Oral evidence: Fake News, HC 363
Oral evidence, 17 April 2018.
Witness
Brittany Kaiser.
Chair
Good morning. Welcome to this further session of the Digital, Culture, Media and Sport Select Committee as part of our ongoing investigation into disinformation and fake news. I am very pleased to welcome Brittany Kaiser to give evidence to the Committee this morning. Brittany has also supplied the Committee with some written evidence, which we have formally agreed to publish and is now available on the Committee website. It includes Brittany Kaiser’s written statement and some other supporting documents. There are also some emails which Britany Kaiser has given to us. We need to do a bit of work on those to remove some of the personal data that is contained in that document, giving personal email addresses and the like. We have agreed that we will publish those emails separately as soon as we have had the chance to prepare them and put them in a format where we can publish them. The bulk of that written evidence is now available for people to view online. I am sure we will be referring to that evidence and the emails during the course of our discussions this morning. For the benefit of the Committee and the record, Brittany Kaiser, can you give us an insight into how you came to work for Cambridge Analytica and SCL and the way in which the company was structured and work was conducted through those companies?
Brittany Kaiser
Of course, good morning, Chairman and hon. Members. Firstly, I want to thank you for this opportunity to bear witness to the inquiry on fake news. As a believer in truth, universal human rights and democracy, I understand that fake news undermines these ideals and much of what we stand for and hold dear. It is evidence of a much deeper problem. We have spent too many years using technology to take advantage of people’s openness and good will, and I will not stand by any longer to observe this privileged abuse of power, intentionally or unintentionally. I want to help shed light in these dark places and to help people be more responsible in their actions, which is why I spoke out originally and why I have returned to the UK willingly and at my own expense to give testimony to this inquiry. Short background about myself: I was born in Houston, Texas and raised in Chicago. I went to high school at Phillips Academy, Andover outside Boston before moving to the United Kingdom to attend Edinburgh University to pursue some of my life-long passions: human rights and international relations. During my academic studies I volunteered on the presidential campaigns of Howard Dean and John Kerry, as well as the senatorial campaign of then State Senator Barack Obama, before later joining a small new media team on his presidential campaign, which is of note because as part of that team I was part of the first-time use of social media in political communications. I spent several years working on human rights projects around the world, first for Amnesty International, and then lobbying at the United Nations and the European Parliament, specifically on stopping crimes against humanity and abuses of power.
Chair
Sorry, if I can stop you there, I appreciate that background detail; a lot of it is included in your written statement, which we have read and published as well. I think we can probably move on to your starting work at Cambridge Analytica and your work there.
Brittany Kaiser
Of course. I first met Alexander Nix when I was working for Democrats Abroad in London—I think it was 2013. Friends of ours thought it would be a good joke to introduce us, as he was consulting for the Republican party, while I was a registered Democrat—I still am today. He was very interested in learning more about my experience with the Democrats, as he was running a political consultancy, which was very well known at the time. He gave me his card and said, “Let me get you drunk and steal your secrets.” I did not call him at the time. But we met coincidentally a few more times after that, and in December 2014 he offered me a job. I wanted to find a way to be more effective in campaigning and in social change projects. His firm, Strategic Communication Laboratories—the SCL Group—boasted of scientifically valid methods of measuring effectiveness in communications. I have always wanted to use this power for good, not ill. The first project I worked on with SCL was designing a communications campaign to drive engagement with the reconstruction of Ebola-stricken countries. I began with the SCL Group on a part-time basis in December 2014. My job title was left to me. I called myself “special adviser”. I later became a full-time employee in February 2015. My title was “director of business development”. As a person who is really an over-glorified salesgirl, these titles tend to be slightly inflated. I was later called “director of business development” at the request of my superiors.
Chair
When you were employed at Cambridge Analytica, were you employed as a staff contractor or as a consultant?
Brittany Kaiser
I was a consultant when I first started and then I was a full-time permanent employee from February 2015.
Chair
Okay, so from a UK basis you paid income tax through the company and you were on a regular full-time contract?
Brittany Kaiser
Yes, until January of this year.
Chair
In January, did you leave the company completely, or did you leave your full-time role but continue to do some part-time or consultancy work for them?
Brittany Kaiser
I left my full-time role at the company, hoping to come on as a consultant and we were not able to agree anything, so I have not been paid or been a part of the company since January.
Chair
So you have not received any money at all from them since January?
Brittany Kaiser
No.
Chair
In terms of your change of status from a full-time role, that ended in January, while you sought to explore working as a consultant. That ultimately went nowhere and you were not working on any projects at that time.
