenty six
1:18:46million people who may have been shown
1:18:48content from a Facebook page associated
1:18:50with the Russian troll farm the internet
1:18:52research agency were the same users
1:18:55whose data was shared with Cambridge
1:18:57analytic ah he replied that he believes
1:18:59it is quote entirely possible that there
1:19:02will be a connection there what can you
1:19:04say about the potential for any overlap
1:19:07between the Facebook users whose data
1:19:09Cambridge analytic ah
1:19:10obtained and the users who were shown
1:19:12content from the internet research
1:19:14agency the thing that I would say is so
1:19:17firstly I I didn't ever deal with the
1:19:20internet research agency so I can't
1:19:21comment specifically on the on that
1:19:24entity but to your point my concern is
1:19:27that information either may have been
1:19:32shared or indeed misappropriated again
1:19:34at the second instance by Russian entity
1:19:37from Cambridge analytics what I would
1:19:39say though is that it's not it's not
1:19:43just whether or not these individual
1:19:45records were then targeted because if
1:19:48they were used to build an algorithm
1:19:49whether that algorithm was built by
1:19:51Cambridge analytical or another entity
1:19:53other users who share similar profiles
1:19:56and patterns in their data could have
1:19:58also been exposed you know in a way that
1:20:02you simply looked at were these records
1:20:04targeted specifically even if the answer
1:20:07is no that doesn't mean that those
1:20:08records weren't used to build a
1:20:10targeting algorithm I get it
1:20:11I'll follow up on them on the second
1:20:14round here one last question as you use
1:20:16the word profiling and these voter
1:20:18disengagement as you called it voter
1:20:20suppression one of the most horrifying
1:20:21things from our hearings that I saw were
1:20:23those ads that were clearly made to
1:20:26suppress the votes of African Americans
1:20:29do you have any knowledge about the
1:20:32scope of this activity and how often it
1:20:35occurred out of the company my knowledge
1:20:39relates to the the tail ends of my
1:20:42engagement with Cambridge analytic at
1:20:43one of the one of the things that did
1:20:47provoke me to leave was the beginnings
1:20:50of discussions about voter disengagement
1:20:52I have seen documents that reference
1:20:55photo disengagement and I recall
1:20:57conversations that that it that it was
1:21:00intended to focus on African American
1:21:02kept those documents I will I will
1:21:05discuss with my lawyers the best way to
1:21:07get you that information okay and just
1:21:09to be clear just as you said it would be
1:21:10illegal an act of a legal act to have an
1:21:14airplane come in with those pamphlets
1:21:16this under American law where you are
1:21:18specifically suppressing votes is also
1:21:21in a legal act so thank you - and to be
1:21:24clear I I didn't I didn't partake myself
1:21:27and that's why you laughed in part I get
1:21:30that and I appreciate you coming to
1:21:31testify very much today thank you thank
1:21:35you actually the chairs asked me to
1:21:37chair the committee and I also happen to
1:21:39be in the next water for I'm not taking
1:21:41chairs prerogative it's my turn so thank
1:21:44you all for being here I I think this is
1:21:47a very helpful hearing I intend to be
1:21:49here for the the second round as well
1:21:51and mr. Wylie I just leaned over to the
1:21:54ranking member and said that she was
1:21:56complimenting you on your technical
1:21:58expertise and I said I love having a
1:22:00witness before the stand that I can
1:22:02actually understand I can't necessarily
1:22:04understand lawyers but I can understand
1:22:06data analytics people because I was one
1:22:08I hope that the result of this hearing
1:22:11is actually trying to figure out what if
1:22:13anything Congress should do from a
1:22:14regulatory framework
1:22:16can compliance framework with with
1:22:18respect to the new Avenue for data use
1:22:23you know all of us in these committees
1:22:25have already used data from aggregators
1:22:28we take the voter data we know what
1:22:30voting propensity czar that's downloaded
1:22:33from the boards of election you use that
1:22:34as a basis for targeting voters and then
1:22:37there was the the next wave data
1:22:39aggregator so that you can overlay
1:22:41people's affiliations with associations
1:22:44the magazine subscriptions that's all
1:22:46really become passe in terms of data
1:22:48matching it's been happening for
1:22:50probably ten or twenty years and in our
1:22:53campaigns and I would dare say that
1:22:55every single member here who's gone I
1:22:56ran up for election has had that now
1:22:59that is where they buy data and they
1:23:01aggregate it they built their and then
1:23:03they build their proprietary platforms
1:23:04around it now with the advent of social
1:23:06media we have entities who have come
1:23:10into play that really don't want to sell
1:23:12their data they want to sell the
1:23:14analytic result of that data so that
1:23:16they can target people on certain social
1:23:17media platforms would you agree with
1:23:19that analysis
1:23:20yes we're it's it's the age of access
1:23:24rather than the age of transfer in terms
1:23:27of data and something baseline is a part
1:23:31of their their intellectual capital a
1:23:33part of their or their institutional
1:23:35value uh yes data is an incredibly
1:23:39powerful it's like the new oil for
1:23:43either of the other two witnesses to to
1:23:47chime in a part of when mr. Zuckerberg
1:23:50was before us just some weeks back I
1:23:55tried to focus on what we should all be
1:23:58looking at as policymakers as other
1:24:01practices beyond the camera gen oolitic
1:24:03and I should say that my firm engaged
1:24:05camera channel it occur I met them the
1:24:07day that they proposed it and I saw him
1:24:09again on election day I do have some
1:24:12questions about the focus on trying to
1:24:16get cyclo graphic data from people
1:24:18through in-person interviews it may or
1:24:20may not have been captured through some
1:24:22of the techniques that we're discussing
1:24:23today but I also want to go back and
1:24:26talk about if we're going to do a
1:24:27thorough impartial
1:24:31nonpartisan review of the facts that we
1:24:34really should go back probably over
1:24:35about the last ten years if I don't know
1:24:38if you've had an opportunity to but I
1:24:39would commend it to you to read in the
1:24:42MIT Technology Review they had a
1:24:43three-part series and the series was
1:24:46titled how Obama's team used big data to
1:24:48rally voters and if you go through that
1:24:51that very well-written review and
1:24:54through articles associated with it
1:24:58there were quotes from campaign workers
1:25:00who said we literally took the whole
1:25:02social graph from Facebook and were able
1:25:05to download it through the use of an
1:25:07application they there's a terminal term
1:25:09in here I don't know their specific page
1:25:10but they talk about gamifying the apps
1:25:13that campaign workers would use and
1:25:14possibly gamifying the apps that would
1:25:18be supporters would use and then the
1:25:20Obama campaign I think it may have been
1:25:22the first one could have been the second
1:25:24they actually asked whether or not if
1:25:27you were a Facebook user if you would
1:25:29mind by clicking on an on a button be
1:25:32willing to share information about all
1:25:34your friends in my case I have about
1:25:37forty nine hundred friends so by
1:25:38clicking that button I was actually
1:25:40giving access to thousands of people's
1:25:43information without their knowledge and
1:25:46this was a document in practice that I
1:25:48think also has to be looked at in the
1:25:51context of creating good policy that's
1:25:55based on various uses contemporary
1:25:57unisys uses of data on social media
1:26:00platforms and so in my remaining time in
1:26:04this and the the first round I would
1:26:07just ask you do you believe that some of
1:26:13the technology players today simply have
1:26:16grown so large so quickly that some of
1:26:20what you thought would have been
1:26:21captured through just good corporate
1:26:23governance and code of conduct
1:26:25having somebody from a social media
1:26:27platform who either leans conservative
1:26:30or leans liberal kind of putting their
1:26:33thumb on the scale and giving them
1:26:35people information that they really
1:26:36shouldn't if they were good stewards of
1:26:38the data and the social media platform
1:26:40that they were on and I'll leave that as
1:26:42an open question
1:26:42anybody on the panel and reserve my
1:26:44remainder for the next round I just say
1:26:49that you know I I don't hold a view that
1:26:51Facebook is somehow necessary for our
1:26:53lives as a instructor of political
1:26:55science I encourage my students never to
1:26:57share information about politics or ads
1:26:59or news online because they're not
1:27:02professional editors they should direct
1:27:04their attention to newer sources that
1:27:06have editors the idea that you must
1:27:08contact your friends and family and
1:27:10upload pictures and that there is no
1:27:12alternative is really something that
1:27:13doesn't strike me not as an expert but
1:27:15as an ordinary citizen as something
1:27:17that's that's true
1:27:17and so I think in answering your
1:27:20question Facebook has a lot of data
1:27:22I think it's acted really
1:27:22inappropriately in terms of how its
1:27:25conveyed news sold ads to just about
1:27:27anyone who wants to target hate groups
1:27:29fine bring it in we have an algorithm we
1:27:31don't pay too much attention to it and
1:27:32it's not our fault if it goes wrong I
1:27:34think it's a really terrible way that
1:27:36they've conducted business and again I
1:27:39would encourage people to use Facebook
1:27:41but I am surprised by by people's
1:27:43continuing interest in the company I
1:27:49concur a lot in what he was saying I
1:27:52think it's difficult to argue that that
1:27:54Facebook per se is necessary for
1:27:56people's lives what about half the
1:27:58Internet users aren't on it and there
1:28:00are lots of other types of social media
1:28:03to your point about would someone with a
1:28:06particular ideological political view
1:28:08tilt the scales one way or another I
1:28:10don't know if people would deliberately
1:28:12do it they might but by the nature they
1:28:15will because because what I view as true
1:28:18is influenced by how I think
1:28:20ideologically and so things that I might
1:28:22screen out and say well this this is bad
1:28:25this is disruptive to the community
1:28:26would be influenced by how I think
1:28:28thank you Sarah Coons Thank You senator
1:28:32tell us and thank you to the panel for
1:28:34the chance to be with you today I think
1:28:36the reason this matters is that we're
1:28:38talking about the intersection of
1:28:40several important developments that are
1:28:42difficult for the average American to
1:28:43understand big data and social media and
1:28:48foreign interference in our 2016
1:28:50election and we're trying to tease out
1:28:52what actually happened or didn't happen
1:28:55mr. Riley let me start
1:28:57to Cambridge University professor
1:28:59Michael Kaczynski studied the use of
1:29:01Facebook data and found that based on
1:29:03average of 68 Facebook Likes by a user
1:29:07these are just likes what you liked or
1:29:09disliked you could predict sexual
1:29:12orientation political party affiliation
1:29:14race with 85% accuracy further factors
1:29:19like religious affiliation alcohol or
1:29:21drug use even whether your parents were
1:29:23divorced could be deduced is that your
1:29:26understanding of that analysis and do
1:29:29you think Cambridge analytic oh and the
1:29:32the work of Professor Cogan used that
1:29:35predictive power to develop algorithms
1:29:38that then weaponized differences between
1:29:40Americans for electoral game so the
1:29:44basis of the research that we were doing
1:29:47at Cambridge analytic I was from the
1:29:52papers that you're citing from from dr.
