YT YouTube2Text

YouTube2Text → Transcripts → Christopher Wylie Testifies Before U.S. Senate

Christopher Wylie Testifies Before U.S. Senate — Transcript

by Bloomberg News · 18,062 words · 2,794 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00the public seemed generally unaware this
0:03story has forced both the public and
0:06lawmakers to confront serious issues
0:09that need to be addressed including what
0:12role Congress should play in promoting
0:15transparency for consumers regarding
0:18data collection and use while ensuring a
0:22well-functioning marketplace for our
0:25data dependent technologies to drive
0:28further innovation we started that
0:32conversation with mr. Zuckerberg last
0:35month I hope that today we can continue
0:39a productive and meaningful debate about
0:42these serious policy issues
0:45unfortunately events like these more
0:49often than not seem to get muddled by
0:51partisanship and efforts to score a
0:54quick sound bite the Facebook story
0:59first broke December 2015 because the
1:03Guardian identified that dr. Kogan had
1:07allegedly transferred Facebook data to
1:10gain Cambridge analytic in violation of
1:13Facebook's data policies according to
1:17Cambridge press releases and a recent
1:20internal report in July 2016
1:23Facebook requested Cambridge and his
1:26affiliates to remove any data received
1:29from dr. Cogan Cambridge said that they
1:33removed the data and filed legal
1:35certification to Facebook saying as much
1:40I had requested that Cambridge analytic
1:43I appear at this hearing to explain
1:45these facts and tell their side of the
1:48story
1:49Cambridge however recently commenced
1:52insolvency proceedings and therefore
1:54determined it could not participate in
1:57this area the underlying story has not
2:00changed much since 2015 except for two
2:03important events first Cambridge began
2:08doing work for the Trump campaign and
2:11secondly
2:13the Trump President Trump won the 2016
2:17election these two facts sounded an
2:22alarm that revived the cambridge story
2:27this does not diminish the importance of
2:30this discussion but only highlights the
2:34extreme partisanship at play and more
2:37importantly that this conversation could
2:40have easily taken place in 2015 in fact
2:45this conversation could have taken place
2:47much earlier advertising agencies and
2:51political campaigns have utilized data
2:54analytic tools for many years campaigns
2:59including those of presidential
3:01candidates in every election year since
3:04at least the 1990s use data to
3:07micro-target during the past three
3:10presidential elections these strategies
3:13have expanded to social media platforms
3:16specifically Facebook President Obama's
3:20campaign developed an app utilizing the
3:23same Facebook feature the Cambridge used
3:26to capture the information of not just
3:28the apps users but also millions of
3:32their friends President Obama's apt
3:34potentially pulled even more information
3:38than Cambridge's app a former Obama
3:42campaign official Carol Davidson
3:44recently wrote quote Facebook was
3:47surprised that we were able to suck out
3:50the whole social graft end of quote in
3:54the 2012 election we also we could also
3:59be talking about more recent events like
4:02BuzzFeed's partnering with multiple
4:06Democratic and anti-trump super PACs in
4:082016 in a 2016 interview BuzzFeed's vice
4:13president of politics and advocacy said
4:16that one of the problems BuzzFeed was
4:19working with other partners to solve was
4:23quote how are we going to get whim
4:26who do not like Hillary Clinton to vote
4:30for her end of quote
4:32that type of order our reach is not
4:34surprising to many that's because it
4:37happens all the time
4:39similarly it shouldn't be surprising
4:41that President Trump's campaign used
4:44consultants to help reach voters as well
4:48regardless of these events and whether
4:51such tactics are actually effective it
4:54is clear that the use of data across the
4:56political spectrum is only increasing
4:59and so instead of just treating this as
5:03a partisan issue to score political
5:06points the important policy discussion
5:09should really have is whether tech
5:12consumer and Congress where we should go
5:16from here our tech companies have access
5:19to some of our most sensitive data are
5:22these companies doing enough to properly
5:25disclose their data policies and protect
5:30their user date many of the services
5:36offered by such tech companies provide
5:39huge benefits to consumers at little to
5:43no cost our consumers blissfully unaware
5:46or are they making informed choices with
5:51respect to how their data is collected
5:54and used in 2015 the US consumer
5:59technology sector directly provided for
6:02and 7/10 million jobs and generated one
6:05a 9/10 trillion in output four hundred
6:08and thirty five billion in labor income
6:11and 192 billion in tax payments how do
6:15we ensure the proper amount of
6:17regulation to protect consumers without
6:20damaging an industry that hasn't been
6:23vital to the economy these are the
6:26questions we should be asking I hope
6:28today's hearing will allow us to
6:29continue that discussion senator
6:32Feinstein thanks very much mr. chairman
6:36for holding the hearing
6:39in March of this year a series of
6:42articles and videos who are published
6:44online regarding Cambridge analytic and
6:47its efforts to use personal Facebook
6:50data of millions of Americans to
6:53influence United States elections to
6:56date numerous governments have launched
6:58formal investigations into the company
7:01including the United Kingdom Australia
7:04Canada Nigeria Kenya and India there is
7:09much we do not know about Cambridge
7:12analytical but there are significant
7:14facts already in the public record
7:17we know that Cambridge analytical was
7:20established by Robert and Rebecca Mercer
7:22in 2013 at the urging of former White
7:26House chief strategists Steve Bannon as
7:29an American subsidiary of a london-based
7:32firm SCL group it has been reported that
7:37the intent of creating an American shell
7:40was to give the appearance of compliance
7:43with the United States election law that
7:46prohibits foreigners from working on
7:49United States elections according to CEO
7:52Alexander Nicks Cambridge analytic I
7:55worked for candidates in 44 united state
7:59elections in 2014 during the 2016
8:04election cycle mr. Nick stated that
8:08Cambridge analytic and I quote did all
8:11the research all the data all the
8:14analytics all the targeting and quote we
8:19ran all of the digital campaign the
8:22television campaign and our data
8:25informed all the strategy for the Trump
8:28campaign in addition Cambridge analytic
8:32obtained detailed personal information
8:35on approximately 87 million people from
8:39Facebook without their knowledge the
8:43massive data set which reportedly
8:45included approximately 4,000 data points
8:49on each individual was used by Cambridge
8:53and SC ella Canada to develop a
8:57comprehensive voter voter targeting an
9:00online behavioural influenced tool
9:03called Project rip on reportedly project
9:08rip on was a software program that used
9:11sophisticated algorithms to allow
9:14campaigns to segment voters into groups
9:18based on psychological characteristics
9:20such as neurotic or introverted once
9:25individuals were identified and grouped
9:28the platform then provided pre-selected
9:32and focused group tested images and
9:35keywords that were most likely to alter
9:38the behavior of those individuals
9:41examples of the messages developed and
9:44used by Cambridge analytical included
9:48keywords such as drain the swamp and
9:51deep state as well as images of border
9:55walls in an undercover video Cambridge
9:59analytic of managing director mark
10:02Turnbull explained that Cambridge
10:04analytical also created the brand defeat
10:08crooked Hillary the company then created
10:12hundreds of different online
10:14advertisements for that brand including
10:17online videos that were viewed 30
10:20million times through Project rip on
10:24selected images were then sent to the
10:27relevant individuals through online
10:30advertising services like Google and
10:33Facebook these websites provided
10:37feedback on an individual's reactions to
10:41those advertisements which were then fed
10:44automatically back into the targeting
10:47program this is what we've learned in
10:49the past several months however
10:52significant questions remain and there's
10:55much we still do not know about
10:57Cambridge analytical we do not know the
11:01extent to which it worked with hackers
11:03to illegally obtain
11:06damaging information on candidates
11:09including the United States it was
11:12reported in 2015 that Cambridge analytic
11:16as parent company facilitated the
11:19hacking and theft of sensitive medical
11:22records from a Nigerian presidential
11:25candidate and published them online we
11:29do not know whether Cambridge channel
11:31that it could use these tactics in the
11:33United States but this pattern of
11:35activity was certainly used by Russian
11:38intelligence during the 2016 election we
11:43do not know the extent of Cambridge
11:45analytics connections to weaken
11:48WikiLeaks and other Russian interests it
11:51has been reported that Alexander Nicks
11:55contacted WikiLeaks in June of 2016
11:59mr. Nix has said this was his only
12:02contact with WikiLeaks however his
12:05former partner mr. Nigel Oakes has
12:08suggested that Cambridge analytic as
12:11first contact with Julian Assange was
12:15between 12 and 18 months prior it has
12:20also been reported that other employees
12:23at Cambridge analytic had direct
12:26connections to mr. Assange including
12:29through Assange as former attorney in
12:342016 Alexander Nicks
12:37also provided white papers and briefings
12:41to executives from Lukoil Russia's
12:44second largest oil firm about Cambridge
12:47analytic as political activities in the
12:50United States
12:51Luke Oil is currently under United
12:55States sanctions related to the Russian
12:57government's activities in the Ukraine
13:00and in March 2017 Lukoil revealed its
13:05formal information sharing partnership
13:08with the Russian Federal Security
13:10Service the FSB the successor to the KGB
13:15finally we still do not know whether the
13:18data
13:19by Cambridge analytic I was ever shared
13:22with or obtained by a third party the
13:26data was originally obtained through a
13:28Facebook application developed by a
13:31russian-born professor named Alexander
13:34Cogan
13:35professor Cogan maintains a teaching
13:38position at st. Petersburg University in
13:41Russia a state-funded institution and
13:44has traveled frequently back and forth
13:47to Russia
13:48these are concerning questions not only
13:52for the United States but for all
13:54democracies around the world based on
13:58what we have already learned there is no
14:00question that the future of data privacy
14:03will have a significant impact on every
14:07aspect of our lives including our basic
14:10constitutional rights today we're going
14:14to hear testimony from Christopher Wiley
14:17who served as research director at
14:19Cambridge analytic ah
14:21from June of 2013 until November of 2014
14:26I understand he will be able to share
14:30insight into some of these significant
14:33questions based on his first-hand
14:35experience so I very much look forward
14:38to hearing from mr. Wiley and I thank
14:41you again mr. chairman for holding this
14:43hearing I'm now going to introduce our
14:46witnesses and after I do that I would
14:51ask you to stand and be sworn also it
14:56was our intention senator Feinstein and
14:58I to come down and shake hands with you
15:00and welcome you I'm sorry that
15:03everything was talked here so we didn't
15:05want to mess everything up but we wanted
15:07both of us want to thank you for
15:09participating and particularly you mr.
15:12Wiley coming as far as you have to help
15:15help us with proper testimony our first
15:19witness is dr. Aden Hirsch as dr. Hirsch
15:24is an associate professor at Tufts
15:27University foking focusing on American
15:30politics
15:32he spent seven years as an assistant
15:34professor at Yale University is a
15:37nationally published author he earned a
15:40PhD degree and master's degree in
15:42political science from Harvard
15:43University and a back and a bachelor's
15:46degree from Tufts next person besides
15:54thanking you we welcome you to this
15:56hearing and hopefully you feel
15:58comfortable and everything I know you've
16:00testified elsewhere
16:03Christopher Wiley is a former director
16:05of research at Cambridge and let Iike
16:08mr. Wiley worked with President Obama's
16:11campaign for the Canadian Liberal Party
16:13and for the United Kingdom's Liberal
16:15Democratic Party from 2013 to 2014 he
16:19worked for Cambridge analytics while
16:21also studying for a PhD and fashioned
16:25trending forecasting he has an
16:27undergraduate degree in law from the
16:29London School of Economics lastly dr.
16:33Mark Jameson is a visiting scholar at
16:36American Enterprise Institute he's also
16:39the director and günther professor of
16:42public utility Research Centre
16:44University of Florida's Warrington's
16:47College of Business dr. Jameson earned a
16:50PhD in economics from Warrington College
16:54of Business at the University of Florida
16:56and a master's of bachelor's degree from
16:59Kansas State University so would you
17:02three stand now please and so we'll
17:20start with Professor Hirsch and then mr.