Brittany Kaiser
That is correct. At no time was I part of the executive leadership or top management of the SCL Group or Cambridge Analytica, although I wanted to be. I really believe that I contributed hugely to the success of the company. I was never an officer of the board of any of the companies. I was also not a data scientist and never had access to any data sets, let alone the controversial data sets I hope to explore today. Since I left, I have been learning just how many secrets the management team was keeping me out of—looking back with fresh eyes at the emails I was copied into and other documents with the benefit of hindsight. My job at Cambridge Analytica was primarily to meet potential clients, hear what they wanted and required, and then to brief the operations team in order to produce a sales proposal, often under the guidance of senior management and executives, such as CEO Alexander Nix, CFO and COO Julian Wheatland, and chief data officer, Dr Alexander Taylor. I had access to company materials on products and services, the account management in those teams and data on potential and current clients. My testimony today is informed by what I witnessed and experienced during my time at Cambridge Analytica and the SCL Group, and materials I worked on or was copied into. When I originally spoke to The Guardian in March of this year, I provided relevant documents from my Cambridge Analytica inbox, which I was subsequently locked out of. But I later found an offline copy of my SCL Group inbox, with further materials that have been provided to the Committee today.
Chair
Thank you very much for that. Could you explain a little about the way in which SCL works as a group of companies? We have had some evidence that everything is SCL and that Cambridge Analytica is really just a fascia—that no one really works for it, everyone works for SCL. It is just another brass plate behind which work is done, and there are lots of other entities like that. Is that a fair reflection, based on your experience, of how the company works?
Brittany Kaiser
I was never privy to the full company structure, perhaps for reasons that have now become known to the Committee. But it was my understanding when I joined the SCL Group that it was a parent company of a few different divisions. One would be SCL Commercial, SCL Elections, SCL Defence, SCL Social, and then Cambridge Analytica, which I understood was the US-acting subsidiary that was created with new investment in order to explore advanced data science methods.
Chair
Would you say that your sense of the organisation was that it was effectively one organisation that had different divisions that worked on different projects?
Brittany Kaiser
Yes, most staff were shared between the companies. I don’t believe that anyone was even ever employed directly by Cambridge Analytica. We had another subsidiary called SCL USA, which was the contracting entity in the United States.
Chair
What can you tell us about Emerdata? Is that a new project being funded through SCL or is it something completely different?
Brittany Kaiser
The first time I heard the word Emerdata was in the press. I believe that that was in relation to the private raise that my executives were undertaking for investment, which was closed in January. They created a new holding company structure. I never knew what the name of that structure was until I saw that in the press. I do assume that was the same entity that I had heard about.
Chair
Do you believe it was a new vehicle through which they were raising money in order to launch new businesses? Would that be fair?
Brittany Kaiser
I thought it was specifically to invest in the development of Cambridge Analytica and the SCL Group.
Chair
Could you tell us a little about where else in the world you have worked? We may later have some more detailed questions about specific areas, but could you just give us a sense of which countries that you have worked in or that you know SCL worked in during your time there?
Brittany Kaiser
The only countries that I worked in for Cambridge Analytica were the United Kingdom, the United States and Mexico. I worked on projects in other countries, or on winning contracts for those projects in countries in Asia, eastern Europe and Latin America, but I was never physically present as part of the campaign team, due to the nature of my business development role.
Chair
Can you say which specific countries?
Brittany Kaiser
Of course. First, Nigeria, Ethiopia, Lithuania, Malaysia; I have already said Mexico. Most other places were a pitch, where I don’t believe that I contributed to work that was ever really undertaken. Ethiopia was also just a pitch, not work that was undertaken. I did spend quite a long time on commercial business development in the United States, which was my main focus for the latter half of my time with the company.
Chair
Did you work in Malta?
Brittany Kaiser
No.
Chair
Are you aware that Cambridge Analytica or SCL worked in Malta?
Brittany Kaiser
No, sir.
Chair
Or Israel?
Brittany Kaiser
No, sir.
Chair
People have asked whether there was work done with Germany.
Brittany Kaiser
Oh. I was part of preparing a pitch with Germany for political work, which was never undertaken, and for a variety of commercial projects, some of which were contracted for.
Chair
Who was that pitch directed to?
Brittany Kaiser
The CDU.
Chair
To the CDU?
Brittany Kaiser
Yes.
Chair
And in France? Was there any work related to the French presidential election?
Brittany Kaiser
I did go there with Alexander Nix—I can’t remember; it must have been in 2015 at some point—to explore partnering with a consultancy firm there. In the end, similar to Germany, they said that using data for political communications was not advisable in that work.
Chair
Or legal.
Brittany Kaiser
They were not interested in pursuing further work together.
Chair
Was it advisable or legal, in their words?
Brittany Kaiser
They said that if the use of data—which was legal—in a variety of different ways for political communications after an opt-in were found out by the press, they would probably lose business over that, and were not interested in putting at risk the rest of their work that they were already undertaking.
Chair
But as you know, under European data protection law there are already quite strict rules around holding data relating to people’s political affiliations and opinions. There are restrictions on doing that outside political parties.