1:29:54Kaczynski so so the firm replicated his
1:29:59approach and then sought to improve it
1:30:02and did I correctly summarize the
1:30:05incredibly high correlation that you
1:30:08could show in terms of really knowing
1:30:09the individual Facebook user if you had
1:30:12access to their likes and their friends
1:30:14and their social media activity yeah it
1:30:16it increases the particular paper that
1:30:19you're citing the the number of likes
1:30:22once you surpass a hundred and 200 you
1:30:25can get to the same level of accuracy at
1:30:29predicting for example personality
1:30:31traits as as your spouse would if they
1:30:35were answering questions about you and
1:30:36in in comparison to how you would result
1:30:38was the point that Professor Hirsch made
1:30:40earlier all of us who have stood for
1:30:42election have struggled with the
1:30:44difficulty of actually targeting our
1:30:46voters effectively because the data sets
1:30:48we've had access to mostly publicly
1:30:51available data are very thin very narrow
1:30:53there's very little that we have the
1:30:55data sets that Facebook has access to
1:30:57that's why it's a multi-billion dollar
1:30:59company are unbelievably deep and rich
1:31:02and so it's unlike anything we've had to
1:31:05confront before correct yeah III don't I
1:31:08don't can contest what Professor
1:31:10Hirsch was saying in the sense that
1:31:11there are you know it is true persuading
1:31:16somebody compared to motivating them to
1:31:19to turn out as much more difficult and
1:31:21the the datasets that traditionally were
1:31:24used are often very sparse and not
1:31:27necessarily reliable but that data it's
1:31:29that available now through Facebook are
1:31:31orders number attitude yes yes and
1:31:33that's why Cambridge analytic ended up
1:31:36pursuing that path because it found that
1:31:39in comparison to traditional marketing
1:31:43datasets the the the data that you could
1:31:45procure from social networking sites was
1:31:48the was much more dense and actually
1:31:50much more reliable to create a precise
1:31:52algorithm and to be clear a billionaire
1:31:55American mega donor and supporter of the
1:31:57Trump campaign funded a shell
1:32:00corporation Cambridge analytic Oh still
1:32:02run by foreign nationals to take
1:32:04advantage of this research this
1:32:06understanding and gained access to
1:32:08eighty seven million Americans Facebook
1:32:11information and developed some of the
1:32:15algorithms that came out of that I want
1:32:17to ask you in the time I have remaining
1:32:19about your experience working with Steve
1:32:20Bannon one of president Trump's senior
1:32:23campaign advisers and the goals that he
1:32:25used Cambridge analytical to achieve was
1:32:27one of his goals to suppress voting or
1:32:29discourage certain individuals in the
1:32:31United States from voting that was my
1:32:34understanding yes was voter suppression
1:32:35a service that US clients could request
1:32:38in their contracts with Cambridge
1:32:40analytical while Bannon was vice
1:32:41president yes and so Steve Bannon is
1:32:46running an organization where you could
1:32:48as a client request in contracts voter
1:32:51suppression using this remarkable data
1:32:54set I don't know if it was referenced in
1:32:57contracts but I have seen documents that
1:32:59make reference to it in relation to
1:33:01client requests and services providing
1:33:03has got a last question you testified
1:33:06that back in 2014
1:33:07Cambridge analytic a set up focus groups
1:33:09message testing and polling on Americans
1:33:12views on the leadership of Vladimir
1:33:14Putin and Russia's expansionism in
1:33:16Eastern Europe
1:33:18you've also testified that it's entirely
1:33:20possible would have been relatively easy
1:33:22for Russian intelligence to
1:33:24put a keylogger on professor Cogan's
1:33:26computer and get access to this full
1:33:29data set why do you think Cambridge
1:33:31analytic Oh was testing Putin's
1:33:33aggressive actions and what threat would
1:33:36it pose to our 2018 elections if this
1:33:39entire data set all the algorithms that
1:33:41go with it are currently in the hands of
1:33:44Russian intelligence so I I don't have a
1:33:48clear answer as to why the company
1:33:50wanted to was so engaged in testing
1:33:52Russian expansionism and the leadership
1:33:54of Vladimir Putin that's a question
1:33:57better put to Steve Benin in terms of
1:34:01the dangers for not just American
1:34:04democracy but you know other democracies
1:34:06around the world this this data is
1:34:09powerful and if it's put into the wrong
1:34:11hands it becomes a weapon and we have to
1:34:14understand that you know companies like
1:34:17Facebook and platforms like Facebook or
1:34:20Twitter are not just social networking
1:34:21sites there are opportunities for
1:34:23information warfare not just by state
1:34:26actors but also non-state actors and so
1:34:28we really do have to look at protecting
1:34:30cyberspace as a national security issue
1:34:34just in the same way that we have
1:34:35agencies to protect our borders you know
1:34:38Lance Ian Ayre thank you very much mr.