17:23Wiley and then dr. Jamison Kathleen
17:27distinguished members of the committee
17:29thank you for inviting me here today
17:32Bob just a minute we may have to do
17:35something with your microphone try talk
17:40again and see if it says hear me
17:43is it okay
17:52is there anything weak here we have
17:57another person coming with another
17:58microphone we've had this happen before
18:02so it's not it's not you and kind of
18:18pull it as close as you can go ahead
18:32could you
18:37but okay we have to turn it on here is
18:42what I'm told
18:54okay good thank you
18:58the controversy surrounding Cambridge
19:00analytic and Facebook has raised a
19:02number of serious concerns foreign
19:05interference in u.s. elections the
19:07personal privacy of Facebook users and
19:09Cambridge analytic has voter targeting
19:11strategies my expertise is on civic
19:14engagement personal data and voter
19:15targeting
19:16and I hope to be able to answer
19:17questions you have in these areas let me
19:20briefly summarize a few points in my
19:21written testimony every election brings
19:24exaggerated crane claims about the
19:26effects of the latest technologies after
19:29an election there is always a demand to
19:30figure out why the winning campaign won
19:32and the latest technology used by the
19:34winning campaign is often a good
19:36storyline even if it's false campaign
19:39consultants also have a business
19:40interest in appearing to offer special
19:43skills and products and so they often
19:44embellish their role to the media
19:46Cambridge analytic Oh was relatively new
19:49to the scene in 2016 and it promoted new
19:51strategies of psychographic targeting to
19:53persuade voters from everything I've
19:56seen publicly disclosed about Cambridge
19:57analytic a I am skeptical of the idea
20:00that its strategies of voter persuasion
20:01were unusually effective or contributed
20:04meaningfully to the election outcome to
20:06understand why it's useful to divide a
20:08campaign strategy into mobilization and
20:11persuasion mobilization entails finding
20:14supporters and encouraging them to vote
20:16persuasion entails encouraging voters to
20:19support your candidate compared to
20:21mobilization persuasion is very very
20:23hard and Facebook data or data from
20:26other sources doesn't help very much let
20:28me explain why first persuasion effects
20:31decay rapidly research has shown that an
20:34ad may change a voters mind for a
20:36fleeting moment but the effect goes away
20:38amidst all the other ads news posts and
20:41stimuli that accompany an election
20:43season second persuasion is hard because
20:46it is actually very hard even with
20:48top-notch data to figure out which
20:50voters are persuadable the reason for
20:54this is no person is persuadable all the
20:56time persuade ability isn't a stable
20:58disposition you might be persuadable now
21:01but not tomorrow by one kind of message
21:03or messenger but not in
21:04other third persuasions hard because
21:07when you try to figure out who is
21:09persuadable your predictions of who is
21:11persuadable are often wrong they're
21:13imperfect and you end up sending
21:15targeted messages to the wrong
21:17population for example campaigns often
21:20want to predict the racial identity of a
21:22voter predictions of which voters are
21:25black or Hispanic are wrong about 25% of
21:27the time so when a campaign sends a
21:30message targeted to these voters a
21:31quarter of the people receiving the
21:33message will be missed targeted research
21:36suggests that voters penalize candidates
21:38who missed target them and if campaigns
21:41get race wrong a quarter of the time
21:42they're going to have much much more
21:44error in estimating something nuanced
21:46like a personality how extroverted or
21:49neurotic you are based on your Facebook
21:51lights which is what Cambridge analytic
21:53acclaimed to be doing the idea that
21:55Cambridge analytic oculd use Facebook
21:57Likes to predict personalities then use
21:58those predictions to effectively target
22:00ads strikes me as implausible given what
22:02we know about the stet the challenges of
22:04persuasion and campaigns no evidence has
22:06been produced publicly about the firm's
22:08profiling or targeting to suggest that
22:10its efforts were effective or that it
22:12overcame the difficulties in persuasion
22:14that I just articulated in spite of my
22:17skepticism there is a lot we do not know
22:19about how campaigns are using social
22:21media to target voters this controversy
22:24gives us some anxiety in part because we
22:26don't really know where the line is
22:27between ads attempting to persuade
22:30voters and ads attempting to manipulate
22:32or deceive voters this anxiety is all
22:35the more understandable because Facebook
22:37hasn't really taken seriously its solemn
22:39civic role as a facilitator of news and
22:42political communications in my written
22:45testimony I described an initiative
22:47currently underway by which independent
22:49researchers can measure the effects of
22:51Facebook ad targeting and new sharing
22:52the success of this program depends on a
22:55serious commitment by Facebook to share
22:58its data even and especially in cases
23:00that will bring negative press to the
23:01company in raising skepticism about
23:04Cambridge analytics I do not at all mean
23:06to suggest that there aren't serious
23:08concerns stemming from this controversy
23:10related to privacy foreign interference
23:12and the troubling role of social media
23:14in transmitting news information
23:17fake news and fake information rather I
23:20hope we can focus on the more important
23:22of these issues as I discuss in my
23:24written testimony thank you again for
23:26your attention to the subject matter and
23:27I welcome your questions talk all right
23:38how's that good mr. chairman senators
23:42thank you for the opportunity to invite
23:45me to speak today
23:46Cambridge analytics is the canary in the
23:49coal mine we must address the Digital
23:52echo chambers that are being exploited
23:54to algorithmically segregate American
23:56society online communities should unite
23:59us and not divide us data is the new
24:02electricity of our date of our digital
24:04economy and just like electricity we
24:07cannot escape data online platforms
24:10terms and conditions present users with
24:12a false choice because using the
24:14Internet is no longer a choice Americans
24:17cannot opt out of the 21st century all
24:20revolutions throw up new power
24:22structures the American Revolution
24:24required the Constitution to ensure that
24:26citizens of the young Republic were
24:28protected from the excesses of arbitrary
24:30government the Industrial Revolution
24:32required protections for workers against
24:35hazardous conditions in the workplace
24:37and in the environment so too with the
24:40digital revolution unless we realize
24:41that there are there is a new game being
24:43played and that the rights to life
24:46liberty and the pursuit of happiness
24:47need to be defended from those who might
24:50undermine them using new technologies
24:52whether they are corporations whether
24:54they are nations or whether they're
24:56non-state actors I've come here today
24:58voluntarily as a witness and as a
25:00whistleblower
25:01I have already reported these matters to
25:04the authorities and it should be made
25:05clear that I am NOT a target of these
25:07investigations I was the director of
25:10research at SCL and Cambridge analytic
25:12Oh from mid 20 2013 to late 2014
25:16SDL group was a British military
25:18contractor specializing in information
25:20operations Cambridge analytic Acree 'it
25:23was created to allow SCL to work in the
25:26united states cambridge analytic i did
25:28not have any employed staff and all of
25:30its clients were
25:31handled by SCL although lawyers warned
25:34about using foreign citizens the firm
25:36installed a non-us citizen as its CEO
25:39and embedded non US citizens in American
25:42campaigns Cambridge analytic I offered
25:45voter disengagement services in the
25:47United States and there are internal
25:49documents that I have seen that make
25:51reference to this tactic my
25:52understanding was that this was targeted
25:54at African American voters when I was at
25:57Cambridge analytic I I was also made
25:59aware of the firm's black ops capacity I
26:02have seen documents relating to
26:04instances where the firm sought to
26:06procure hacked material some of these
26:08targets of some of the targets of these
26:11special intelligence services are
26:13currently heads of state in various
26:14countries I've also seen internal
26:16documents that make reference to the use
26:19of specialised insert intelligence
26:21services from former members of Israeli
26:23and Russian state security services a
26:27further concern is Cambridge analytic
26:29--is employment of people closely
26:31associated with WikiLeaks and Julian
26:33Assange in addition to making approaches
26:35to WikiLeaks the firm hired two senior
26:37staff
26:38both of whom were previously aides to
26:40the British lawyer representing Julian
26:41Assange Cambridge analytic has sought to
26:44identify mental vulnerabilities in
26:46voters and worked to exploit them by
26:49targeting information designed to
26:51activate some of the worst
26:52characteristics in people such as
26:53neuroticism paranoia and racial biases
26:56to be clear the work of Cambridge
26:58analytics is not equivalent to digital
27:00to to traditional marketing Cambridge
27:03analytic is specialized in
27:04disinformation spreading rumours comprar
27:06matte and propaganda for those who claim
27:09the profiling does not work this
27:11contradicts copious amounts of
27:12peer-reviewed literature in top
27:14scientific journals even Facebook
27:16applied for a patent on quote
27:18determining user personality
27:19characteristics from social networking
27:21systems dr. Alexander Cogan was selected
27:24by Cambridge analytics to lead the data
27:26harvesting operation and over 80 million
27:28data subjects had their personal data at
27:30misappropriated given this scale this
27:33could be one of the largest breaches of
27:34Facebook's data Cambridge analytic akan
27:37tractors had also worked on pro-russian
27:39political operations in Euston Europe
27:41including with suspected Russian
27:43intelligence agents
27:45which analytic has set up focus groups
27:47and polling on Americans views on the
27:49leadership of Vladimir Putin and Russian
27:51expansionism in Eastern Europe dr. Kogan
27:54was also working on Russian state funded
27:57research projects the Russian project at
27:59st. Petersburg was also building
28:01algorithms using Facebook data for
28:03psychological profiling and researched
28:05social media trolling in fact Cambridge
28:08analytic had pitched quote the
28:10interesting work Alex Cogan has been
28:11doing for the Russians end quote to his
28:13other clients Cambridge analytical also
28:16was in close contact with senior
28:17executives at Lukoil one of Russia's
28:19largest oil companies Cambridge analytic
28:21a presented Lukoil
28:22with documents outlining his experience
28:24with foreign test information rumour
28:26campaigns and it's American data assets
28:29Facebook knew about Cambridge analytic
28:31as schemes since 2015 before the story
28:34broke Facebook threatened to sue the
28:36Guardian and then banned me for
28:38whistleblowing responding on behalf of
28:40the British government the UK Secretary
28:42of State for culture called my ban
28:44outrageous because it reveals the
28:47unrestrained power technology companies
28:49can exercise over ordinary citizens when
28:51a person's entire online presence can be
28:53so quickly and so thoroughly eliminated
28:56from existence there is no check on this
28:58power and it raises a serious question
29:00for Republicans and Democrats alike what
29:03happens to our democracy when these
29:05country when these companies can delete
29:07people at will when they speak out mark
29:10zuckerberg continual refusal to
29:11cooperate with the British inquiry
29:13reveals the challenge that other
29:14countries whole face to hold companies
29:16like Facebook to account the British
29:19Parliament is now considering a standing
29:20summons on mr. Zuckerberg after Facebook
29:22refused and failed to answer 40 of their
29:25questions
29:26the Cambridge analytical scandal has
29:28exposed that social platforms are no
29:30longer safe for users these platforms
29:32are critical part of American cyber
29:34space and are in desperate need a
29:36protection and oversight I'm still
29:38optimistic about the future of
29:40technology but we should not walk into
29:42the future blind and is the job of
29:44lawmakers to ensure that technology
29:46serves citizens and not the other way
29:48around
29:48thank you
29:53chairman Grassley ranking member
29:55Feinstein and members of the committee
29:56thank you for the opportunity to appear
29:58before you today I can summarize my
30:01testimony in three sentences first using
30:04Facebook and other social media data in
30:06ways that are not transparent to users
30:08it's not unusual in modern political
30:11activity second Facebook's problems
30:13appear to result from a rapidly changing
30:15company allowing itself to drift from
30:18forming communities to serving
30:20advertisers and developers not from a
30:22lack of regulation third new regulations
30:25aimed at Facebook's errors are more
30:27likely to protect the business from
30:28competition than benefit consumers
30:31let me summarize my thoughts on each the
30:34best known are probably most effective
30:36political use of Facebook and other data
30:38analytics were the national campaigns of
30:40President Barack Obama so I'll focus
30:42there these campaigns were uniquely
30:44capable at leveraging Facebook Facebook
30:47co-founder Chris Hughes put together the
30:49new media strategy for Obama's US Senate
30:52and 2008 presidential campaigns and the
30:552012 campaign took things to a new level
30:58the campaign's obtained Facebook data of
31:00supporters and their Facebook friends
31:02through Facebook apps and by having
31:04supporters log into my Barack Obama com
31:07using their Facebook accounts the 2012
31:10campaign's tech team was reputed to be
31:13able to exploit Facebook capabilities
31:15before even Facebook knew those
31:17capabilities existed one Facebook
31:20outreach program let's supporters know
31:22if they're close friends had not voted
31:23during an election and encouraged the
31:25supporters to contact those friends this
31:28required intimate knowledge of
31:29supporters Facebook friends and the
31:32nature of their relationships regarding
31:35Facebook's failings the company's
31:37primary failure is not being clear and
31:39candid with its users I distinguish
31:41between users who are the subscribers
31:43and the customers who are those entities
31:46that buy ads and other services for
31:48markets to perform well users should
31:51have complete and understandable
31:53information on the nature of the
31:55services they're using even those that
31:57have zero monetary price as in the case
32:00of Facebook
32:00this isn't happening the company has
32:03changed how it serves and uses
32:05subscribers without
32:06ensuring that they fully understand
32:08Facebook has morphed from a connector of
32:11communities to someone that investigates
32:13people's lives and filters their
32:15Facebook communications the pivotal
32:17moment appears to be in 2007 when
32:20Facebook's growth stalled from
32:22Facebook's beginning adding users was a
32:24primary goal the stall triggered the
32:27company to hire a team that according to
32:29some of its former members attempted to
32:31use psychological manipulation to
32:33increase users time on Facebook since
32:37then it seems fair to describe the
32:39company's business as gathering people
32:41into a context where they reveal
32:43information about themselves so that
32:45others can market products and ideas in
32:48a sense Facebook's users are its
32:50products one of Facebook's methods for
32:53expanding its reach was a development of
32:55newsfeed newsfeed changed Facebook yet
32:58again making it into a discussion
33:00monitor determining who's allowed voice
33:03and who hears what voices on the
33:05platform Facebook shifts appear to be
33:09the case of a company allowing itself to
33:11drift each step over the years probably
33:13made sense by itself but taken as a
33:16whole they constitute a change in who
33:19the company is many businesses have gone
33:22down this road when things go too far
33:24competition steps in and now turn my
33:27attention to my last point new
33:30regulations are likely to harm Facebook
33:32users there are two reasons one is that
33:35the problems are business problems that
33:37commonly occur so new regulations are
33:39unlikely to help they won't be focused
33:42on the main issue the second reason is
33:44that new regulations would likely serve
33:47to protect Facebook and other large
33:49Internet companies from competition the
33:52European Union's new general data
33:54protection regulation is a case in point
33:56the new regulations are driving some
33:59tech companies especially small ones out
34:01of Europe such regulations impede
34:03freedom stifle innovation and reduce
34:06competition I'd be glad to answer any
34:09questions
34:11thanks to everybody for your testimony
34:14and we'll go to questions now and we'll
34:16have five minute rounds for each
34:18individual I'm gonna start with mr.
34:22Whitely I think these will be easy
34:24questions for you but listen while I
34:26lead up to them you joined SCL group
34:31Cambridge analytic a group of companies
34:34in June of 2013 in May of 2014 the group
34:40received Facebook data from Alexander
34:44Cogan you stopped working for the group
34:47in July 2014 in early 2016 Facebook
34:52contacted the group and yourself and
34:54asked that Facebook data received from
34:57mr. Colgin be deleted in March of 2017
35:01following an internal audit the group
35:04certified it had purged all of mr.
35:07Cullen's Facebook data from its servers
35:10the group was retained by the Trump
35:13campaign in the summer of 2016 so these
35:19three questions that I think will be
35:21easy for you to answer
35:22so SCL Cambridge had not been retained
35:26by the Trump campaign during your
35:28employment when I worked at Cambridge
35:33analytic had the Trump campaign was not
35:35quiet enough to and you did not work for
35:40the Trump campaign while at SCL
35:42Cambridge I have never worked for the
35:45Trump campaign and you weren't there
35:47when the company made its certification
35:49to Facebook I was not in the company or
35:53or engaged of the company when they when
35:55they had that dealing with Facebook now
35:57thank you for that now mister dr. Hirsch
36:01there's been a lot of media attention
36:03around allegations that Cambridge
36:05analytic helped President Trump
36:07improperly influenced the 2016 election
36:11by utilizing data received from Facebook
36:13Cambridge claims that the Facebook data
36:16was ineffective was deleted upon request
36:19from Facebook and was not used in their
36:22work for the Trump campaign if Cambridge
36:25so my question if Cambridge had kept and
36:28utilized the Facebook data what impact
36:31do you think the organisation like
36:33Cambridge and live Iike and strategies
36:36which mr. Riley described quote as
36:39military style information warfare end
36:42quote
36:43can have on influencing the outcome of
36:45an election given the kind of data
36:48received from facebook thanks for
36:50question you know as I said in my
36:51testimony it is hard to move people it's
36:55easier to mobilize or potentially
36:56demobilize people in to persuade people
36:58but as I said there's has been any
37:00evidence presented from Facebook or
37:02Cambridge analytic data that probably
37:04does exist that could answer this
37:06question in other words when Facebook
37:09ads are run or when companies like
37:11Cambridge analytical run ads they do it
37:12so with experimentation they have a
37:14control group in a treatment group and
37:15they could know the answer to this
37:17question so if these things were
37:19actually effective I think cameras only
37:20would know and Facebook wouldn't know
37:22based on what we've seen from public
37:24reports and from this history of
37:25targeting about why it's hard to move
37:27people I you know I'm skeptical of of
37:31this this data in these ads moving
37:33people in a substantial way dr. Davidson
37:38the media has portrayed the Trump
37:41campaign use of data and firms like
37:43Cambridge analytic as nefarious actions
37:47to manipulate the public quote or
37:50question are these strategies and use of
37:53data is something new in the political
37:55world and how about advertising
37:57generally now thank you mr. mr. chairman
38:01know this this is not new in the
38:03political world it's been around for a
38:04long time it feels new to people because
38:06it's never gotten into the public press
38:08before at least not in this kind of a
38:10volume so it's it's understandable that
38:13people feel that this has violated a
38:15norm but it's been a norm for a long
38:17time okay
38:18I'll reserve my time and go to senator
38:21Feinstein thanks mr. chairman mr. Wylie
38:27in February 2018 special counsel Muller
38:32indicted 13 Russian national
38:34and three companies for their part in a
38:37well-funded coordinated campaign of
38:40information warfare using social media
38:43this information warfare campaign
38:46spearheaded through the russian-backed
38:49internet research agency started as
38:52early as 2014 what can you tell us about
38:57possible connections between the SCL
39:01group or Cambridge analytic and Russia
39:05thanks for your questions so one of my
39:09concerns is the the level of engagement
39:13that the company had with the company
39:16being Cambridge analytic out with Luke
39:18Oil and executives from Luke Oil which
39:20is Russia's second largest oil company
39:23the the firm Cambridge analytic I made
39:27presentations and sent documents to Luke
39:29oil that made reference to its
39:32experience in disinformation that made
39:34reference to zzyx its experience in
39:36rumor campaigns attitudinal inoculation
39:40Alexander Nicks emailed me to say that
39:44he passed on a white paper to the CEO of
39:48Luke oil that white paper discussed the
39:53data assets that the company had in the
39:55United States and the strategies that it
39:57was employing the lead researcher that
40:01Cambridge analytical used to harvest the
40:04Facebook data dr. Kogan was also working
40:06on projects in Russia on psychological
40:09profiling at the University of st.
40:11Petersburg the the company had also
40:15engaged contractors who had previously
40:21worked in Eastern Europe for pro-russian
40:25parties and and indeed the company
40:29decided to test Americans views on the
40:33leadership style of Vladimir Putin and
40:36Americans views on Eastern European
40:39issues relating to Russian expansionism
40:42in the region to be clear the only
40:45foreign leader that was tested when I
40:47was there was
40:48Damir Putin and and the the bulk of the
40:52foreign issues that were being tested
40:53focused on on Russian expansionism in
40:552014 one of the concerns that I have is
40:59that there was a lot of contact with
41:05Russia both Russian companies and then
41:07also via dr. Kunze research
41:08presentations that were being made in
41:10Russia that made that made it known that
41:12this research was being done so what I
41:14can't I can't say definitively that this
41:16had any relation to the internet
41:18research agency for example but what I
41:20can say is that a lot of noise is being
41:23made to to companies and individuals who
41:26are connected to the Russian government
41:28and and for me that is of substantial
41:31concern do you think it's possible or
41:37even likely that the Facebook data
41:40harvested by Cambridge analytic ended up
41:43in Russia what I can say is that the
41:47lead researcher dr. Kogan who was
41:50managing the Facebook harvesting project
41:53for Cambridge analytic I was at the time
41:56working on projects that related to
41:59psychological profiling in Russia with a
42:01Russian team as that was going on I also
42:03know that he was traveling to Russia I
42:05also know based on conversations that I
42:07had with him at the time that he was
42:09making it known to his colleagues in
42:11Russia about the project and so I can't
42:15I can't say definitively one way or the
42:17other if if these datasets did end up in
42:20Russia but what I can say is that it
42:22would have been very easy to facilitate
42:24that you told the UK House of Commons
42:27that Cambridge analytic a pitched the
42:30Russian oil company Lukoil and its
42:33services you have also said that
42:36Alexander Nicks gave Luke oil a white
42:38paper you wrote that explained Cambridge
42:41analytic as data collection and online
42:44targeting of Americans lukoil is on the
42:48united states sanctions list and is said
42:50to be tied to the Russian Federal
42:53Security Service they also known as the
42:56former KGB when and where were these
42:59meetings taking place
43:02in London in the United Kingdom and and
43:05also via via the phone
43:07what did Cambridge analytic Attell
43:10Lukoil about its data on Americans did
43:14it share any of that data or is it
43:17possible Russia acquired any of its data
43:20on Americans in sending in sending the
43:25white paper and discussions that I had
43:28with Alexander Knicks about what he was
43:29speaking to the company about I know
43:32that the scale the scale of the of the
43:36data and the location of the data was
43:38made known and also that dr. Kogan was
43:42involved in in that day in that data
43:45collection project which and the concern
43:48that I have is that if if you were
43:51intending on acquiring the data even if
43:54you were not intending to acquire the
43:55data with the willing participation at
43:57Cambridge analytics what was made known
43:59was that this kind of this data could
44:01have been easily acquired by something
44:03as simple as a key logger on dr. Cohen's
44:05computer when he was visiting Russia
44:07thank you my time is up
44:09senator Kennedy Thank You mr. chairman
44:13professor Hersh I don't really have a
44:17question but if what you I'll come back
44:19to you have time if what you're saying
44:21is that people in America are not
44:25persuadable or persuaded by advertising
44:29I think that's rubbish I think some
44:34really smart people spent two hundred
44:36billion dollars two hundred and six
44:38billion dollars last year on advertising
44:39and I hear kids all the time walk around
44:44saying dilly dilly they didn't just
44:46dream that up okay but mr. Wiley let me
44:51ask you a couple questions I'm not
44:54interested in innuendo or speculation or
44:58rumor other than facebook lists for me
45:02the sources of all of the data that you
45:05know Cambridge analytically used while
45:08you work for there was several different
45:13consumer
45:15data vendors that were used can you name
45:18them please I'd like you be specific I
45:20believe I believe it
45:24Experion data was used that Believix
45:27axiom data was used I'm not sure but I
45:32can get back to you on whether contracts
45:34work directly with axiom and Experian or
45:36reseller the state state voter roles and
45:42registration
45:43who else there were smaller firms which
45:46I can't give you just off the top of my
45:48head who had more specialize in each
45:50data and then in terms of online data
45:53there were experiments being done on
45:57collecting other social media data so
45:59for example Twitter that was also used
46:02but the the basis of the modeling that
46:05was being done at the time that I was
46:07there was primarily using Facebook data
46:09okay
46:10did Cambridge a tablet let up put
46:13Facebook aside for a second the Justice
46:17Department FBI get to the bottom of that
46:19did did Cambridge on Avantika obtained
46:23any of this information and lawfully
46:25while you were there I'm not a lawyer in
46:31the United States so I couldn't comment
46:32on whether it was lawful or unlawful
46:35well did they do it improperly I mean
46:38you've been making normative judgments
46:41quite often I you know don't get
46:44religion now on me so so the the faith
46:48the Facebook data I believe was other
46:52than Facebook other than Facebook the
46:55the consumer data lists that were used
46:59were acquired via contracts that were
47:02signed and painted it did Cambridge
47:04analytic ahac anybody while you were
47:08there did you know of the III I have
47:11seen documents that make reference to
47:14special intelligence services and
47:17information gathering networks do you
47:19have copies of those documents not with
47:21me no okay
47:22did Cambridge analytic could get money
47:24from WikiLeaks or get to information
47:27from weaken leaks not when I was there
47:28now
47:29okay do you know if they got it after
47:32you were there in terms of money I don't
47:35know you know that money I misspoke a
47:37data I'm still worked up over Delhi
47:40Delhi I'm I I I wasn't there when the
47:46when the request was made while you were
47:48there scuse me finn erupted and i'm not
47:50trying to be rude but but we have them
47:52into mount of time sure
47:53while you were in cambridge analytical
47:55listen even give me the names of all its
47:57clients boy there was issue or candidate
48:01so at risk of speaking i am happy to
48:07give you a complete list of the clients
48:10that that were being used at the time
48:13okay after well there were i know that
48:17there were packs in various candidates
48:19that were being like who I mean what
48:21can't what issues in cannabis so there
48:25was a network of packs that I believe
48:28were primarily financed by Robert Mercer
48:32that were and then several senatorial
48:37and congressional candidates I believe
48:38any issues there were issue campaigns so
48:43for example John Bolton's pack was if
48:45you consider that an issue based
48:46campaign then that was an issue based
48:48campaign did Cambridge analytical work
48:50for Russia or anybody connected with
48:52Russia while you were there work the we
48:56did not have a Russian client at the
48:58time that I was there
48:59okay I yield back my 9 seconds thanks
49:07Thank You mr. Wallace
49:08Senator Leahy and looking back to the
49:18earlier 2014 under the leadership of
49:22Steve Bannon Cambridge analytic I
49:25reportedly began testing slogans like
49:27build the wall deep state drain the
49:32swamp is also message testing images and
49:36policies of Russian President Vladimir
49:39Putin
49:42now then later on of course her dad
49:46built the wall deep state and drain the
49:47swamp in the Trump campaign but they
49:50were telling this they said before there
49:53was any Trump campaign is that correct
49:56yes
49:57so the the company was testing as you as
50:01you said slogans like during the swamp
50:04images of walls paranoia about the deep
50:07state in 2014 before the Trump campaign
50:11was in existence the the company learns
50:19that there were segments of the
50:21population that responded to messages
50:25like drain the swamp or or images of
50:29walls or indeed paranoia about the deep
50:32state that weren't necessarily always
50:35reflected in mainstream polling or
50:37mainstream political discourse that that
50:40Steve Bannon was interested in in using
50:42to build his movement actually have any
50:50employed staff it was a front group for
50:53the uk-based SCL group they had Rebecca
50:58Mercer Steve band among others on the
51:00board of directors client work was
51:05handled by SEL Mike Wright for all of
51:07that yes not Americans from working on
51:14US campaigns to protect our elections
51:18from foreign interference a 2014 memo
51:24from Cambridge analyticals outside legal
51:27counsel Beck emergency Bennett made
51:31these restrictions very clear did
51:35Cambridge analytic up all of these legal
51:38requirements his work on American
51:40campaigns yeah so so I'm the the source
51:43that provided the media with thought
51:44that memo which I saw at the tail ends
51:47of my engagement at Cambridge analytic
51:49era D to my understanding that memo was
51:53disregarded because
51:54Alexander Nix continued to be the CEO
51:56and they continue to send people to the
51:58United States who weren't American
52:00citizens one thing that I would just say
52:02is that many of the people who were
52:04indeed sent to the United States who
52:05weren't American citizens were not privy
52:07to that memo and were not made aware
52:09that that potentially there were
52:11violations of US law discusses been used
52:20an election advertising for sometimes
52:24they know somebody lives in particular
52:26state they could target ads that reflect
52:31the interests of that state
52:33not all Cambridge or Cambridge
52:37analytical and Craig but if I'm wrong in
52:41this obtained the unauthorized Facebook
52:44data of 87 million people and then
52:48target them was manipulative
52:51disinformation that is that a correct
52:53statement that wasn't everything that
52:56they did but yes that is something that
52:57they did do how does traditional online
53:04marketing compare with how Cambridge and
53:08lilyc obtained individuals information
53:11how they used it so when you are looking
53:16at traditional marketing traditional
53:19marketing first of all doesn't
53:21misappropriate tens of millions of
53:23people's data if they're performing
53:25their duties legally and it is not it is
53:30not or should not be targeted at
53:32people's mental vulnerabilities such as
53:35neuroticism or paranoia or indeed racial
53:38biases traditional marketing does not
53:40exacerbate people's innate prejudices
53:45and and mmin coerce them and make them
53:50believe things that aren't necessarily
53:52true full service propaganda machine the
53:57discouraged voters were more prone to
53:59voting for democratic or liberal
54:02candidates and some of their US clients
54:06extremely close races why did Cambridge
54:11analytical's American clients investors
54:14like the Mercer family invested twenty
54:16million dollars believe these my tactics
54:19might be helpful so Steve Bannon is a
54:23follower of something called the
54:25Breitbart doctrine which which posits
54:27that politics is downstream from culture
54:30so if you want to have any lasting or
54:32enduring changes in politics you have to
54:34focus on the culture and when when Steve
54:38Benin uses the term culture war he uses
54:41that term pointedly and they were
54:43seeking out companies that could build
54:48an arsenal of informational weapons to
54:50fight that war which is why they went to
54:52a British military contractor which
54:55specialized in information operations
54:56senator Lee especially in Russian state
55:06security services let me announce to
55:09everybody that record stays open for a
55:12week and then a few folks get written
55:15questions for people that are here like
55:17Senator Leahy or people that don't come
55:19we'd appreciate if you'd answer them and
55:22get them back to us as soon as you can
55:25senator Lee Thank You mr. chairman mr.