1:34:40Whaley thank you
1:34:43Thank You mr. chairman welcome to each
1:34:46of the witnesses thank you for being
1:34:47here you know I think Americans are
1:34:50rightly concerned about privacy and the
1:34:52security of our data and we're also
1:34:55concerned about the power that is being
1:34:58collected in Silicon Valley if a handful
1:35:00of companies having enormous troves of
1:35:02data that they're able to use an employ
1:35:06with very few rules governing what they
1:35:10do with that information much of the
1:35:13media attention in recent weeks and
1:35:15months has focused on the data operation
1:35:17of the Trump campaign but the Trump
1:35:20campaign was hardly the first to employ
1:35:23data in a very significant way in a
1:35:25political campaign dr. Jameson in your
1:35:28written testimony you talked about the
1:35:30Obama campaign in 2008 and 2012 and and
1:35:34their data operations can you share with
1:35:36this committee what
1:35:38the Obama campaign did regarding data in
1:35:40O eight and twelve sure during the the
1:35:442008 campaign what they did is they had
1:35:47as an advisor a co-founder of Facebook
1:35:49who helped him understand how they could
1:35:52use Facebook and obtain data from
1:35:54Facebook and a lot of that work was done
1:35:57by consultants but that was the center
1:36:00of it in 2012 the campaign changed his
1:36:02strategy and then pulled all of that
1:36:04in-house and made it much much more
1:36:05effective they were able to combine
1:36:07their Facebook data more cleverly more
1:36:10carefully with other sources of data and
1:36:13do a much better job of understanding
1:36:14what individual voters were like who was
1:36:17connecting with whom and how they could
1:36:20understand those conversations and his
1:36:23Facebook approached access to data on a
1:36:28fair and even-handed matter allowing
1:36:31candidates from whichever party the same
1:36:35access to data or have they been more
1:36:37political and partisan players in that
1:36:38regard I have no knowledge of that
1:36:41you know I will note that Carol Davidson
1:36:44who was the director of data integration
1:36:46and media analytics for the Obama for
1:36:48America 2012 she said quote Facebook was
1:36:53surprised that we were able to suck out
1:36:55the whole social graph but they didn't
1:36:58stop us once they realized that was what
1:37:02we were doing and she also said that
1:37:05Facebook quote came to office in the
1:37:09days following election recruiting and
1:37:11they were very candid that they allowed
1:37:14us to do things that they wouldn't have
1:37:16allowed someone else to do because they
1:37:19were on our side so that's the head of
1:37:22data analytics for the Obama campaign
1:37:24saying Facebook is giving preferential
1:37:27treatment to the Obama campaign because
1:37:31that's the that that's the political
1:37:33side they're on
1:37:36did they give the Romney campaign the
1:37:39same access to data in the in that
1:37:41election in 2012 not to my knowledge
1:37:44all right how about in 2016 there's been
1:37:47a lot of discussion of what the Trump
1:37:48campaign did with data
1:37:51did the Hillary Clinton campaign have a
1:37:53date operation I presume so but I'm not
1:37:56familiar but that's a lot wasn't excuse
1:37:58me there not was not very much written
1:38:00about her campaign on any of this so
1:38:04does anyone on the panel know what the
1:38:05Hillary Clinton campaign did on the data
1:38:07side does anyone on the panel think that
1:38:13there is a chance in a million years
1:38:15that the Hillary Clinton campaign didn't
1:38:18have a substantial investment in in data
1:38:20analytics mr. Wylie sure um so campaigns
1:38:26across the across the spectrum in the
1:38:28United States use data and and just to
1:38:30to your point you know when we look at
1:38:33you know a lot of people are concerns
1:38:36about for example the ability of big
1:38:38government to inhibit our liberties and
1:38:41choice big data can engineer a situation
1:38:45that limits our choice and our freedom
1:38:48and it's not a partisan issue so to your
1:38:51points about other parties using types
1:38:55of access on Facebook I actually agree
1:39:00with you in the sense that there is a
1:39:03substantial risk of distorting the
1:39:05electoral process if a company like
1:39:08Facebook decides to pick aside whether
1:39:10that is Democrat or Republican and so
1:39:12the thing that I would hope that this
1:39:15committee and others really you know
1:39:18internalize is that this is we're
1:39:20talking about Cambridge analytic ah but
1:39:22it's not a partisan issue we're talking
1:39:25about the future of how these technology
1:39:28companies operate and the risks to
1:39:31ordinary American citizens and the risks
1:39:34to the integrity of our democratic
1:39:36processes here in the United States and
1:39:37around the world and it's that that's
1:39:39not a partisan issue I very much agree
1:39:42with that and I would note that there is
1:39:43an overlay on top of that that Facebook
1:39:46and other social media companies are now
1:39:49the vehicle through which some 70% of
1:39:53Americans get their political news and
1:39:55so the specter of censorship I I think
1:39:59is a profound threat to Liberty and so I
1:40:02appreciate this panel being here and
1:40:03thank you thank you for your testimony
1:40:09senator Blumenthal thanks mr. chairman
1:40:14welcome to you all thank you for being
1:40:16here mr. Wylie are you aware of
1:40:20conversations between Cambridge
1:40:23analytics executives and any
1:40:26representatives of the Russian
1:40:28government or people associated with the
1:40:30Russian government um I'm I'm aware of
1:40:33meetings that the company had with
1:40:35Lukoil which has close connections with
1:40:39with the Russian government those
1:40:41conversations documented anywhere in any
1:40:44letters or emails or any other kinds of
1:40:50evidence there are documents pertaining
1:40:53to the conversations and and
1:40:56presentations made to to lukoil and I
1:41:00have passed on some of those documents
1:41:01to the authorities to this committee I
1:41:07know I get to this committee you know
1:41:09would you be willing to provide them to
1:41:12this committee um I will consult my
1:41:14lawyers instead the best the best way to
1:41:16get you that information thank you
1:41:17during your time at Cambridge analytic
1:41:19oh my understanding is that you are
1:41:23aware of the founders or funders
1:41:30including Robert Mercer providing monies
1:41:36so that they would not be quote
1:41:39necessarily considered declara Beltaine
1:41:42contributions are you aware of
1:41:45conversations to that effect by mr.
1:41:48Mercer or anyone else what I'm
1:41:51referencing is what was explained to me
1:41:54after I inquired as to the relatively
1:41:59convoluted setup that was happening in
1:42:01the United States with respect to the
1:42:03setup of Cambridge analytic huh what was
1:42:06explained to me is there was an and this
1:42:08is necessary to say that this was the
1:42:09primary reason so it could be an
1:42:11ancillary benefit but that when you
1:42:14invest money as an investor into a
1:42:16company that you own
1:42:17it doesn't necessarily constitute an
1:42:22electoral contribution which is declara
1:42:24bill let me let me just cut right to the
1:42:27intent of my question was there specific
1:42:32explanation to you that the purpose of
1:42:35structuring these funds as investments
1:42:38was to in effect avoid the reporting
1:42:42requirements or other provisions of the
1:42:43United States election laws it was
1:42:46explained to me as a benefit of the set
1:42:49up
1:42:50were there any firewalls during the time
1:42:54he worked at Cambridge analytic oh
1:42:56that's separated work on different
1:42:58campaigns or were the funds in effect
1:43:01commingled in all the campaign whilst I
1:43:09was there I did not see firewalls being
1:43:13set up or or or any sort of barriers
1:43:19between people or or conversations if
1:43:21that's your question so there was no
1:43:23recognition of the legal responsibility
1:43:26to separate campaigns I'm aware of memos
1:43:31from the company's lawyers to some of
1:43:35the executives of the company that
1:43:37outlined the responsibilities in the
1:43:38United States to separate contact
1:43:41between staff and activities that relate
1:43:44to different campaigns or PACs etc but
1:43:47those memos were instructions those
1:43:49memos were instructions from or rather
1:43:51advice from lawyers to executives but
1:43:54when I was there I did not see in
1:43:57practice those instructions were not
1:43:58followed No
1:44:02can you provide some specific examples
1:44:05of your direct knowledge of either focus
1:44:11groups or other efforts to suppress
1:44:13voting so in terms of specifics I'm
1:44:20happy to work with the committee to give
1:44:23a more full explanation just I
1:44:25understand there's limits on time but I
1:44:27am aware of
1:44:28of research that was being looked at
1:44:32about what what motivates and indeed
1:44:35what demotivates different types of
1:44:37people and so if you you focus on
1:44:39messaging that demotivates certain types
1:44:41of people you decrease the amount of
1:44:44turnout you dr. Jameson you served as
1:44:49part of the agency landing team in the
1:44:53transition for the Federal
1:44:55Communications Commission for president
1:44:59elect Trump correct yes during that time
1:45:02did you have any contact with Michael
1:45:04Kohn about FCC policy no during the
1:45:08transition did you ever meet him no not
1:45:10to my knowledge
1:45:11thank you thanks mr. chairman senator
1:45:15Harris
1:45:21so we're here today to talk about how
1:45:23Cambridge analytic obtain sensitive data
1:45:25about millions of Americans from
1:45:27Facebook without the users knowledge or
1:45:29consent and then used that data to
1:45:30target voters and influence our
1:45:32elections and there are broader issues
1:45:35of privacy that are highlighted by this
1:45:37incident and I think it's worth stepping
1:45:39back to pull all this in context for the
1:45:42American public to put it plainly most