55:28Wiley I'd like to start with you if
55:29that's okay mr. Wiley why you fault
55:33Cambridge analytic Oh for using its data
55:36that it obtained you say without
55:39authorization its vote or targeting
55:41efforts at the same time you took that
55:44same data with you upon leaving the
55:47company isn't that right so so because
55:51there were no staff at Cambridge
55:53analytic ah most people were contractors
55:56or indeed had companies so I received a
56:00copy of the data okay so you had that
56:03data and then you started your own
56:04company
56:04no I my company was in existence before
56:08I left Cambridge and okay okay
56:10because I was a contractor but after
56:12leaving the company you you had a
56:15meeting or perhaps a series of meetings
56:17with a major campaign to do
56:20discuss micro-targeting techniques in
56:22short yeah that's not that's not true I
56:25didn't I didn't meet another campaign to
56:28discuss that if you're in reference to
56:31that data set well it appears that you
56:34tried to use some of the same market as
56:37your former company you were gonna use
56:39that data for something right we're just
56:41going to leave it idle well to be clear
56:45the data was was never used on any
56:48commercial so why did you take it with
56:50you when it's not that I took the data
56:54with me it's just that it was still in
56:56existence at the time that I left okay
56:58what I didn't to be clear I didn't take
57:00any data from camera gentlemen okay
57:03you didn't take it in the sense that it
57:06was already with you yes okay
57:09what type of work did you anticipate
57:11that your company would perform after
57:12you left Cambridge and I let it go
57:14so I I work mostly in data analytics
57:18looking at different kinds of social
57:21trends so after after I left Cambridge
57:25analytic I you know I continued working
57:28on independent projects but to be clear
57:31I didn't use that data on any commercial
57:38contract but couldn't that data have
57:40proven useful in some of your work it
57:43could have but I didn't I didn't go on
57:45and sell it right but you could happen
57:48have you been successful with that and
57:49had you gone on and your business become
57:51successful
57:52couldn't you have now been at the
57:53receiving end of some of the same
57:55questions now going to Cambridge
57:56analytical yes but it didn't happen
57:58understood want to go over some
58:01statements that you've made in paragraph
58:0415 of your testimony you say and I quote
58:08when I was at SCL and seeing I made it
58:12was made aware of the firm's black ops
58:14capacity which I understood to include
58:16using hackers to break into computer
58:19systems to acquire comprar mat or other
58:21intelligence for its clients the firm
58:23referred to these operations as special
58:25intelligence services or special IT
58:27services how did you learn about these
58:29blackops capacities Alexander Nicks told
58:33me
58:33and who was a practical matter was
58:36involved in these black ops my
58:40understanding was that in in some of the
58:44projects that SEL group pants in
58:49different parts of the world you know
58:52misappropriated information was used as
58:56as compromised in in elections in
58:59particular against opposition candidates
59:02okay
59:04in paragraph 30 if you say quote the
59:09Russian project undertaken by doctor
59:12Corrigan had a particular focus on the
59:14dark triad traits of narcissism
59:17Machiavellianism and psychopathy the
59:21Russian project also conducted behavior
59:23research on online trolling how did you
59:27learn all of this and can you describe
59:29those projects in more detail so at the
59:32time dr. Kogan told me about it some of
59:35the research that he was doing he also
59:37told the company about some of the
59:38research that was was being done by that
59:42team in Russia so there's email
59:44correspondence from the firm that for
59:47example referenced the quote work that
59:50he was doing for the Russians so it was
59:54initially through conversations that I
59:57had with dr. Kogan and then later
59:59through the investigative reporting that
1:00:02is being done for the past year with the
1:00:04Guardian and others more details have
1:00:07emerged as well okay professor Hershey
1:00:10the use of social media to micro-target
1:00:13is a fairly new practice but it's my
1:00:15understanding that micro targeting
1:00:18itself is not and sadly the use of
1:00:22provocative information to either divide
1:00:24the electorate or to mobilize portions
1:00:26of the electorate has a long history in
1:00:28our country's political campaigns it was
1:00:31the use of social media to micro-target
1:00:33different than what it's been done how
1:00:37it's been done in the past so there's a
1:00:39lot we don't know we don't know about
1:00:40its effectiveness in some ways but it
1:00:42often looks the same and also responding
1:00:44to senator Kennedy you know just because
1:00:47campaign spend a lot of money on a
1:00:48particular ad kind doesn't mean it works
1:00:51for a long time campaigns are spending
1:00:52money on robocalls countless experiments
1:00:54have shown that robo calls do nothing in
1:00:57an environment where there's lots of
1:00:59stimuli a lot going on in the campaign a
1:01:01lot of campaign ads don't really work
1:01:03probably nobody in this room or nobody
1:01:05that anyone in this room knows changed
1:01:07their mind as a result of any campaign
1:01:09ad in the 2016 election now for someone
1:01:11who's a director of research for
1:01:13Cambridge analytical they should know
1:01:15I'd be shocking if they didn't know the
1:01:16actual effect an estimate of which
1:01:19campaign ads do what part of my
1:01:22skepticism comes because given that
1:01:24there's been whistleblowing there's been
1:01:25no presentation of any such evidence in
1:01:28randomized controlled environments in
1:01:29which someone shows here is the effect
1:01:32this ad targeting neuroticism or
1:01:35whatever had some effect so again it's
1:01:38very hard to sort this out as technology
1:01:40changes from two thousand two thousand
1:01:42four eight twelve sixteen lots of things
1:01:44are new but one point we often come back
1:01:47to is that in the presidential election
1:01:49particularly when there's so much going
1:01:50on the effect of one ad or one kind of
1:01:53ad or one robocall is usually zero okay
1:01:56that's Senator Whitehouse thank you
1:01:59hello mr. Wylie you've said that
1:02:01Cambridge analytic and the SCL group are
1:02:04effectively the same thing and that
1:02:05Cambridge analytic who was the
1:02:07front-facing company for SDL's American
1:02:10operations is that correct yes
1:02:12what is SCL elections so there is a
1:02:17group company in the UK or was called
1:02:21SEL group which had several different
1:02:23divisions the largest division when I
1:02:26first joined was defense so SEL defense
1:02:30SEL Elections was one of the other
1:02:32divisions SEL commercial etc and they
1:02:34all handled different markets for the
1:02:37company so I see all elections handled
1:02:39political what is what are SEL Canada
1:02:43and aggregate IQ those were
1:02:47subcontractors that were set up during
1:02:50the time that I was there
1:02:52to build out a software infrastructure
1:02:55they played a very significant role in
1:02:57building the the actual infrastructure
1:02:59of RIP on the the software product
1:03:01upon our they essentially the same
1:03:02entity SEL Canada and aggregate IQ you
1:03:05could you could think of them like a
1:03:06franchise and the Ripon program was the
1:03:10program that developed the software to
1:03:11use the Facebook data yes once you have
1:03:16algorithms and a target a set of targets
1:03:21you need something to then actually
1:03:22connect those targets with an online
1:03:24Display Network so that that's part of
1:03:26the role that rapun played what is
1:03:28global science research global science
1:03:30research was the company that was set up
1:03:33by dr. Kogan and you've said that it
1:03:36became a company simply to service
1:03:38Cambridge and a let it go correct it it
1:03:40became a company as as I understand it
1:03:44it became a company so that it could
1:03:45sign a contract with Cambridge Anglet
1:03:47okay or rather technically SEO is it
1:03:50fair to describe the entities that I
1:03:53have all just described as a coordinated
1:03:56network yes and what was the role of
1:04:00Robert Mercer and funding that
1:04:01coordinated network he was the primary
1:04:04funder who put in tens of millions of US
1:04:07dollars into Cambridge analytic which
1:04:09then distributed that money to that
1:04:11network did that Cambridge analytic
1:04:13network including SCL have a recurring
1:04:17contracting relationship with black cube
1:04:22when I was there we did not have a
1:04:25contract with black cube have you since
1:04:27become aware of a connection between a
1:04:30CL group and black cube and working
1:04:33together on projects I've become aware
1:04:36of relationships that the company had
1:04:40with former members of Israeli security
1:04:44services how about ASI data science
1:04:47you've said that that company CTO worked
1:04:49on SCL projects that it was a
1:04:52subcontractor to Cambridge analytic and
1:04:55that there was a revolving cast of data
1:04:57scientists between that organization and
1:04:59Cambridge analytic ah yes there are a
1:05:02frequent contractor when I was there I
1:05:06believe their role was as a supplier of
1:05:09as the company was growing there was a
1:05:11an increased demand for
1:05:13more and more data scientists and they
1:05:15were I believe the contract that they
1:05:18had was to provide data scientists as
1:05:21your company consider them part of the
1:05:25Cambridge analytical network we
1:05:26described um if if your definition of
1:05:29network is anybody who has an ongoing
1:05:31relationship then sure you said that the
1:05:34company Palantir had staff that were
1:05:37working on the data at Cambridge
1:05:39analytic
1:05:40that they were meetings with Palantir at
1:05:42Palantir offices and that Palantir staff
1:05:44helped build the models for the Rippon
1:05:47program is that all correct yes although
1:05:51count to clarify Palantir said that all
1:05:54of the work that was being done by
1:05:55Palantir staff was done in a personal
1:05:57capacity so let's just take I've got a
1:06:00minute left let's just take a quick look
1:06:01at
1:06:02brexit some of the forces behind brexit
1:06:05were vote leave be leave and veterans
1:06:08for Britain correct yes all of them had
1:06:11contracts with aggregate IQ yes
1:06:13is there anything peculiar about their
1:06:15contracts with aggregate IQ it's highly
1:06:19suggestive of coordination and and data
1:06:22sharing but that is currently being
1:06:24investigated by the Electoral Commission
1:06:26in the UK because a grid' IQ would have
1:06:27been really hard to find at the time it
1:06:29didn't even have a website it didn't it
1:06:31didn't have a website now and do we know
1:06:32that vote leave actually funneled money
1:06:34into believe and into veterans for
1:06:37Britain that then went to fund aggregate
1:06:38IQ contracts yes we do do we also have
1:06:43connections between Levy you and you Kip
1:06:46and Eldon insurance yes related to a IQs
1:06:52Briggs efforts our not to a IQs Brescia
1:06:56efforts but rather that that side of the
1:06:59leave campaign engaged with Cambridge
1:07:00analytic at Hema general it occurred
1:07:02directly okay well my time has expired
1:07:04I will if there's a second round I'd
1:07:06love to have one senator corner and then
1:07:08then it'll be senator klobuchar I'm
1:07:10going to step out for a few minutes so
1:07:12at the end of his five minutes you take
1:07:13over mr. widely let me start with you
1:07:17please the sort of data mining that
1:07:20you've been describing in the targeting
1:07:23of messages
1:07:26multiple applications correct it could
1:07:28be a commercial application for example
1:07:31and buy something on amazon or navy or
1:07:35netflix they can send me information
1:07:37about something else I might like it
1:07:41could be used to persuade people in a
1:07:43political campaign for against a
1:07:45candidate or for against a particular
1:07:46issue and it's also can be used for
1:07:50covert information operations by
1:07:53governments correct all of that is
1:07:56correct yes data is like any kind of
1:07:59tool you can use it for various various
1:08:01means some very beneficial and legal and
1:08:04others not right so did SC l or
1:08:06Cambridge analytic serve all comers in
1:08:10other words were you open for business
1:08:11to whoever wanted to purchase the
1:08:13services or was that they that well that
1:08:16was the that was the impression that I
1:08:19got Alexander Nicks was quite keen on
1:08:22selling contracts
1:08:24however after Robert Mercer put his
1:08:28investment into Cambridge analytic I do
1:08:30know that we weren't the only
1:08:32restriction that we had was to not work
1:08:34with Democrats okay okay well like for
1:08:37example fusion GPS much in the news and
1:08:41they provided opposition research to the
1:08:44DNC and the against the Trump campaign
1:08:48the the data itself in the the means by
1:08:52which you analyze it and use it is
1:08:55pretty much agnostic correct in other
1:08:57words you can use it for or against a
1:08:59political candidate product or for an
1:09:03information campaign like we said yes
1:09:05okay so mr. Zuckerberg when he was here
1:09:09the other day I he kept saying we don't
1:09:13sell data and I responded to him well
1:09:18you clearly Rin it I don't know whether
1:09:21that's a fair characterization or not
1:09:23how would you characterize the social
1:09:27media platforms like Facebook use of
1:09:29personal data they say they don't sell
1:09:33it
1:09:33how would you characterize it they've
1:09:36created a platform that that encouraged
1:09:39the use of data so it's true that you
1:09:43can't go to Facebook and simply buy
1:09:45Facebook's data but they make it readily
1:09:48available to its customers via its
1:09:51network of applications or the fact that
1:09:56the layouts of people's profiles on
1:09:59Facebook makes it very conducive to
1:10:01scraping data for example although
1:10:03Facebook would say that they don't allow
1:10:05that they still create an asset up which
1:10:09which catalyzes its misuse in my view
1:10:14well mr. Zuckerberg also said that the
1:10:17Terms of Service that that consumers
1:10:21agreed to when they sign on to Facebook
1:10:24he said people probably don't read it or
1:10:27if they read it they don't really
1:10:29understand it is your impression that
1:10:31most of the public has no idea about
1:10:36what they are sharing with these these
1:10:39companies when you even talk to lawyers
1:10:42who read through the terms and
1:10:43conditions some of it's even dense for a
1:10:45lawyer
1:10:46so IIIi think that it's unreasonable to
1:10:49expect a regular ordinary person to to
1:10:54have the burden of understanding dense
1:10:56legal text and the other thing that I
1:10:58would say is that you know social media
1:11:01is not really a choice for most people
1:11:03the Internet is not really a choice for
1:11:05most people it is very difficult to be a
1:11:07functioning member of the workforce or
1:11:09society and refused to use the Internet
1:11:11I don't know a job that would let you go
1:11:13in and refuse to use Google for example
1:11:15I don't know
1:11:16you know most most hiring requires a
1:11:18LinkedIn profile now so so although
1:11:21although we use this narrative of choice
1:11:24because someone's pushing a button even
1:11:26if they had read the terms and
1:11:28conditions and understood it they
1:11:29substantively don't really have a choice
1:11:31because in the modern workforce you have
1:11:33to use social media and you have to use
1:11:34the internet I don't really have other
1:11:35options I don't know a job that would
1:11:38hire somebody who refuses to use the
1:11:39internet so do you think it's too much
1:11:41to expect that these social media
1:11:44platforms get consumers informed consent
1:11:47I mean there is this idea in the law for
1:11:50example if you're going to consent to a
1:11:51surgical procedure
1:11:53by your doctor that the doctor must that
1:11:55your consent must be informed in other
1:11:57words you have to understand what you're
1:12:00agreeing to do you think that's too much
1:12:01to ask for in this context it's not that
1:12:04it's too much is not too much to ask for
1:12:06people absolutely should have informed
1:12:08consent but the the analogy is not is
1:12:10not equivalent when you when you go and
1:12:13see a doctor and you need a surgery you
1:12:15need something and that that you can
1:12:18sending to that surgery is proportionate
1:12:19to the to the the benefit that you're
1:12:21getting when somebody when somebody
1:12:23consents quote-unquote to something
1:12:26online even if they understand that
1:12:28they're consenting to their data come
1:12:31you know being harvested if if that's
1:12:33the only way that you can get a job it's
1:12:34not really genuinely a fair situation
1:12:36and so the the point that I that I would
1:12:39make to you is that it takes a step back
1:12:43from this narrative of consent and start
1:12:44to look at the fact that people don't
1:12:46have a choice they have to use a lot of
1:12:48these platforms to be functional in
1:12:49society in the workforce well I don't
1:12:51have to use Facebook I can use Twitter
1:12:53or I can use but but all of these
1:12:55platforms do the same thing right they
1:12:58all they all as soon as you sign up to
1:13:01Twitter as soon as you sign up to Google
1:13:02as soon as you sign up to anything they
1:13:05all will be collecting a hugely
1:13:07disproportionate amount of data compared
1:13:10to the utility that they provide to you
1:13:11and further the other thing that I would
1:13:13say is that when you're looking at
1:13:14technology you know there's a question
1:13:17of not just informed consent in the
1:13:18present but reasonable expectations in
1:13:20the future when I first signed up to
1:13:22Facebook it didn't have facial
1:13:23recognition I posted all of my photos
1:13:25and then Facebook developed facial
1:13:27recognition algorithms that then can go
1:13:30and search the internet and find other
1:13:31things that I'm doing right so so it's
1:13:34not a question of just informed consent
1:13:36it's about is it proportionate to the
1:13:38benefit that the consumer is getting is
1:13:39it reasonably expected in the future if
1:13:41there's future developments in that
1:13:42technology that that the consumer did
1:13:44actually consent to that at the time and
1:13:47more broadly this this narrative of
1:13:49consent is slightly problematic in the
1:13:51sense that when people have to use these
1:13:54platforms it doesn't matter whether or
1:13:56not they understand if they say have if
1:13:58they have to use it to get a job they
1:14:00will still use it and so we are we are
1:14:02sort of coercing and compelling people
1:14:03to hand over a lot of information which
1:14:05which
1:14:06not only now could be dangerous but you
1:14:08have to imagine in the future also what
1:14:10what the developments of technology you
1:14:13know what developments will see moving
1:14:15forward and what kinds of risks will be
1:14:17exposing to people ten years from now
1:14:18when that data about them still exists
1:14:20my my time is up I would just conclude
1:14:22by saying that companies now can and do
1:14:26market their services and products based
1:14:28upon people's expectations of privacy
1:14:31and so if consumers are fully informed
1:14:34about what they're doing what the
1:14:36consequences are they can make their
1:14:38choices that may create markets for
1:14:40other alternative platforms of where
1:14:42people's data will be more protected
1:14:47okay
1:14:50thank you very much mr. chairman thank
1:14:52you for your succinct answers mr. Wiley
1:14:54and they're to the point
1:14:56last month senator Kennedy and I
1:14:57introduced a bill to protect consumers
1:15:00privacy online and included in that
1:15:05Bill's giving consumers the right to
1:15:07control their own data by allowing
1:15:09people to opt out of having their data
1:15:11collected and requiring companies to
1:15:13notify consumers of a privacy violation
1:15:15within 72 hours you've previously
1:15:18expressed support for allowing users to
1:15:20opt-out of all targeting criterion three
1:15:23clicks or less as well as rules to
1:15:25require the data collected by each app
1:15:27be proportional to the apps actual
1:15:30purpose would Cambridge analytic have
1:15:32been able to harvest the data of
1:15:33Facebook users and their friends if the
1:15:35users had opted out of having their data
1:15:38attracted by Facebook it would well the
1:15:43the the the the the problem was that
1:15:45Facebook had had set up applications
1:15:49that that that physically allowed the
1:15:52collection of friends data so one of the
1:15:53other things that I would say is that in
1:15:56addition to giving consumers rights we
1:15:58should be putting obligations on on
1:16:00companies themselves so a principle of
1:16:03privacy by design which treats privacy
1:16:06as an engineering problem as a safety
1:16:08problem would also be incredibly
1:16:10beneficial because privacy is not just a
1:16:14governance issue or a terms issue or a
1:16:15policy issue it's a physical engineering
1:16:17issue when it comes to software
1:16:19and you think that they would just do
1:16:20that on their own or do you think you
1:16:22know good to have some thank you God is
1:16:23my response to one of my friends on the
1:16:26other side of the Allen white center
1:16:28Kennedy and I have proposed this bill
1:16:29secondly the honest ads act a bill that
1:16:32I've done with Senator McCain and
1:16:33Senator Warner
1:16:35I understand you've supported the idea
1:16:38that there should be more transparency
1:16:40of political ads absolutely no there's
1:16:42no requirements in place in your written
1:16:44testimony you stated if a foreign actor
1:16:46drop propaganda leaflets by airplane
1:16:49over Florida or Michigan that would
1:16:51universally be condemned as a hostile
1:16:54act but this is happening online as you
1:16:57know Facebook Twitter Microsoft are now
1:16:59supporting my bill but and they are
1:17:02taking measures to disclose these ads
1:17:06especially Facebook do you think that
1:17:08this patchwork of voluntary measures
1:17:10though will be the answer no I think
1:17:13that just just in the same way that we
1:17:15require safety standards and everything
1:17:16else that we care about we should be
1:17:18requiring safety standards and
1:17:20transparency standards in software and
1:17:22online platforms thank you
1:17:24at last month hearing I asked mr.