1:45:45Americans have entered into a bargain
1:45:47with Facebook and other web service
1:45:49providers in which users unknowingly
1:45:51give those companies huge amounts of
1:45:54personal data in exchange for the free
1:45:56service of social networking in turn
1:45:59Facebook uses this data to show its
1:46:02users carefully targeted ads which are
1:46:04the source of 98% of Facebook's revenue
1:46:08interesting interestingly enough this
1:46:11business model makes the Facebook user
1:46:13the product and makes the advertisers
1:46:16the customer but let's be clear this
1:46:20arrangement is not always working in the
1:46:22best interests of the American people
1:46:24first users have little to no idea just
1:46:28how Facebook collects their data
1:46:30including tracking the users location
1:46:33the device they are using their IP
1:46:36address and activities on other web
1:46:38sites to be clear this occurs whether or
1:46:42not they are logged into Facebook and
1:46:44whether or not they even use Facebook
1:46:47let me put this in perspective in the
1:46:50real world this would be like someone
1:46:53following you every single day as you
1:46:56walk down the street watching what you
1:46:58do where you go for how long and with
1:47:02whom you're with for most people it
1:47:05would feel like an invasion of privacy
1:47:06and they called the cops and yes social
1:47:12network sites technically lay all this
1:47:15out in their Terms of Service but let's
1:47:17be honest
1:47:18few Americans can decipher or understand
1:47:21what this contract means second as this
1:47:26hearing illustrates Americans do not
1:47:29have real control over the data
1:47:31collected on them and there's almost
1:47:33nothing that users can
1:47:35once data is shared with third parties I
1:47:37believe it is therefore government's
1:47:40responsibility to help create rules that
1:47:43yield a better and more fair bargain for
1:47:45the American people one that respects
1:47:48their rights as consumers and their
1:47:50privacy in the meantime of a few
1:47:53questions I have for the witnesses today
1:47:55mr. Wiley in particular you've mentioned
1:47:57there's been a lot of discussion about
1:47:59how Canberra Cambridge analytic a
1:48:01targeted african-american voters and
1:48:03discouraged them for participating in
1:48:05elections when Steve Annan was the vice
1:48:07president what specifically did Steve
1:48:10Bannon or anyone else decide motivates
1:48:13or demotivates african-americans to vote
1:48:16so it's it's not just focusing on racial
1:48:21characteristics of people actually when
1:48:24you pull a random sample of African
1:48:26Americans they're not all the same
1:48:28people oftentimes are very different at
1:48:30very different lives and different
1:48:31motivators so when you're looking at any
1:48:35any program whether it is motivating or
1:48:37demotivating someone understanding their
1:48:40internal characteristics is a very
1:48:41powerful thing because you don't treat
1:48:44them just as a black person you treat
1:48:46them as who they are that can be used to
1:48:49encourage people to vote or discourage
1:48:50people to vote so how specifically then
1:48:53did they target african-american voters
1:48:57understanding as you do that the African
1:49:00American population is not a monolith
1:49:01how did they then decipher and determine
1:49:04who was African American so they would
1:49:06target them in their intent to suppress
1:49:09the vote so racial characteristics can
1:49:12be modeled and I'm not sure about the
1:49:17the studies that my colleague here was
1:49:19referencing but we were able to get an a
1:49:23UC score which is a way of measuring
1:49:25accuracy for race that was 0.89 I
1:49:28believe only a you see is what area
1:49:31under the receiving operator operates
1:49:33characteristic it's a way of measuring
1:49:35precision which means it's very high so
1:49:40but to be clear I didn't participate on
1:49:45any
1:49:47voter suppression programs so I can't
1:49:50comment on the specifics of those
1:49:52programs I can I can comment on their
1:49:53existence and I can comment more
1:49:55generally on my understanding of what
1:49:57they were doing but those questions are
1:50:00better placed for or Steve Benin okay
1:50:03and I've joined in the request for any
1:50:04documents you can share with us that are
1:50:06evidence of the conduct you've described
1:50:08on a different note what should Facebook
1:50:10have done to verify that either you or
1:50:12Cambridge analytic in fact had deleted
1:50:15the unauthorized data so I can't speak
1:50:20for Cambridge analytic I can speak for
1:50:22myself in 2016 they sent me a letter
1:50:27that said we're aware that you may still
1:50:31have data from this harvesting program
1:50:34it informed me that dr. Cogan actually
1:50:38didn't have permission to use the
1:50:40application that he was using for
1:50:41commercial purposes and only academic
1:50:42purposes that was actually new
1:50:44information to me I did not know that at
1:50:46the time and then it requested that I if
1:50:51if I still had the data to delete it and
1:50:53then sign a certification of that that I
1:50:59no longer had the data did it require
1:51:01you to sign that under the supervision
1:51:05of a notary or just sign it and send it
1:51:06back it did not require a notary or any
1:51:11sort of legal procedure so I signed the
1:51:16certification and sent it back and they
1:51:18accepted it just as a housekeeping we
1:51:24will have a vote called
1:51:25around noon on time we know that means
1:51:28we need to be there within about 25
1:51:30minutes so we I think we'll be able to
1:51:33get the first round we have senators
1:51:34Durbin Hirono and Booker and I'm happy
1:51:37to stick around to the end of the vote
1:51:38for anyone if we can agree to
1:51:40three-minute rounds in the remaining
1:51:41time before the votes call senator
1:51:42durbin thanks mr. chairman thanks to the
1:51:45witnesses let me say initially senator
1:51:47Harris
1:51:48I thought your presentation on the issue
1:51:50of privacy was spot-on
1:51:53it really described what we are about at
1:51:56least one of the things we should be
1:51:57about I asked a question of mr.
1:52:00Zuckerberg whether he felt comfortable
1:52:02telling me the name of the hotel he
1:52:04stayed in last night and after a couple
1:52:07minutes or seconds of hesitation he said
1:52:09no I've asked a lot of questions in the
1:52:12Senate that one got a lot of attention I
1:52:15think because it really got to the heart
1:52:18of the issue in terms of mr. Zuckerberg
1:52:20and his own feeling about personal
1:52:23privacy and where he would draw the line
1:52:25we know of course from what senator
1:52:28Harris has said what we all know from
1:52:29life experience is the last hotel I
1:52:32stayed in is probably a matter of some
1:52:34record with my name attached to it
1:52:37somewhere who knows but privacy is a
1:52:41critical issue here and the right of
1:52:44Facebook or any entity to use my
1:52:47information without my express
1:52:49permission I think is over the line I
1:52:51sent to Senator Coburn or Senator Cornyn
1:52:54my friend amended earlier when he said
1:52:58that he thought consumers were aware
1:52:59when information was being gathered on
1:53:02him but I'm sure he's wrong we have now
1:53:04put a little piece of electric tape over
1:53:07the camera on my laptop most people do
1:53:11now because they're being watched and
1:53:14they may not even know it but there are
1:53:16two other issues here that I'd like to
1:53:18spend a moment addressing and one of
1:53:21them relates to mr. Nix mr. Nix is a
1:53:25British national
1:53:26I said correct yes and he was clearly
1:53:29involved in some of the campaign
1:53:32activities of Cambridge analytics
1:53:35he was the CEO of the company so he was
1:53:40the point person for all of the clients
1:53:42you know he often made the presentations
1:53:44and recommendations to those clients and
1:53:47the Federal Election Commission says
1:53:49expressly that regulations prohibit
1:53:51foreign nationals from directing
1:53:53dictating controlling or directly
1:53:55indirectly participating in the
1:53:57decision-making process of any person
1:53:59with regard to any election related
1:54:00activities in the United States
1:54:02so there is a red flag or red Union Jack
1:54:06whatever you want to call it that should
1:54:09make it clear that we're in a territory
1:54:12here that it may be may be a violation
1:54:15of law you have said that the
1:54:17involvement of Cambridge and successor
1:54:20organisations and Russia came after you
1:54:25left is it correct I know sorry can you
1:54:30clarify slightly what you mean by my my
1:54:34my experiences direct as it relates to
1:54:37Lukoil
1:54:38as it relates to understanding you know
1:54:43the research that was being done in
1:54:44terms of the focus groups and all of
1:54:45that in the use of information by the
1:54:49Russians in the election campaign oh I'm
1:54:51sorry I misunderstood yes that was
1:54:54either happened some other place or
1:54:55after you had left generally yeah I was
1:54:58I was not aware at the time that there
1:55:01was activities in Russia to influence
1:55:05the United States elections so I would
1:55:08just add that to the second category the
1:55:10first question privacy the second
1:55:12question is the involvement of foreign
1:55:13nationals in the United States campaign
1:55:15whether mr. Knicks personally or Russia
1:55:17as a country trying to influence the
1:55:20impact the third has been brought up on
1:55:22colleagues and I think gets to the heart
1:55:24of another very important issue and that
1:55:26is the issue of the secrecy of this
1:55:27activity the fact that we know that mr.
1:55:31Mercer was engaged in this is because of
1:55:33something called open secrets and other
1:55:35sources that weren't disclosed in the
1:55:39ordinary course of business in this and
1:55:42did you have any guidance from Cambridge
1:55:45when you were there in terms of the
1:55:47secrecy of the clients that you were
1:55:49working for so to to your first point
1:55:53about non-us nationals working in u.s.
1:55:56elections Cambridge analytic I received
1:56:00formal legal advice and which I've
1:56:03disclosed to the media and that legal
1:56:07advice did make clear that the company
1:56:10should not be sending non-us nationals
1:56:12to
1:56:13work on American elections Minh with
1:56:20respect to the the secrecy everybody had
1:56:24to sign a nondisclosure agreement a very
1:56:26thorough non-disclosure agreements and
1:56:28in fact after I left you know when the
1:56:31when the company engaged in a protracted
1:56:35legal correspondence with me also
1:56:37ensured that myself and other people who
1:56:40decided to leave signed a an undertaking
1:56:44of confidentiality which was sort of a
1:56:47higher threshold of NDA you've been
1:56:49asked for some documents earlier and
1:56:51those two you just referred to the legal
1:56:52opinion about the involvement of
1:56:54Cambridge yeah
1:56:56employees it's I've made it public so
1:57:00it's if you wouldn't mind sure it with
1:57:01the canary as well as I'll make a note
1:57:03of hood a copy of the non-disclosure
1:57:04agreement that you signed or others
1:57:06might have signed that would be helpful
1:57:08too thanks mr. chairman Thank You
1:57:09Senator Durbin senator Hirono Thank You
1:57:12professor Hirsch elections very much
1:57:15turn on voter turnout so you would agree
1:57:19that efforts to suppress voter
1:57:21engagement or voter turnout should be a
1:57:23matter of serious concern to us
1:57:25certainly you testified that it is
1:57:28easier to demobilize people than to
1:57:30mobilize them in fact you've testified
1:57:32on the impact of voter ID requirements
1:57:36on voter turnout so if Congress if we
1:57:41were to consider regulating anything and
1:57:42I realized that there are privacy issues
1:57:44it's all very complicated but if you
1:57:46want to focus on some kind of a
1:57:48regulatory scheme should we focus on
1:57:50regulating ads or messages that
1:57:52demobilize people so thank you for the
1:57:55question
1:57:55I think this issue is really complicated
1:57:57personally I've worked as an extra
1:57:58witness for the ACLU in the Obama
1:58:00Justice Department on government voter
1:58:01suppression yes voter ID laws nonsense
1:58:04including regulations this is really
1:58:06different and it's different because
1:58:07campaigns do things that are not nice
1:58:09all the time if a Democratic campaign
1:58:11where to go and remind Trump voters of
1:58:13all of the moral failings and Trump's
1:58:15past and say don't vote for him is that
1:58:18to mobilization or not I don't know I
1:58:20really I know that it's not the thing to
1:58:22figure out whether a message actually
1:58:24demobilize does anybody but lets us know
1:58:26we can come up with some way to define
1:58:29what demobilization because that is a
1:58:31sea of matter serious concern you said
1:58:32that that is easier to affect so that
1:58:36may be an area for us to pursue in terms
1:58:39of any kind of regulation in this very
1:58:42complex space that we're in right now
1:58:44that might be right I I mean I'm not a
1:58:47regulatory expert in terms of you know
1:58:49First Amendment issues and so forth you
1:58:52know if I were to ask someone who
1:58:54they're voting for and they said
1:58:55president Trump and I said don't vote
1:58:57for him I'd really rather you not vote
1:58:59it doesn't seem to me that that is a
1:59:01form of voter suppression I know we're
1:59:04talking about basically people paying to
1:59:07suppress both as mr. Wylie testified
1:59:09that Steve Bannon was running an
1:59:11operation where clients could request
1:59:13voter suppression messages which they
1:59:16would then pay for we're talking about a
1:59:18very different kind of circumstance than
1:59:20somebody just saying yeah well don't
1:59:22vote for so-and-so so I think you
1:59:23understand the differences mr. Wylie you
1:59:26obviously have an awareness of the use
1:59:27of misuse of massive amounts of data so
1:59:30I want to ask you this the US
1:59:32Immigration and Customs Enforcement has
1:59:33proposed a new extreme vetting
1:59:35initiative or life cycle vetting their
1:59:38plan is to hire a contractor to exploit
1:59:40publicly available information such as
1:59:43media blogs public hearings conferences
1:59:46academic websites social media websites
1:59:48such as Twitter Facebook LinkedIn
1:59:51LinkedIn to extract pertinent
1:59:53information regarding targets to
1:59:55determine who will be a productive
1:59:58member of society and who will commit
2:00:00crimes and terrorist acts
2:00:01we're talking about people who are
2:00:03wanting visas to come to our country so
2:00:06according to the nonpartisan brennan
2:00:08center for justice isis plan would
2:00:11automatically fat people for deportation
2:00:13or visa denial based on the exact
2:00:16criteria from the original Muslim ban so
2:00:18it was the original Muslim ban of the
2:00:20president that set up this this extreme
2:00:23vetting program and so do you think that
2:00:28that kind of prediction of human
2:00:30behavior as to whether somebody is gonna
2:00:32become an upstanding contributing member
2:00:35or whether that person is likely to
2:00:36become a terrorist or or because
2:00:38Curnow is even possible from the sorts
2:00:40of information that could be available
2:00:42to the US government so two points that
2:00:47there is no Universal definition of a
2:00:50bad person and therefore it is a social
2:00:53construct and it is it is laden with
2:00:57people's moral judgments so there is no
2:01:00mathematical way to determine whether
2:01:02someone is a bad person in the abstract
2:01:04in the sense that there is no Universal
2:01:07definition of a bad person so the second
2:01:11point that I would make though is that
2:01:13just because you are using data Science
2:01:18and Mathematics you could have the most
2:01:21advanced neural network made yet but if
2:01:27the underlying training set for that
2:01:30algorithm uses systemically biased
2:01:32information so for example if you're
2:01:35looking at criminal justice statistics
2:01:36and you have a model that ends up
2:01:39predicting that this particular
2:01:41african-american is more likely to
2:01:42commit a crime because more
2:01:44african-americans end up in prison you
2:01:46you create an algorithm which is simply
2:01:48reflecting social and moral biases of
2:01:51the of the of the data sets in the PLA
2:01:54okay I I think that you know the
2:01:56predictive value of what the government
2:01:59is pursuing is very questionable and
2:02:01problematic in terms of all kinds of
2:02:03issues privacy you name it and and the
2:02:06accuracy and the ability of that kind of
2:02:08initiative and yet they are pursuing it
2:02:10by the way
2:02:11so I think it's very problematic of
2:02:13professors would you agree that you know
2:02:15we admit government getting their hands
2:02:18and all this kind of information so that
2:02:20they can determine whether somebody
2:02:21would commit crimes or somebody's gonna
2:02:23become an upstanding citizen does that
2:02:25even make sense to you
2:02:26no thank you Thank You mr. chairman
2:02:30Sarah Booker thank you very much mr.
2:02:33Wylie um I have a lot of concerns about
2:02:36how these platforms can be used to pit
2:02:38people against each other one of the
2:02:40greatest values of America is this idea
2:02:42of indivisibility that we have the lines
2:02:44of the tires together are stronger than
2:02:46than the lines that divide us as a
2:02:48country
2:02:49and you know this morning when I woke up
2:02:51I listened to the New York Times podcast
2:02:53which was all about Sri Lanka the
2:02:55headline is when Facebook rumors incite
2:02:57real violence and it was about how the
2:03:00platform Facebook was being used by
2:03:02sinister forces to incite hatred between
2:03:06groups fear between groups and
2:03:08ultimately violence in this circumstance
2:03:11I was just so deeply disappointed and
2:03:16angered when Nigel Oakes who's a founder
2:03:18of the SEL group is you wrote would say
2:03:21things and record a conversation where
2:03:23he's it's things like this that
2:03:25resonates sometime to attack the other
2:03:27group and know that you're going to lose
2:03:29them is going to reinforce and resonate
2:03:32your group which is why Hitler attacked
2:03:35the Jews because he didn't have a
2:03:36problem with the Jews at all but people
2:03:40didn't like the Jews so he just
2:03:41leveraged in our official enemy well
2:03:44that's exactly what Trump did he
2:03:46leveraged a Muslim Trump had the balls
2:03:49and I mean really the balls to say what
2:03:51people wanted to hear this is to me a
2:03:54really frightening reality a threat to
2:03:58the very ideal of a nation that wasn't
2:04:00founded because we all pray alike
2:04:01because we all looked alike but we had a
2:04:03set of common aspirations and democratic
2:04:05principles that united this country our
2:04:07founders talked about pledging to each
2:04:09other not religious alliances not racial
2:04:13alliances but our our sacred honor and
2:04:15and I'm wondering first and foremost
2:04:18your experience with this organization
2:04:21that you ultimately left did you look to
2:04:24see did you feel that this was a
2:04:26manipulation of the data to prey upon
2:04:30divisions prejudices biases inflaming
2:04:34them in order to produce and as a
2:04:37demagogues often do a certain elect
2:04:39electoral outcome sure so the United
2:04:43States went went through a civil rights
2:04:44movement a couple decades ago in an
2:04:47attempt to desegregate society and one
2:04:50of the things that we're seeing now is a
2:04:52resegregate of society that is catalyzed
2:04:57by algorithms so some people call that
2:05:00echo chambers
2:05:01an echo chamber is where you start to
2:05:04receive information from one side and
2:05:07you stop to see information from another
2:05:08side which ultimately interferes with
2:05:11that common fabric because what we're
2:05:13doing is we're tearing that fabric apart
2:05:15with respect to what Cambridge analytic
2:05:19who was doing it was looking to find
2:05:21ways of exploiting certain
2:05:24vulnerabilities in certain subsets of
2:05:26the population so not everyone certain
2:05:28subsets of the population to send them
2:05:31information that will start to then
2:05:32remove them from the commonly the the
2:05:36public forum if you will where they
2:05:38start to see more and more conspiracy
2:05:40theories or they start to see more and
2:05:42more vitriolic messaging and they start
2:05:44to internalize that messaging and they
2:05:47start to you know discount mainstream
2:05:50media and in fact if if there's enough
2:05:52money being spent on targeting these
2:05:55individuals they'll stop seeing any
2:05:56mainstream media online and from the a
2:05:59and from that point they've now been
2:06:00removed from the public forum and for me
2:06:04that's deeply problematic because when
2:06:06you look at what democracy is supposed
2:06:07to be about it's supposed to be there's
2:06:08a an element of a common experience that
2:06:10voters need to have in order to
2:06:12collectively make a decision if you are
2:06:14segregating people in terms of the
2:06:17information space so that one set of
2:06:19voters only see one thing and another
2:06:21set of voters only see another thing we
2:06:23have destroyed the public forum and I
2:06:25know from your your statements first of
2:06:27all that was an excellent observation
2:06:28and from your public statements on your
2:06:32statement you talked about leaving as
2:06:33they're creating videos to try to incent
2:06:37to inflict hatred towards Muslims fear
2:06:40of Muslims that's correct right yes so
2:06:43there were videos created by the firm
2:06:46that were in there's no other way to
2:06:51describe it other than sadistic and
2:06:53Islamophobia and and and I wouldn't I've
2:06:56three seconds left but just to shift my
2:06:59colleague Chairman Tillis talked about
2:07:02Obama using big data to rally voters
2:07:04which i think is I'm somebody that also
2:07:08is very interested in big data and the
2:07:10potentials it has predictive analytics
2:07:12how they can help us with everything
2:07:13from
2:07:14empowering police officers to you name
2:07:16it and we can talk about that but the
2:07:18question this was stunning to me was to
2:07:20use big data not to rally voters but to
2:07:22suppress voters based upon their racial
2:07:25ethnicity and until your testimony am I
2:07:28correct that you're saying that this was
2:07:30a determined effort to suppress not just
2:07:33black votes but the votes of Asians and
2:07:36the most votes of other ethnic
2:07:37minorities it was it was as I understand
2:07:40it so I didn't participate on it but my
2:07:42understanding was that it was to target
2:07:44anybody with characteristics that would
2:07:46lead them to vote for the Democratic
2:07:48Party including and in particular
2:07:50african-americans and it seemed that mr.