1:17:25Zuckerberg if Facebook had determined
1:17:27whether they up to 87 million Facebook
1:17:30users whose data was shared with
1:17:32Cambridge analytic a were concentrated
1:17:34in certain states mr. Zuckerberg said
1:17:37that he would follow up with a
1:17:38state-by-state breakdown of those users
1:17:40I want to ask if you have any knowledge
1:17:42as to whether those Facebook users were
1:17:44mostly in any particular states i I
1:17:48can't say off the top of my head what
1:17:50the density was in each state but I do
1:17:52know that there were particular focused
1:17:54if there was a particular focus on on
1:17:57States from the company's activities on
1:17:59on swing states and states that were
1:18:01winnable by Republicans states like
1:18:04Wisconsin Michigan yes
1:18:06Pennsylvania and I we again await that
1:18:10information from Facebook mr. Wiley I
1:18:13also asked mr. Zuckerberg went whether
1:18:15any of the roughly a hundred 26 million
1:18:17people who may have been shown content
1:18:20from a Facebook page associated with the
1:18:22Russian troll farm the internet research
1:18:24agency were the same users whose data
1:18:27was shared with Cambridge analytic ah he
1:18:29replied that he believes that his quote
1:18:31entirely
1:18:32possible that there will be a connection
1:18:34there what can you say about the
1:18:37potential for any overlap between the
1:18:39Facebook users whose data Cambridge
1:18:41analytic I obtained and the users who
1:18:43were shown content from the internet
1:18:45research agency the thing that I would
1:18:48say so firstly I I didn't ever deal with
1:18:51the internet research agency so I can't
1:18:53comment specifically on on the on that
1:18:55entity but to your point my concern is
1:18:59that information either may have been
1:19:03shared or indeed misappropriated again
1:19:06at the second instance by Russian entity
1:19:08from Cambridge analytic a what I would
1:19:11say though is that it's not manat
1:19:15whether or not these individual records
1:19:17were then targeted because if they were
1:19:20used to build an algorithm whether that
1:19:22algorithm was built I Cambridge
1:19:23analytical or another entity other users
1:19:26who share similar profiles and patterns
1:19:29in their data could have also been
1:19:30exposed you know in a way that if you
1:19:33simply looked at were these records
1:19:36targeted specifically even if the answer
1:19:39is no that doesn't mean that those
1:19:40records weren't used to build a
1:19:41targeting algorithm and get it I'll
1:19:43follow up on on the second round here
1:19:46one last question as you use the word
1:19:48profiling and these voter disengagement
1:19:50as you called it voter suppression one
1:19:52of the most horrifying things from our
1:19:53hearings that I saw were those ads that
1:19:56were clearly made to suppress the votes
1:19:59of African Americans do you have any
1:20:02knowledge about the scope of this
1:20:04activity and how often it occurred out
1:20:07of the company my knowledge relates to
1:20:11the the tail ends of my engagement with
1:20:14Cambridge analytic out one of the one of
1:20:16the things that did provoke me to leave
1:20:20was the beginnings of discussions about
1:20:22voter disengagement I have seen
1:20:25documents that reference photo
1:20:26disengagement and I recall conversations
1:20:29that that it that it was intended to
1:20:32focus on African American kept those
1:20:34documents I will I will discuss with my
1:20:38lawyers a the best way to get you that
1:20:39information okay and just to be clear
1:20:41just as you said it would be illegal an
1:20:43act of illegal act to have an airplane
1:20:46come in with those pamphlets this under
1:20:48American law where you are specifically
1:20:51suppressing votes is also an illegal act
1:20:53so thank you to end to be clear I I
1:20:56didn't I didn't partake myself and and
1:20:59that's why you left in part so I get
1:21:01that and I appreciate you coming to
1:21:03testify very much today thank you thank
1:21:07you actually the chair has asked me to
1:21:08chair the committee and also happen to
1:21:11be and in the next water for I'm not
1:21:13taking chairs prerogative it's my turn
1:21:14so thank you all for being here I I
1:21:18think this is a very helpful hearing I
1:21:20intend to be here for the the second
1:21:21round as well and mr. Wylie I just
1:21:24leaned over to the ranking member and
1:21:27said that she was complimenting you on
1:21:29your technical expertise and I said I
1:21:31love having a witness before the stand
1:21:33that I can actually understand I can't
1:21:35necessarily understand lawyers but I can
1:21:37understand data analytics people because
1:21:38I was one I hope that the result of this
1:21:42hearing is actually trying to figure out
1:21:44what if anything Congress should do from
1:21:46a regulatory framework and compliance
1:21:48framework with reform with respect to
1:21:50the new Avenue for data use you know all
1:21:56of us in these committees have already
1:21:58used data from aggregators we take the
1:22:01voter data we know what voting
1:22:02propensity czar that's downloaded from
1:22:04the boards of election you use that as a
1:22:06basis for targeting voters and then
1:22:09there was the the next wave data
1:22:10aggregator so that you can overlay
1:22:13people's affiliations with associations
1:22:16the magazine subscriptions that's all
1:22:18really become passe in terms of data
1:22:20matching it's been happening for
1:22:21probably ten or twenty years and in our
1:22:24campaigns and I would dare say that
1:22:26every single member here who's gone I
1:22:28ran up for election has had that now
1:22:30that is where they buy data and they
1:22:32aggregate it they built there and then
1:22:34they build their proprietary platforms
1:22:36around it now with the advent of social
1:22:38media we have entities who have come
1:22:42into play that really don't want to sell
1:22:44their data they want to sell the
1:22:45analytic result of that data so that
1:22:47they can target people on certain social
1:22:49media platforms would you agree with
1:22:50that analysis
1:22:51yes we're it's it's the age of access
1:22:56rather than the age of transfer in terms
1:22:58and something just simply that that data
1:23:01that baseline is a part of their their
1:23:03intellectual capital a part of their or
1:23:05their institutional value yes
1:23:09data is an incredibly powerful it's it's
1:23:12like the new oil I'd be happy for either
1:23:15of the other two witnesses to to chime
1:23:19in a part of when mr. Zuckerberg was
1:23:22before us just some weeks back I tried
1:23:27to focus on what we should all be
1:23:29looking at as policymakers as other
1:23:32practices beyond the cameras analytic oh
1:23:34and I should say that my firm engaged
1:23:36Cambridge analytical I met them the day
1:23:39that they proposed it and I saw him
1:23:41again on election day I do have some
1:23:43questions about the focus on trying to
1:23:47get cyclo graphic data from people
1:23:49through in-person interviews it may or
1:23:51may not have been captured through some
1:23:53of the techniques that we're discussing
1:23:54today but I also want to go back and
1:23:57talk about if we're going to do a
1:23:58thorough impartial nonpartisan review of
1:24:04the facts that we really should go back
1:24:06probably over about the last 10 years if
1:24:09I don't know if you've had an
1:24:10opportunity to but I would commend it to
1:24:11you to read in the MIT Technology Review
1:24:14they had a three-part series and the
1:24:16series was titled how Obama's team used
1:24:19big data to rally voters and if you go
1:24:22through that that very well-written
1:24:25review and through articles associated
1:24:28with it there were quotes from campaign
1:24:31workers who said we literally took the
1:24:34whole social graph from Facebook and
1:24:36we're able to download it through the
1:24:38use of an application they there's a
1:24:39terminal term in here I don't know their
1:24:41specific page but they talk about
1:24:42gamifying the apps that campaign workers
1:24:45would use and possibly gamifying the
1:24:48apps that would be supporters would use
1:24:51and in the Obama campaign I think it may
1:24:54have been the first one could have been
1:24:55the second they actually asked whether
1:24:58or not if you were a Facebook user if
1:25:00you would mind by clicking on an on a
1:25:02button be willing to share information
1:25:05about all your friends in my case I have
1:25:08about forty nine hundred friends so by
1:25:10clicking that button I was actually
1:25:12giving act
1:25:13to thousands of people's information
1:25:15without their knowledge and this was a
1:25:18document in practice that I think also
1:25:20has to be looked at in the context of
1:25:23creating good policy that's based on
1:25:27various uses contemporary unis uses of
1:25:30data on social media platforms and so in
1:25:34my remaining time in this and the the
1:25:37first round I would just ask you do you
1:25:43believe that some of the technology
1:25:46players today simply have grown so large
1:25:49so quickly
1:25:51that some of what you thought would have
1:25:52been captured through just good
1:25:54corporate governance and code of conduct
1:25:56having somebody from a social media
1:25:58platform who either leans conservative
1:26:01Arlene's liberal kind of putting their
1:26:05thumb on the scale and giving them
1:26:06people information that they really
1:26:08shouldn't if they were good stewards of
1:26:10the data and the social media platform
1:26:12that they were on and I'll leave that as
1:26:13an open question anybody on the panel
1:26:15and reserve my remainder for the next
1:26:17round I just say that you know I I don't
1:26:22hold the view that Facebook is somehow
1:26:23necessary for our lives as a instructor
1:26:26of political science I encourage my
1:26:28students never to share information
1:26:29about politics or ads or news online
1:26:32because they're not professional editors
1:26:35they should direct their attention to
1:26:37newer sources that have editors the idea
1:26:39that you must contact your friends and
1:26:41family and upload pictures and that
1:26:43there is no alternative is really
1:26:45something that doesn't strike me not as
1:26:46an expert but as an ordinary citizen as
1:26:48something that's that's true and so I
1:26:50think in answering your question
1:26:52Facebook has a lot of data I think it's
1:26:53acted really inappropriately in terms of
1:26:56how its conveyed news sold ads to just
1:26:59about anyone who wants to target hate
1:27:00groups fine bring it in we have an
1:27:02algorithm we don't pay too much
1:27:03attention to it and it's not our fault
1:27:05if it goes wrong I think it's a really
1:27:07terrible way that they've conducted
1:27:08business and again I would encourage
1:27:11people to use Facebook but I am
1:27:13surprised by by people's continuing
1:27:15interest in the company I concurred a
1:27:21lot in what he was saying I think it's
1:27:24difficult to argue that that Facebook
1:27:26person
1:27:27is necessary for people's lives when
1:27:29about half the Internet users aren't on
1:27:30it and there are lots of other types of
1:27:33social media to your point about would
1:27:36someone with a particular ideological
1:27:39political view tilt the scales one way
1:27:41or another I don't know if people would
1:27:43deliberately do it they might but by the
1:27:46nature they will because because what I
1:27:48view as true is influenced by how I
1:27:51think ideologically and so things that I
1:27:54might screen out and say well this this
1:27:56is bad this is disruptive to the
1:27:57community would be influenced by how I
1:27:59think thank you senator Coons
1:28:02Thank You senator Tillis and thank you
1:28:05to the panel for the chance to be with
1:28:06you today
1:28:06I think the reason this matters is that
1:28:09we're talking about the intersection of
1:28:12several important developments that are
1:28:14difficult for the average American to
1:28:15understand big data and social media and
1:28:19foreign interference in our 2016
1:28:22election and we're trying to tease out
1:28:24what actually happened or didn't happen
1:28:26mr. wildly let me start with you
1:28:29Cambridge University professor Michael
1:28:31Kaczynski studied the use of Facebook
1:28:33data and found that based on average of
1:28:3668 Facebook Likes by a user these are
1:28:39just likes what you liked or disliked
1:28:41you could predict sexual orientation
1:28:44political party affiliation race with
1:28:4785% accuracy further factors like
1:28:50religious affiliation alcohol or drug
1:28:53use even whether your parents were
1:28:55divorced could be deduced is that your
1:28:57understanding of that analysis and do
1:29:00you think Cambridge analytic oh and the
1:29:03work of Professor Cogan used that
1:29:07predictive power to develop algorithms
1:29:09that then weaponized differences between
1:29:12Americans for electoral game so the
1:29:16basis of the research that we were doing
1:29:18at Cambridge analytic I was from the
1:29:23papers that you're citing from from dr.
1:29:26Kaczynski so so the firm replicated his
1:29:30approach and then sought to improve it
1:29:33and did I correctly summarize the
1:29:36incredibly high correlation that you
1:29:39could show in terms of
1:29:40really knowing the individual Facebook
1:29:42user if you had access to their likes
1:29:45and their friends and their social media
1:29:47activity yeah it it increases the
1:29:50particular paper that you're citing the
1:29:52the number of likes once you surpass a
1:29:55hundred and two hundred you can get to
1:29:57the same level of accuracy at predicting
1:30:01for example personality traits as as
1:30:04your spouse would if they were answering
1:30:07questions about you and in comparison to
1:30:09how you would resolve to a point that
1:30:11Professor Hirsch made earlier all of us
1:30:13who have stood for election have
1:30:15struggled with the difficulty of
1:30:16actually targeting our voters
1:30:18effectively because the datasets we've
1:30:20had access to mostly publicly available
1:30:22data are very thin very narrow there's
1:30:25very little that we have the datasets
1:30:27that Facebook has access to that's why
1:30:30it's a multi-billion dollar company are
1:30:32unbelievably deep and rich and so it's
1:30:35unlike anything we've had to confront
1:30:37before correct yeah III don't I don't
1:30:40can contest what Professor Hirsch was
1:30:42saying in the sense that there are you
1:30:46know it is true persuading somebody
1:30:48compared to motivating them to to turn
1:30:51out as much more difficult and the the
1:30:53datasets that traditionally were used
1:30:56are often very sparse and not
1:30:58necessarily reliable but that's that
1:31:00available now through Facebook are
1:31:02orders number two yes yes
1:31:05and that's why Cambridge analytic ended
1:31:08up pursuing that path because it found
1:31:10that in comparison to traditional
1:31:13marketing datasets the the the data that
1:31:17you could procure from social networking
1:31:18sites was the was much more dense and
1:31:21actually much more reliable to create a
1:31:24precise algorithm and to be clear a
1:31:26billionaire American mega donor and
1:31:28supporter of the Trump campaign funded a
1:31:31shell corporation Cambridge analytic Oh
1:31:33still run by foreign nationals to take
1:31:35advantage of this research this
1:31:37understanding and gained access to
1:31:40eighty seven million Americans Facebook
1:31:43information and developed some of the
1:31:47algorithms that came out of that I want
1:31:48to ask you in the time I have remaining
1:31:50about your experience working with Steve
1:31:52band
1:31:52one of president Trump's senior campaign
1:31:55advisors and the goals that he used
1:31:57cambridge analytical to achieve was one
1:31:59of his goals to suppress voting or
1:32:01discourage certain individuals in the
1:32:03united states from voting that was my
1:32:05understanding yes was voter suppression
1:32:07a service that US clients could request
1:32:09in their contracts with cambridge
1:32:11analytical while Bannon was vice
1:32:12president yes and so steve Bannon is
1:32:17running an organization where you could
1:32:20as a client request in contracts voter
1:32:23suppression using this remarkable data
1:32:26set i don't know if it was referenced in
1:32:28contracts but I have seen documents that
1:32:31make reference to it in relation to
1:32:33client requests and services provided as
1:32:35a last question you testified that back
1:32:38in 2014
1:32:39Cambridge analytic a set up focus groups
1:32:41message testing and polling on Americans
1:32:44views on the leadership of Vladimir
1:32:46Putin and Russia's expansionism in
1:32:48Eastern Europe
1:32:49you've also testified that it's entirely
1:32:52possible would have been relatively easy
1:32:53for Russian intelligence to put a
1:32:56keylogger on professor Cogan's computer
1:32:58and get access to this full data set why
1:33:02do you think Cambridge analytic Oh was
1:33:03testing Putin's aggressive actions and
1:33:06what threat would it pose to our 2018