2:07:52Bannon was particularly focused on on
2:07:54black voter suppressing african-american
2:07:56voters do you understand where that was
2:07:59coming from do you have any insight into
2:08:01motivations there you'd have to ask mr.
2:08:05Bannon about his own views thank you
2:08:07very much I think we have a few more
2:08:10minutes
2:08:11senator Kennedy would you be open to a
2:08:15three-minute round with a hard gavel of
2:08:17the three minute mark
2:08:19senator Kennedy Jamison Thank You
2:08:26Brittany dr. James I want to be sure I
2:08:28understand your testimony in response to
2:08:32answering mr. Cruz's question in the
2:08:36presidential campaign between Governor
2:08:39Romney and Senator Obama now President
2:08:43Obama the Facebook co-founder shared
2:08:48Facebook information with the Obama
2:08:51campaign that it didn't share with the
2:08:54Romney campaign because the Facebook
2:08:57co-founder wanted mr. Obama to win is
2:09:00that accurate
2:09:00I didn't mean to say that it was that
2:09:03Chris Hughes was working for the Obama
2:09:07campaign who is Christian he was a
2:09:09co-founder of Facebook okay and he was
2:09:11working for the Obama campaign I don't
2:09:15have any information that says he took
2:09:16information from Facebook but he
2:09:18explained to the campaign here's how you
2:09:20use Facebook
2:09:24did he share that information with mr.
2:09:27Romney divided always no he wasn't wrong
2:09:30name okay and do you know why he shared
2:09:34this information with the Obama campaign
2:09:36well he actually volunteered to work for
2:09:39for Obama when he was running for US
2:09:41Senate so they'd had a long relationship
2:09:43all right so he wanted the President
2:09:46Obama to become personal yes okay here
2:09:54is one of the things that to me they're
2:09:56there too there are two issues
2:09:59well they're many but they're at least
2:10:01two issues regarding the social media
2:10:03platforms first is the privacy issue I
2:10:07happen to believe that social media
2:10:09platforms led by Facebook has they have
2:10:12the ability to influence what we believe
2:10:15how we vote what we buy how we feel
2:10:19about ourselves and they have a
2:10:23responsibility to disclose that to
2:10:25people and how they do it and then
2:10:28people still want to use Facebook fine I
2:10:30I don't believe in the lottery for
2:10:33example I think it's a tax on poor
2:10:35people but as long and people but people
2:10:38understand that the odds of winning it
2:10:40are not great I'm I'm pretty libertarian
2:10:42people want to play the lottery more
2:10:44power to them
2:10:45but here's the problem gentlemen and
2:10:47this is the problem mr. Allen I think we
2:10:50could spend several days talking about
2:10:52we can all agree that poison is being
2:10:55spread on social media here's the tough
2:11:00part to find poison I don't want
2:11:04Facebook censoring what I see I don't it
2:11:10doesn't violate or does violate the
2:11:12First Amendment if somebody wants to run
2:11:14an ad on Facebook that says don't
2:11:16believe the mainstream media they're
2:11:19biased if you tell them they can't do
2:11:22that then you don't believe in the First
2:11:25Amendment now it's a lot different if
2:11:29somebody wants to run an ad that says go
2:11:32kill every Rohan go Muslim that you
2:11:36and fine in Burma but where you draw
2:11:39that line is tough talk I'll let you
2:11:47finish but if you've got a great memory
2:11:49we're gonna take a brief recess or I'll
2:11:52let you respond very quickly we're gonna
2:11:53take a brief recess I'm gonna run to and
2:11:56from the chamber and there are at least
2:11:57three members who have asked and we will
2:11:59keep it to three-minute rounds 30
2:12:01seconds I just because I think that
2:12:02Senator Booker and Senator Kennedy's
2:12:04comments really get to my view of the
2:12:06heart of the matter is which is that we
2:12:09have a basic human response that we are
2:12:11attracted to provocation and extremism
2:12:13and and what the platforms are doing
2:12:15online at the basic level is just
2:12:16encouraging that behavior letting us
2:12:19click and share and we're not drawn to
2:12:21things that are civically responsible or
2:12:22truthful we're drawn to extremism and I
2:12:25think to me that's that's the hardest
2:12:28part about this is that it's not about
2:12:29ads it's about what we want and want to
2:12:32share and whose job it is whether it's a
2:12:34cultural a leadership a corporate a
2:12:36governmental response to that very that
2:12:39very human response we have to share
2:12:40things that are not necessarily nice for
2:12:42the world or good Civic professor at
2:12:44some point you've got to trust people
2:12:45yep we're going to go into a recess
2:12:49it'll probably be less than a second
2:12:50round I have mainly now because I have
2:12:53to preside in 20 minutes if we could
2:12:56keep the rounds to about three minutes I
2:12:57would appreciate it or keep the each
2:13:00turn for about three minutes and we'll
2:13:02start with Senator Whitehouse Thank You
2:13:05mr. Wylie I wanted to go back to the
2:13:09brexit campaign and the role that
2:13:14aggregate IQ played in it in the
2:13:18aftermath of that campaign person named
2:13:22Dominic Cummings who was a campaign
2:13:24strategist for vote leave said that
2:13:26tweeted out that Cambridge analytic I
2:13:29had a essentially 0% role in the brexit
2:13:32referendum is that true
2:13:38so it's I I don't I don't I don't agree
2:13:43with that because a IQ was set up to
2:13:47service and build technology for SDL and
2:13:50Cambridge analytic ah so it's true in
2:13:53the sense that Cambridge analytical was
2:13:55the forward facing front for this
2:13:59network for its American operations but
2:14:02not was the front for the brexit
2:14:06operations the the a IQ was only set up
2:14:13to service SEL group and Cambridge
2:14:16analytic have intellectual property
2:14:18agreements that transferred their IP to
2:14:20Cambridge analytic and at the time of
2:14:23the brexit referendum they were also
2:14:25concurrently working with Cambridge
2:14:27analytical on different projects so I so
2:14:30to say that Cambridge analytic I didn't
2:14:32have any role or influence on the brexit
2:14:35campaign I think is looking at it a
2:14:37little too narrowly in fact you've said
2:14:39a IQ played an absolutely pivotal role
2:14:41in the overspending scheme that was set
2:14:43up at vote leave what did you mean by
2:14:46the overspending scheme um when you look
2:14:49at how vote leave transferred money from
2:14:54vote leave into a bunch of different
2:14:56entities some of which it had set up to
2:15:01funnel money into into those schemes and
2:15:05then those schemes those campaigns all
2:15:09coincidentally used the same service
2:15:11provider such that 40% of vote leaves
2:15:14spending went to a IQ forty percent vote
2:15:18leave spending went to an entity that
2:15:19didn't even have a website and would
2:15:20have been virtually impossible despite
2:15:22the fact that Dom Cummings that he found
2:15:24them on the internet I don't know how he
2:15:26would he must be an incredible sleuth if
2:15:28he did there's an individual who you
2:15:30described as playing a pivotal role to
2:15:33use your words in the relationship
2:15:35between Cambridge analytic ah and black
2:15:38cube and you refer to the same woman as
2:15:42playing a pivotal role in the connection
2:15:48of the leave
2:15:50you and Nigeria projects in which a IQ
2:15:54was involved I don't know if I should
2:15:59bring her name into this hearing but you
2:16:02know who I'm referring to I know the
2:16:04person that you're referring to yes what
2:16:07do you mean by pivotal role connecting
2:16:10Cambridge analytic up black cube the
2:16:12Nigeria project and Levy you the person
2:16:15that you're referencing joined the
2:16:20company and played a substantial role in
2:16:24setting up the company's projects in
2:16:28Nigeria which involve black cube which
2:16:31involves the procurement of hacked
2:16:34material and in addition this person was
2:16:40also at the for example press launch of
2:16:44levy you sitting on a panel with other
2:16:47directors of levy you and black cube
2:16:50figured in that how you said she was a
2:16:51pivotal role between the two she as she
2:16:55testified at Parliament
2:16:57she made introductions to a group of
2:17:01Israelis to to Cambridge analytic a and
2:17:05they're the ones who work through the
2:17:06black cube corporation or entity you'll
2:17:11you'll have to ask Cambridge and when I