1:33:09elections if this entire data set all
1:33:12the algorithms that go with it are
1:33:14currently in the hands of Russian
1:33:16intelligence so I I don't have a clear
1:33:19answer as to why the company wanted was
1:33:22so engaged in testing Russian
1:33:25expansionism in the leadership of
1:33:26Vladimir Putin that's a question better
1:33:29put to Steve Benin in terms of the
1:33:33dangers for not just American democracy
1:33:36but you know other democracies around
1:33:38the world this this data is powerful and
1:33:41if it's put into the wrong hands it
1:33:43becomes a weapon and we have to
1:33:46understand that you know companies like
1:33:49Facebook and platforms like Facebook or
1:33:51Twitter are not just social networking
1:33:53sites there are opportunities for
1:33:55information warfare not just by state
1:33:58actors but also non-state actors and so
1:34:00we really do have to look at protecting
1:34:02cyberspace as a national security issue
1:34:05just in the same way
1:34:06that we have agencies to protect our
1:34:08borders you know land sea and air thank
1:34:11you very much mr. Wiley thank you thank
1:34:16you mr. chairman
1:34:17welcome to each of the witnesses thank
1:34:18you for being here you know I think
1:34:21Americans are rightly concerned about
1:34:22privacy and the security of our data and
1:34:25we're also concerned about the power
1:34:28that is being collected in Silicon
1:34:31Valley if a handful of companies having
1:34:32enormous troves of data that they're
1:34:36able to use an employee with very few
1:34:40rules governing what they do with that
1:34:42information much of the media attention
1:34:45in recent weeks and months is focused on
1:34:48the data operation of the Trump campaign
1:34:50but the Trump campaign was hardly the
1:34:52first to employ data in a very
1:34:55significant way in a political campaign
1:34:58dr. Jameson and your written testimony
1:35:00you talked about the Obama campaign in
1:35:032008 and 2012 and and their data
1:35:07operations can can you share with this
1:35:08committee what the Obama campaign did
1:35:10regarding data in O eight and twelve
1:35:13sure during the the 2008 campaign what
1:35:17they did is they had as an advisor a
1:35:20co-founder of Facebook who helped him
1:35:22understand how they could use Facebook
1:35:24and obtain data from Facebook and a lot
1:35:27of that work was done by consultants but
1:35:30that was the center of it
1:35:32in 2012 the campaign changed his
1:35:34strategy and then pulled all of that
1:35:35in-house and made it much much more
1:35:37effective they were able to combine
1:35:38their Facebook data more cleverly more
1:35:41carefully with other sources of data and
1:35:44do a much better job of understanding
1:35:46what individual voters were like who was
1:35:49connecting with whom and how they could
1:35:51understand those conversations and his
1:35:54Facebook approached access to data on a
1:36:00fair and even-handed matter allowing
1:36:03candidates from whichever party the same
1:36:06access to data or have they been more
1:36:08political and partisan players in that
1:36:10regard I have no knowledge of that you
1:36:13know I will note that Carol Davidson who
1:36:16is the director of data integration and
1:36:18media analytics for the Obama for a
1:36:20mirror
1:36:2020:12 she said quote Facebook was
1:36:25surprised that we were able to suck out
1:36:27the whole social graph but they didn't
1:36:30stop us once they realized that was what
1:36:34we were doing and she also said that
1:36:37Facebook quote came to office in the
1:36:41days following election recruiting and
1:36:43they were very candid that they allowed
1:36:46us to do things that they wouldn't have
1:36:48allowed someone else to do because they
1:36:51were on our side so that's the head of
1:36:54data analytics for the Obama campaign
1:36:56saying Facebook is giving preferential
1:36:59treatment to the Obama campaign because
1:37:02that's the that that's the political
1:37:05side they're on
1:37:09did they give the Romney campaign the
1:37:10same access to data in the in that
1:37:12election in 2012 not to my knowledge
1:37:16all right how about in 2016 there's been
1:37:19a lot of discussion of what the Trump
1:37:20campaign did with data did the Hillary
1:37:23Clinton campaign have a date operation I
1:37:26presume so but I'm not familiar with one
1:37:28that's a lot wasn't excuse me they're
1:37:30not was not very much written about her
1:37:33campaign on any of this so does anyone
1:37:36on the panel know what the Hillary
1:37:37Clinton campaign did on the data side
1:37:43does anyone on the panel think that
1:37:45there is a chance in a million years
1:37:47that the Hillary Clinton campaign didn't
1:37:49have a substantial investment in in data
1:37:52analytics mr. Wylie sure um so campaigns
1:37:57across the across the spectrum in the
1:38:00United States use data and and just to
1:38:02to your point you know when we look at
1:38:05you know a lot of people are concerns
1:38:08about for example the ability of big
1:38:10government to inhibit our liberties and
1:38:12choice big data can engineer a situation
1:38:17that limits our choice and our freedom
1:38:19and it's not a partisan issue so to your
1:38:22point about other parties using types of
1:38:27access on Facebook I actually agree with
1:38:32you in the sense that there is a
1:38:34substantial risk of distorting the
1:38:37electoral process if a company like
1:38:39Facebook decides to pick aside whether
1:38:42that is Democrat or Republican and so
1:38:44the thing that I would hope that this
1:38:46committee and others really you know
1:38:49internalize is that this is we're
1:38:52talking about Cambridge analytic ah but
1:38:54it's not a partisan issue we're talking
1:38:57about the future of how these technology
1:39:00companies operate and the risks to
1:39:02ordinary American citizens and the risks
1:39:05to the integrity of our democratic
1:39:07processes here in the United States and
1:39:09around the world and it's that that's
1:39:10not a partisan issue I very much agree
1:39:13with that and I would note that there is
1:39:15an overlay on top of that that that
1:39:17Facebook and other social media
1:39:18companies are now the vehicle through
1:39:21which some 70% of Americans get their
1:39:25political news and so the specter of
1:39:28censorship I I think is a profound
1:39:31threat to Liberty and so I appreciate
1:39:34this panel being here and thank you
1:39:35thank you for your testimony
1:39:41senator Blumenthal thanks mr. chairman
1:39:46welcome to you all thank you for being
1:39:48here mr. Wylie are you aware of
1:39:52conversations between Cambridge
1:39:55analytics executives and any
1:39:58representatives of the Russian
1:40:00government or people associated with the
1:40:01Russian government I'm I'm aware of
1:40:05meetings that the company had with
1:40:07Lukoil which has close connections with
1:40:10with the Russian government those
1:40:12conversations documented anywhere in any
1:40:16letters or emails or any other kinds of
1:40:21evidence there are documents pertaining
1:40:25to the conversations and and
1:40:28presentations made to to lukoil and I
1:40:31have passed on some of those documents
1:40:33to the authorities to this committee
1:40:38know I yet to this committee know would
1:40:41you be willing to provide them to this
1:40:43committee um I will consult my lawyers
1:40:46as to the best the best way to get you
1:40:47that information thank you
1:40:49during your time at Cambridge analytic a
1:40:51my understanding is that you are aware
1:40:56of the founders or funders including
1:41:02Robert Mercer providing monies so that
1:41:09they would not be quote necessarily
1:41:12considered declara Beltaine
1:41:13contributions are you aware of
1:41:16conversations to that effect by mr.
1:41:19Mercer or anyone else what I'm
1:41:22referencing is what was explained to me
1:41:26after I inquired as to the relatively
1:41:31convoluted setup that was happening in
1:41:32the United States with respect to the
1:41:35set up of Cambridge analytic ah what was
1:41:37explained to me is there was an and this
1:41:39is necessary to say that this was the
1:41:41primary reason so it could be an
1:41:43ancillary benefit but that when you
1:41:46invest money as an investor into a
1:41:48company that you own it doesn't
1:41:50necessarily constitute
1:41:53and an electoral contribution which is
1:41:55declara bill let me let me just cut
1:41:57right to the intent of my question was
1:42:02there specific explanation to you that
1:42:05the purpose of structuring these funds
1:42:08as investments was to in effect avoid
1:42:12the reporting requirements or other
1:42:14provisions of the United States election
1:42:17laws it was explained to me as a benefit
1:42:20of the set up
1:42:21were there any firewalls during the time
1:42:26he worked it came back analytical that
1:42:27separated work on different campaigns or
1:42:31were the funds in effect commingled in
1:42:36all the campaign maasai was there I did
1:42:41not see firewalls being set up or or or
1:42:48any sort of barriers between people or
1:42:52or conversations if that's your question
1:42:53so there was no recognition of the legal
1:42:57responsibility to separate campaigns I'm
1:43:01aware of memos from the company's
1:43:05lawyers to some of the executives of the
1:43:08company that outline the
1:43:09responsibilities in the United States to
1:43:11separate contact between staff and
1:43:14activities that relate to different
1:43:16campaigns or PACs etc but those memos
1:43:19were instructions those memos were
1:43:21instructions from or rather advice from
1:43:23lawyers to executives but when I was
1:43:27there I did not see in practice those
1:43:29instructions were not followed No
1:43:33can you provide some specific examples
1:43:36of your direct knowledge of either focus
1:43:42groups or other efforts to suppress
1:43:45voting so in terms of specifics I'm
1:43:52happy to work with the committee to give
1:43:54a more full explanation just I
1:43:56understand there's limits on time but I
1:43:58am aware of of research that was being
1:44:03looked at about what what
1:44:06and indeed what demotivates different
1:44:08types of people and so if you you focus
1:44:11on messaging that demotivates certain
1:44:13types of people you decrease the amount
1:44:15of turn a key dr. Jameson you served as
1:44:21part of the agency landing team in the
1:44:25transition for the Federal
1:44:27Communications Commission for president
1:44:30elect Trump correct yes during that time
1:44:34did you have any contact with Michael
1:44:36Kohn about FCC policy no during the
1:44:39transition did you ever meet him no not
1:44:42to my knowledge
1:44:43thank you thanks mr. chairman senator
1:44:47Harris
1:44:52so we're here today to talk about how
1:44:55Cambridge analytic obtained sensitive
1:44:57data about millions of Americans from
1:44:58facebook without users knowledge or
1:45:00consent and then used that data to
1:45:02target voters and influence our
1:45:04elections and there are broader issues
1:45:06of privacy that are highlighted by this
1:45:09incident and I think it's worth stepping
1:45:11back to pull all this in context for the
1:45:13American public to put it plainly most
1:45:16Americans have entered into a bargain
1:45:18with Facebook and other web service
1:45:20providers in which users unknowingly
1:45:23give those companies huge amounts of
1:45:25personal data in exchange for the free
1:45:28service of social networking in turn
1:45:31Facebook uses this data to show its
1:45:33users carefully targeted ads which are
1:45:36the source of 98% of Facebook's revenue
1:45:39interestingly interestingly enough this
1:45:43business model makes the Facebook user
1:45:45the product and makes the advertisers
1:45:48the customer but let's be clear this
1:45:51arrangement is not always working in the
1:45:53best interests of the American people
1:45:56first users have little to no idea just
1:45:59how Facebook collects their data
1:46:01including tracking the user's location
1:46:04the device they are using their IP
1:46:07address and activities on other websites
1:46:10to be clear this occurs whether or not
1:46:13they are logged into Facebook and
1:46:15whether or not they even use Facebook
1:46:18let me put this in perspective in the
1:46:22real world this would be like someone
1:46:24following you every single day as you
1:46:28walk down the street watching what you
1:46:30do where you go for how long and with
1:46:34whom you're with for most people it
1:46:37would feel like an invasion of privacy
1:46:38and they call the cops
1:46:41and yes social network sites technically
1:46:46lay all this out in their Terms of
1:46:47Service but let's be honest
1:46:50few Americans can decipher or understand
1:46:53what this contract means second as this
1:46:58hearing illustrates Americans do not
1:47:00have real control over the data
1:47:02collected on them and there's almost
1:47:05nothing that users can do
1:47:06once data is shared with third parties I
1:47:09believe it is therefore government's
1:47:11responsibility to help create rules that
1:47:15yield a better and more fair bargain for
1:47:17the American people one that respects
1:47:19their rights as consumers and their
1:47:22privacy in the meantime of a few
1:47:25questions I have for the witnesses today
1:47:26mr. Wiley in particular you've mentioned
1:47:29there's been a lot of discussion about
1:47:30how Canberra Cambridge analytic a
1:47:33targeted african-american voters and
1:47:34discouraged them for participating in
1:47:36elections when Steve Annan was the vice
1:47:39president what specifically did Steve
1:47:42Bannon or anyone else decide motivates
1:47:44or demotivates african-americans to vote
1:47:47so it's it's not just focusing on racial
1:47:53characteristics of people actually when
1:47:55you pull a random sample of African
1:47:57Americans they're not all the same
1:47:59people oftentimes are very different
1:48:01have very different lives and different
1:48:02motivators so when you're looking at any
1:48:06any program whether it is motivating or
1:48:08demotivating someone understanding their
1:48:11internal characteristics is a very
1:48:13powerful thing because you don't treat
1:48:15them just as a black person you treat
1:48:18them as who they are that can be used to
1:48:20encourage people to vote or discourage
1:48:22people to vote so how specifically then
1:48:25did they target African American voters
1:48:29understanding as you do that the African
1:48:31American population is not a monolith
1:48:33how did they then decipher and determine
1:48:36who was African American so they would
1:48:37target them in their intent to suppress
1:48:40the vote so racial characteristics can
1:48:44be modeled and it I'm not sure about the
1:48:49the studies that my colleague here was
1:48:51referencing but we were able to get an a
1:48:54UC score which is a way of measuring
1:48:56accuracy for race that was 0.89 I
1:48:59believe only a you see is what area
1:49:02under the receiving operator operates
1:49:04characteristic it's a way of measuring
1:49:06precision which means it's very high so
1:49:12but to be clear I didn't participate on
1:49:17any
1:49:19voter suppression programs so I can't
1:49:22comment on the specifics of those
1:49:23programs I can I can comment on their
1:49:25existence and I can comment more
1:49:26generally on my understanding of what
1:49:29they were doing but those questions are
1:49:32better placed for or Steve Benning okay
1:49:34and I join in the request for any
1:49:36documents you can share with us that are
1:49:37evidence of the conduct you've described
1:49:39on a different note what should Facebook
1:49:42have done to verify that either you or
1:49:44Cambridge analytic in fact had deleted
1:49:46the unauthorized data so I can't speak
1:49:51for Cambridge analytic I can speak for
1:49:54myself in 2016
1:49:58they sent me a letter that said we're
1:50:01aware that you may still have data from
1:50:04this harvesting program it informed me
1:50:08that dr. Cogan actually didn't have
1:50:10permission to use the application that
1:50:12he was using for commercial purposes and
1:50:13only academic purposes that was actually
1:50:16new information to me I did not know
1:50:17that all the time and then it requested
1:50:20that I if if I still had the data to
1:50:24delete it and then sign a certification
1:50:27of that that I no longer have the data
1:50:32if it require you to sign that under the
1:50:35supervision of a notary or just sign it
1:50:38and send it back it did not require a
1:50:40notary or any sort of legal procedure so
1:50:46I signed the certification and sent it
1:50:49back and they accepted it just as a
1:50:54housekeeping we will have a vote called
1:50:56around noon on send of time we know that
1:51:00means we need to be there within about
1:51:0125 minutes so we I think we'll be able
1:51:04to get the first round we have senators
1:51:06Durbin Orono and Booker and I'm happy to
1:51:08stick around to the end of the vote for
1:51:10anyone if we can agree to three-minute
1:51:12rounds in the remaining time before the
1:51:13votes call senator durbin thanks mr.