2:17:13came black cube that okay Sarah char I'm
2:17:17gonna defer to senator Blumenthal or
2:17:19something and then I'll go after him
2:17:22thank you thanks mr. chairman and thanks
2:17:27to senator Klobuchar mr. Wylie are you
2:17:30aware of services performed by Cambridge
2:17:34analytic afore American clients for free
2:17:38or as in-kind contributions at the
2:17:43expense of another entity specifically
2:17:50no but what I would say to that is that
2:17:52the money invested by Robert Mercer in
2:17:57the tens of millions allowed the company
2:17:59at least when I was there to work on
2:18:01projects
2:18:03and charge client substantially less
2:18:06money for the work that was being done
2:18:10than it would have then it would have
2:18:13actually cost had the client would be
2:18:14paying for it entirely themselves you
2:18:18know so it goes back to my earlier
2:18:21question about the commingling of funds
2:18:23as as an investment in other words
2:18:27characterizing money put into Cambridge
2:18:31analytic as an investment rather than a
2:18:33contribution and thereby used for a
2:18:36variety of different campaigns but
2:18:38reported as a contribution for none of
2:18:41them correct because as I understand it
2:18:44and I'm not an expert in this but as I
2:18:47understand it one of the ancillary
2:18:50benefits of the set up was that when
2:18:53whenever mr. Mercer would invest money
2:18:56it would be an investment in a company
2:18:58for research and development which
2:19:02ultimately was for the benefit of its of
2:19:05its clients various packs and campaigns
2:19:08but that benefit wasn't necessarily
2:19:11reported in whichever reporting
2:19:14mechanisms you have in the United States
2:19:16beyond the direct payments that the the
2:19:19campaigns were were using in their
2:19:22contracts and the the dangers to privacy
2:19:28of collecting data and regarding
2:19:33consumers as the product not the
2:19:36customer or their data is sold they are
2:19:39in effect the product and advertisers
2:19:43are the clients is a very serious issue
2:19:47but just to be clear here data was in
2:19:51effect harvested or collected from
2:19:55Facebook according to Facebook without
2:19:59its knowledge as well and what I would
2:20:04like to ask you is whether there were
2:20:05conversations at Cambridge analytic
2:20:08about in effect that taking or
2:20:12harvesting or collecting of this
2:20:14without Facebook's knowing so whenever
2:20:19you have an application you have to
2:20:23submit to Facebook terms and conditions
2:20:25for that application so at the time I
2:20:29was aware that the terms and conditions
2:20:31that the app was sending or at least
2:20:33when I was told the terms and conditions
2:20:36that dr. Kogan sent to Facebook did
2:20:38include clauses about transfers to third
2:20:42parties and commercialization of that
2:20:44data so in effect Facebook was being
2:20:50notified of at least sort of generally
2:20:55that this app is collecting data for
2:20:57those purposes however I never directly
2:21:02communicated with with Facebook about
2:21:06that project until until after I left
2:21:08the company but my understanding is that
2:21:11dr. Cogan did have conversations with
2:21:14Facebook Facebook has consistently told
2:21:17me that they were under the impression
2:21:19that he was doing solely academic
2:21:20research but I wasn't a party to those
2:21:23conversations so you'd have to ask
2:21:24Facebook well in fact in my questioning
2:21:27of Mark Zuckerberg I showed him the
2:21:30Terms of Service indicating that
2:21:32Facebook was put on notice about the
2:21:35possible sale or the use of this
2:21:37information and he denied knowledge of
2:21:40it but you're confirming that you heard
2:21:42about conversation between Cogan and
2:21:44Facebook well I I recall dr. Cogan
2:21:48informing me of conversations that he
2:21:50had with Facebook but I wasn't a party
2:21:52to those conversations however what I do
2:21:54know is that if you set up an
2:21:56application on Facebook you have to
2:21:59submit the terms and conditions of your
2:22:02app for review before Facebook lets you
2:22:05activate that application which was as I
2:22:08understand it the case for this
2:22:09application which meant that Facebook
2:22:11was notified at least via the app review
2:22:14process that there was some intense of
2:22:18transferring or commercializing the the
2:22:20information whether or not they bothered
2:22:22to read the terms and conditions sort of
2:22:24like how many users don't read
2:22:25Facebook's own terms and conditions
2:22:27is another matter though Thank You
2:22:30senator Klobuchar thank you very much
2:22:33mr. chairman mr. professor Hirsch you
2:22:36and I were talking on the break there
2:22:38about this disengagement issue and I'm
2:22:42following up on senator Hirono questions
2:22:44and I pointed out that while you can
2:22:47have a negative ads and I don't think
2:22:50that is exactly what mr. Whaley is
2:22:51referring to about disengagement you do
2:22:54have blatant examples from ads that were
2:22:57bought with Russian money or through
2:23:01other parties from this last election
2:23:03that actually are blatant criminal
2:23:05violations because they tell people to
2:23:08vote with the african-american face you
2:23:10know save time or avoid the lines and
2:23:12text three five four two three and you
2:23:16can vote now do you consider that
2:23:18disengagement yeah that seems terrible
2:23:21and and you know hopefully legal if it's
2:23:23not okay all right I just wanted to
2:23:25clear up bet that there are gray areas
2:23:27there are areas that you probably would
2:23:30say are just persuasion and then there
2:23:32are areas that are purely illegal but of
2:23:35course using foreign money to influence
2:23:38an election is illegal on its face to
2:23:40begin with you would agree with that I
2:23:42agree yeah it's not helpful to clarify
2:23:43the issues all right Thank You mr. Wiley
2:23:46the in addition to the improper access
2:23:52by Cambridge analytical we've also heard
2:23:55about the misuse of user data by other
2:23:57analytics firms like cube U which sold
2:24:00data that it gained from quizzes that
2:24:02had been misleadingly labeled for
2:24:04nonprofit academic research Facebook has
2:24:08acknowledged that we will hear more
2:24:10about incidents like these my question
2:24:12to you is what do we know about the
2:24:13scope of the problem here how many more
2:24:15Cambridge analytic a--'s are out there
2:24:18that's a question that actually only
2:24:21Facebook can answer and you know it's
2:24:23interesting that their behavior before
2:24:27the story broke was to threaten to sue
2:24:30the Guardian they then banned me as a
2:24:33whistleblower from their platform I
2:24:35think in part because you know this may
2:24:38be highlighting
2:24:40a systemic problem that not only
2:24:43Facebook experiences but I suspect other
2:24:45social media platforms do as well and
2:24:48that's why we need rules for
2:24:49transparency exactly so we were just
2:24:51talking about that that it is as much as
2:24:53Facebook has been the focus and and Mark
2:24:56Zuckerberg testified in some ways with
2:24:59the online political ads they have
2:25:01agreed to go further than some of the
2:25:03other companies leading us in terms of
2:25:05disclosure and disclaimers and so do you
2:25:10think that if you just have and I know I
2:25:12asked this before but could you
2:25:14elaborate on just having a few companies
2:25:16out there doing different things how
2:25:19that is not going to lend itself in the
2:25:20future to rules of the road because in
2:25:23my mind you're gonna have companies
2:25:24emerge that are more blatantly violating
2:25:27the rules stuff goes over their
2:25:28political ads go over here and that you
2:25:31must have not only privacy rules in
2:25:33place but some rules of the road when it
2:25:36comes to political ads could you
2:25:37elaborate on that answer I mean we we we
2:25:41we don't allow you know car companies to
2:25:45make unsafe cars and just put terms and
2:25:47conditions on the outside of that car we
2:25:50require seatbelts in all cars and and
2:25:53because it's important for safety and
2:25:56when you look at industries that are
2:25:59important you know cars food medicine
2:26:02nuclear power airlines we have rules
2:26:06that that require safety and to put
2:26:09consumers first because you know it's
2:26:12very difficult for Americans to get
2:26:15around if they don't have a car and so
2:26:17therefore that car should be safe it's
2:26:18very difficult for people to survive
2:26:20without food so therefore that food
2:26:22should be safe and in the 21st century
2:26:24it is nearly impossible for people to be
2:26:26functional in the workplace and in
2:26:28society more at large without the use of
2:26:30the Internet okay absolutely if this is
2:26:34something that affects everybody
2:26:35everyday then there should be some
2:26:38degree of accountability and public
2:26:41oversight speaking which and then I will
2:26:43and here but you and I were discussing
2:26:47on the break algorithms and the changes
2:26:49that we've seen the great changes so
2:26:51Facebook made some changes a few months
2:26:53ago
2:26:53I've always found it a useful platform
2:26:55to do some things that aren't as caustic
2:26:59actually as we see in our politics
2:27:02whether it's engagement about veterans
2:27:05issues or it's something about I do a
2:27:10photo contest with all the people in my
2:27:12state it's been a very positive thing
2:27:14you get they get to display their photos
2:27:16and things like that or Nate you name it
2:27:20and I've seen a dramatic change in those
2:27:23kinds of more moderate posts about doing
2:27:25things with Republicans for instance
2:27:27senator that was including our burn pit
2:27:31bill a dramatic change in those kinds of
2:27:34posts in this change in algorithm and I
2:27:38to me it seems like when we should have
2:27:41a hearing on this alone I think that I
2:27:44am NOT certain that those companies
2:27:46should be deciding what political speech
2:27:50gets exposed it seems like the most
2:27:53volatile I get on Twitter when I you do
2:27:55a post and you get more base retweets if
2:27:58it's really a volatile but Facebook it's
2:28:01just been a dramatic change on the way
2:28:04the posts are being shared and and who's
2:28:06commenting because I have the numbers
2:28:08I've done my own little sting operation
2:28:11as pathetic as it is to check different
2:28:13posts on different days from before this
2:28:15change in algorithm and I know it's
2:28:17happening to me and I've heard other
2:28:19members saying it's happening to them
2:28:20and to me this is free speech and our
2:28:22own democracy as opposed to the Russians
2:28:24obviously I'm gonna talk to Facebook
2:28:25about it I still think it's an
2:28:27incredible platform I want it to work
2:28:29for everyone but could you just comment
2:28:31about this change in algorithms recently
2:28:34and what kind of speech that is
2:28:35reinforced sure so the first thing that
2:28:37I would say is the the concern that you
2:28:39have has to do with distortion right and
2:28:42when we look at what other sectors we
2:28:45you know we regulate for example in
2:28:47financial markets or in mergers and
2:28:49acquisitions and competition if if if
2:28:52there's potential disruptive distortions
2:28:54in the market we we regulate that and
2:28:57what you're talking about is distortion
2:28:59in the access to information that people
2:29:01have an engagement with that information
2:29:03which arguably is even more important
2:29:05because it's the information that we
2:29:06consume
2:29:07to know what's happening around the
2:29:08world so in terms of the specifics with
2:29:12respect to the changes in algorithms
2:29:14Facebook is still a company and
2:29:17therefore it creates algorithms which
2:29:19prioritize paid engagement and penalize
2:29:23or organic engagement and even when it's
2:29:26looking at you know what what pieces of
2:29:31information to show or not show your
2:29:33friend network or your followers things
2:29:35that the algorithm suspect are going to
2:29:38induce more clicks are prioritized you
2:29:42know that's happening before it a change
2:29:45in that somehow so that has become more
2:29:47exacerbated so yes because the other
2:29:50thing you have to understand about
2:29:51algorithms is that they have no
2:29:53awareness they just continue to optimize
2:29:54until they're told to stop right so
2:29:56again getting back to oversight somebody
2:29:59at some point has to look at that and
2:30:01tell a company that's managing those
2:30:04algorithms what is appropriate and not
2:30:05appropriate to to to continue to
2:30:08prioritize and not project it's a
2:30:09message I'm getting is if you're gonna
2:30:10do a more sort of moderate policy
2:30:13oriented post you better pay for that to
2:30:16get that out there because or if you
2:30:18want to do a really a polarizing post
2:30:21that's gonna they're gonna they're gonna
2:30:23prioritize that and I get that things go
2:30:25viral I understand all that I have I
2:30:27write my own tweets blah blah but I'm
2:30:30telling you that yeah that something has
2:30:32dramatically changed and I think I'm
2:30:34hoping that they will look at that and
2:30:37my final comments you're gonna relate to
2:30:40this I think a part of what we have to
2:30:42talk about thank you senator klobuchar a
2:30:45part of what we have to talk about is
2:30:47these models came up with a purely ad
2:30:50based revenue stream
2:30:52I found it surprising that we had a
2:30:54member here who was somewhat somehow
2:30:56offended by the idea that maybe the next
2:30:58generation model is having someone who
2:31:01engages on any social media platform -
2:31:04as they do with Pandora say I pay a
2:31:07certain fee and I get no ads I get none
2:31:10of that engagement that would be the
2:31:11target of these algorithms so we have to
2:31:13make sure that as we go through these
2:31:15social media platforms that suddenly we
2:31:18don't make them I'm somewhere between
2:31:19professor professor
2:31:21and mr. Wiley on just how much we need
2:31:25social media to be a functioning society
2:31:28but I was in talking about paid ads I'm
2:31:31talking about a company deciding how
2:31:32they're going to amplify unpaid but if I
2:31:36may finish all of that targeting at the
2:31:38end of the day creates an interaction
2:31:40that goes beyond the organic interaction
2:31:42that would come by virtue of someone
2:31:44deciding to engage with you and whatever
2:31:45group that is and so it would never
2:31:48become relevant if the industry moves to
2:31:51that model but we have to have a
2:31:52thoughtful conversation around this if
2:31:54we want to continue to have growing
2:31:56social media platforms because we can
2:31:58talk about all the bad they've done but
2:32:00I could also talk about life that
2:32:01they've saved and crowdsourcing that
2:32:04they've been responsible for that has
2:32:06literally fed people and prevented
2:32:08suicides so there's a good social
2:32:10outcome from these platforms but they're
2:32:12going to have to move to a different
2:32:14model in order to sustain themselves and
2:32:17to provide people who want to engage a
2:32:19safer path of interaction I'd like if I
2:32:24may I'd like to offer for for the record
2:32:27without objection a couple of the
2:32:29articles that I mentioned that we're
2:32:30from the MIT Technology Review they were
2:32:33written I think in 2012 and they're
2:32:36mostly they're not taking an opinion but
2:32:38they're talking about some of the uses
2:32:40of these tools and things that we should
2:32:42be mindful of is we're crafting
2:32:44legislation and then I also wanted to
2:32:46offer for the record the it's an
2:32:48editorial I don't know that I agree with
2:32:50all the conclusions in the editorial but
2:32:52they do have quotes in there that have
2:32:54been referenced in fact that I think are
2:32:56also important for people who want to
2:32:57come with a thoughtful outcome to the
2:32:59extent the government has to get
2:33:01involved and address some of the issues
2:33:02that I think have been legitimately so
2:33:05without objection they'll be submitted
2:33:06for the record have been legitimate
2:33:08legitimately brought up thank you for
2:33:10the extended hearing and the time on the
2:33:12committee we welcome your feedback and
2:33:14certainly you should be tracking what
2:33:15we're doing here to form your own
2:33:18opinions about whether or not they're
2:33:19helpful or harmful to addressing some of
2:33:21the issues and then the future of these
2:33:22platforms this meeting is adjourned

About this transcript

This page contains the full transcript of Christopher Wylie Testimony on Cambridge Analytica U. S. Senate Committee 5 -16 -18 by A Sees, generated from the public captions YouTube serves with the video. The transcript has 22,656 words across 3,464 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.