1:51:16chairman thanks to the witnesses let me
1:51:18say initially senator Harris
1:51:20I thought your presentation on the issue
1:51:22of privacy was spot-on
1:51:25it really described what we are about at
1:51:28least one of the things we should be
1:51:29about I asked a question of misters
1:51:31burg whether he felt comfortable telling
1:51:34me the name of the hotel he stayed in
1:51:36last night and after a couple minutes or
1:51:39seconds of hesitation he said no I've
1:51:43asked a lot of questions in the Senate
1:51:44that wouldnt get a lot of attention I
1:51:46think because it really got to the heart
1:51:50of the issue in terms of mr. Zuckerberg
1:51:52in his own feeling about personal
1:51:54privacy and where he would draw the line
1:51:56we know of course from what senator
1:51:59Harris has said and what we all know
1:52:01from life experience is the last hotel I
1:52:04stayed in is probably a matter of some
1:52:06record with my name attached to it
1:52:08somewhere who knows but privacy is a
1:52:13critical issue here and the right of
1:52:15Facebook or any entity to use my
1:52:18information without my express
1:52:20permission I think is over the line I
1:52:22sent to Senator Coburn senator Cornyn my
1:52:26friend amended earlier when he said that
1:52:29he thought consumers were aware when
1:52:31information was being gathered on him
1:52:33but I'm sure he's wrong we have now put
1:52:36a little piece of electric tape over the
1:52:39camera on my laptop that most people do
1:52:43now because they're being watched and
1:52:45they may not even know it but there are
1:52:47two other issues here that I'd like to
1:52:49spend a moment addressing and one of
1:52:53them relates to mr. Nix mr. Nix is a
1:52:56British national and said correct yes
1:52:58and he was clearly involved in some of
1:53:03the campaign activities of Cambridge
1:53:05analytic but he was the CEO of the
1:53:10company so he was the point person for
1:53:12all of the clients and he often made the
1:53:15presentations and recommendations to
1:53:17those clients and the Federal Election
1:53:19Commission says expressly that
1:53:21regulations prohibit foreign nationals
1:53:24from directing dictating controlling or
1:53:26directly indirectly participating in the
1:53:28decision-making process of any person
1:53:30with regard to any election related
1:53:32activities in the United States so there
1:53:35is a red flag or red Union Jack whatever
1:53:39that should make it clear that word
1:53:42we're in a territory here that may be
1:53:44maybe a violation of law you have said
1:53:48that the involvement of Cambridge and
1:53:51successor organizations and Russia came
1:53:55after you left is it correct I the the
1:54:00no sorry can you clarify slightly what
1:54:02you mean by my my my experiences direct
1:54:08as it relates to Lukoil as it relates to
1:54:12understanding you know the research that
1:54:15was being done in terms of the focus
1:54:16groups and all of that but in the the
1:54:18use of information by the Russians in
1:54:21the election campaign oh I'm sorry I
1:54:23misunderstood yes that was either
1:54:26happened some other place or after you
1:54:27had left Kate generally yeah I was I was
1:54:29not aware at the time that there was
1:54:33activities in Russia to influence the
1:54:37United States elections so I would just
1:54:39add that to the second category the
1:54:41first question privacy the second
1:54:43question is the involvement of foreign
1:54:45nationals in the United States campaign
1:54:47whether mr. Nix personally or Russia as
1:54:49a country trying to influence the impact
1:54:52the third has been brought up on my
1:54:53colleagues and I think gets to the heart
1:54:55of another very important issue and that
1:54:57is the issue of the secrecy of this
1:54:59activity the fact that we know that mr.
1:55:02Mercer was engaged in this is because of
1:55:05something called open secrets and other
1:55:07sources that weren't disclosed in the
1:55:10ordinary course of business in this and
1:55:14did you have any guidance from Cambridge
1:55:16when you were there in terms of the
1:55:18secrecy of the clients that you were
1:55:21working for so to to your first point
1:55:25about non-us nationals working in u.s.
1:55:27elections Cambridge analytic I received
1:55:31formal legal advice and which I've
1:55:35disclosed to the media and that legal
1:55:38advice did make clear that the company
1:55:41should not be sending non-us nationals
1:55:43to work on American elections
1:55:48Minh with respect to the the secrecy
1:55:55everybody had to sign a nondisclosure
1:55:57agreement a very thorough non-disclosure
1:55:59agreement and in fact after I left you
1:56:02know when the when the company engaged
1:56:05in a protracted legal correspondence
1:56:08with me also ensured that myself and
1:56:11other people who decided to leave signed
1:56:14an undertaking of confidentiality which
1:56:18was sort of a higher threshold of NDA
1:56:20you've been asked for some documents
1:56:22earlier and those two you just referred
1:56:23to the legal opinion about the
1:56:25involvement of Cambridge yeah employees
1:56:29it's I mean I made it public so it's if
1:56:32you wouldn't mind share it with the
1:56:33country as well as I'll make it out of
1:56:35hood a copy of the non-disclosure
1:56:36agreement that you signed or others
1:56:38might have signed that would be helpful
1:56:39too thanks mr. chairman Thank You
1:56:41Senator Durbin senator Hirono Thank You
1:56:43professor Hirsch elections very much
1:56:46turn on voter turnout so you would agree
1:56:50that efforts to suppress voter
1:56:52engagement or voter turnout should be a
1:56:54matter of serious concern to us
1:56:56certainly you testified that it is
1:57:00easier to demobilize people than to
1:57:02mobilize them in fact you've testified
1:57:04on the impact of voter ID requirements
1:57:08on voter turnout so if Congress if we
1:57:12were to consider regulating anything and
1:57:14I realized that there are privacy issues
1:57:15it's all very complicated but if you
1:57:17want to focus on some kind of a
1:57:19regulatory scheme should we focus on
1:57:22regulating ads or messages that
1:57:24demobilize people so thank you for the
1:57:26question I think this is really
1:57:28complicated personally I've worked as an
1:57:29extra witness for the ACLU in the Obama
1:57:31Justice Department on government voter
1:57:33suppression yes
1:57:34voter ID laws nonsensical regulations
1:57:36this is really different and it's
1:57:38different because campaigns do things
1:57:40that are not nice all the time if a
1:57:42Democratic campaign were to go and
1:57:44remind Trump voters of all of the moral
1:57:46failings and Trump's past and say don't
1:57:47vote for him is that demobilization or
1:57:50not I don't know I really I know there's
1:57:53nothing to figure out whether a message
1:57:55actually demobilize does anybody but
1:57:57let's assume that we can come up with
1:57:59some
1:57:59way to define what demobilize because
1:58:02that is a sea of matter serious concern
1:58:04you said that that is easier to affect
1:58:06so that may be an area for us to pursue
1:58:10in terms of any kind of regulation in
1:58:13this very complex space that we're in
1:58:15right now that might be right i I mean
1:58:18I'm not a regulatory expert in terms of
1:58:21you know First Amendment issues and so
1:58:23forth you know if I were to ask someone
1:58:24who they're voting for and they said
1:58:26president Trump and I said don't vote
1:58:29for him I'd really rather you not vote
1:58:31it doesn't seem to me that that is a
1:58:33form of voter suppression I know we're
1:58:36talking about basically people paying to
1:58:39suppress opposed as mr. Wiley testified
1:58:41that Steve Bannon was running an
1:58:43operation where clients could request
1:58:45voter suppression messages which they
1:58:48would then pay for we're talking about a
1:58:50very different kind of circumstance than
1:58:52somebody just saying yeah well don't
1:58:53vote for so-and-so so I think you
1:58:55understand the differences mr. Wiley you
1:58:57obviously have an awareness of the use
1:58:59of misuse of a massive amounts of data
1:59:01so I want to ask you this the US
1:59:03Immigration and Customs Enforcement has
1:59:05proposed a new extreme venting
1:59:07initiative or life cycle vetting their
1:59:10plan is to hire a contractor to exploit
1:59:12publicly available information such as
1:59:14media blogs public hearings conferences
1:59:17academic websites social media websites
1:59:20such as Twitter Facebook LinkedIn
1:59:22LinkedIn to extract pertinent
1:59:24information regarding targets to
1:59:27determine who will be a productive
1:59:29member of society and who will commit
1:59:31crimes and terrorist acts
1:59:33we're talking about people who are
1:59:35wanting visas to come to our country so
1:59:38according to the nonpartisan Brennan
1:59:40Center for Justice
1:59:41Isis plan would automatically fat people
1:59:44for deportation or visa denial based on
1:59:47the exact criteria from the original
1:59:49Muslim ban so it was the original Muslim
1:59:51ban of the president that set up this
1:59:53this extreme vetting program and so do
1:59:59you think that that kind of prediction
2:00:01of human behavior as to whether somebody
2:00:03is going to become an upstanding
2:00:05contributing member or whether that
2:00:07person is likely to become a terrorist
2:00:08or or become a criminal is even possible
2:00:11from the sorts of information that could
2:00:13be
2:00:13available to the US government so two
2:00:17points that there is no Universal
2:00:20definition of a bad person and therefore
2:00:24it is a social construct and it is it is
2:00:27laden with people's moral judgments so
2:00:31there is no mathematical way to
2:00:33determine whether someone is a bad
2:00:34person in the abstract in the sense that
2:00:37there is no Universal definition of a
2:00:39bad person so the second point that I
2:00:43would make though is that just because
2:00:46you are using data Science and
2:00:49Mathematics
2:00:50you could have the most advanced neural
2:00:55network made yet but if the underlying
2:01:00training set for that algorithm uses
2:01:02systemically biased information so for
2:01:05example if you're looking at criminal
2:01:07justice statistics and you have a model
2:01:10that ends up predicting that this
2:01:12particular african-american is more
2:01:14likely to commit a crime because more
2:01:16african-americans end up in prison you
2:01:18you create an algorithm which is simply
2:01:20reflecting social and moral biases of
2:01:23the of the of the data sets in the
2:01:25p-a-k-i I think that you know the
2:01:28predictive value of what the government
2:01:30is pursuing is very questionable and
2:01:33problematic in terms of all kinds of
2:01:35issues privacy you name it and and the
2:01:37accuracy and the ability of that kind of
2:01:39initiative and yet they are pursuing it
2:01:42by the way so I think it's very
2:01:44problematic of professors would you
2:01:46agree that you know we admit this
2:01:47government getting their hands and all
2:01:49this kind of information so that they
2:01:51can determine whether somebody would
2:01:53commit crimes or somebody's gonna become
2:01:55an upstanding citizen does that even
2:01:57make sense to you no thank you
2:02:00Thank You mr. chairman Sarah Booker
2:02:02thank you very much mr. Wylie um I have
2:02:07a lot of concerns about how these
2:02:08platforms can be used to pit people
2:02:10against each other one of the greatest
2:02:13values of America is this idea of
2:02:14indivisibility that we have the lines of
2:02:16the Tyus together are stronger than than
2:02:18the lines that divide us as a country
2:02:20and you know this morning when I woke up
2:02:23I listened to the New York Times
2:02:24Podcast which was all about Sri Lanka
2:02:27the headline is when Facebook rumors
2:02:28incite real violence and it was about
2:02:31how the platform Facebook was being used
2:02:33by sinister forces to incite hatred
2:02:37between groups fear between groups and
2:02:40ultimately violence in this circumstance
2:02:43I was just so deeply disappointed and
2:02:48angered when Nigel Oakes who's the
2:02:50founder of the SEL group as you wrote
2:02:52would would say things and record a
2:02:53conversation worries it's things like
2:02:57this that resonates sometime to attack
2:02:58the other group and know that you're
2:03:00going to lose them is going to reinforce
2:03:02and resonate your group which is why
2:03:05Hitler attacked the Jews because he
2:03:08didn't have a problem with the Jews at
2:03:10all but people didn't like the Jews so
2:03:13he just leveraged in our official enemy
2:03:15well that's exactly what Trump did he
2:03:17leveraged a Muslim Trump had the balls
2:03:21and I mean really the balls to say what
2:03:23people wanted to hear this has to meet a
2:03:26really frightening reality a threat to
2:03:30the very ideal of a nation that wasn't
2:03:31founded because we all pray alike
2:03:32because we all looked alike

About this transcript

This page contains the full transcript of Christopher Wylie Testifies Before U.S. Senate by Bloomberg News, generated from the public captions YouTube serves with the video. The transcript has 18,062 words across 2,794 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.