Brittany Kaiser
Of course, yes.
Chair
I’m sure you saw the Channel 4 investigation into Cambridge Analytica, or had the chance to watch it on catch-up. They talk about ghosting in and out of countries: setting up shadow companies and then closing them down after elections. Does that tally with your experience?
Brittany Kaiser
No, not at all. It is incredibly shocking how far sometimes a sales pitch could go through to suggesting things that I believe are illegal or at least quite shadowy. I have never heard most of what was uttered on those videos ever before in any meeting that I have been a part of, nor was I aware or had any inkling that that might have been suggested when I was not present.
Julie Elliott
You mentioned eastern Europe in passing. That came out in the Channel 4 recording. What work are you aware of that was done in the Czech Republic?
Brittany Kaiser
I was not a part of any pitches in the Czech Republic. I do remember it maybe being mentioned in a sales call once but I was not a part of that.
Julie Elliott
But are you aware that Cambridge Analytica were working in the Czech Republic?
Brittany Kaiser
Were we? I have no idea. Sorry about that.
Julie Elliott
So you wouldn’t know who they would be working for there?
Brittany Kaiser
No.
Julie Elliott
The other place that I am interested in is Argentina. Are you aware of any work that was done in Argentina?
Brittany Kaiser
The first commercial project I was ever aware of was in Argentina for an agricultural company that was mostly communications and branding. I also was very well aware of political pitches being undertaken in Argentina, although I do not believe any work was ever actually done.
Julie Elliott
Who were the pitches to?
Brittany Kaiser
To the current President.
Brendan O'Hara
You were director of programme development at Cambridge Analytica between 2014 and 2018. Where were you predominantly based?
Brittany Kaiser
For the first six or seven months I was based out of our London headquarters, which at the time was in Yarmouth Place in Mayfair. Then I moved to our office in Alexandria, Virginia for about nine months, and then to our office in New York for nine months. Since last August I have been based in Mexico City until this past January, when I left the company.
Brendan O'Hara
On 27 February in front of this Committee Alexander Nix said: “I have been with the company for about 14 years and I have never worked on a campaign in the UK”. Is that your understanding?
Brittany Kaiser
I don’t believe that before our involvement with the Leave.EU campaign that Cambridge Analytica or the SCL Group ever undertook any political work in the United Kingdom. But in September 2015, I believe, we began conversational engagements at first with Leave.EU and the UK Independence party—
Brendan O'Hara
Excuse me for interrupting. So, as the director of Cambridge Analytica or SCL, you were unaware of that. Looking at the website, it says: “SCL Elections has worked on three UK election campaigns over the past 15 years. During the first of these elections, SCL was contracted to undertake advanced behavioural polling and Target Audience Analysis, and was subsequently credited as being the only agency to correctly predict the outcome of that General Election. For the two subsequent elections, SCL Elections provided campaign consultancy and strategic guidance for specific candidates.” Are you utterly unaware of that?
Brittany Kaiser
That was brought to my attention and I do not believe any of that is true. It was likely put on there as sales material to back up the fact that we were worthy of being hired for the Leave.EU campaign.
Brendan O'Hara
So you, in your position as a director of a company, are saying that what was on the company’s website is complete lies?
Brittany Kaiser
It could be. Just to differentiate between my directorship and an actual director of the company, I was a glorified sales person. I never held an actual company directorship. Nor was I privy to any management meetings or any higher-level executive authority.
Brendan O'Hara
It appears to me that these are quite specific boasts. For this to be a complete fabrication—to be, as you say, a complete tissue of lies—I don’t want to put words in your mouth, but are you telling me that this is a complete tissue of lies?
Brittany Kaiser
I have never pitched that to absolutely anybody. It was not on our website when I was engaged with Leave.EU. To Arron Banks, Andy Wigmore, Liz Bilney, Matthew Richardson—to everybody I was pitching—I said that we had never undertaken work in the United Kingdom, which I believed to be true. I have seen that before, and I believe it is a cached version of an old website that was never live, or was never live when I was pitching on behalf of the company.
Brendan O'Hara
How would you describe a company that then promotes itself as having worked in three UK elections, having advanced behavioural polling analysis, and worked for two subsequent elections, even down to the level of providing guidance for specific candidates?
Brittany Kaiser
I don’t believe that was ever actually claimed. If it was put up on a website, I have never seen that live—but I have been shown that piece of paper before.
Brendan O'Hara
So you believe Alexander Nix when he says, “I have been with the company for about 14 years and I have never worked on a campaign in the UK”, despite the company itself claiming to have worked extensively in the United Kingdom.
Brittany Kaiser
I am not sure that a cached version of an old website is an actual claim, but I do know that we have been in pitches and negotiations with UK parties in the past, such as the SNP.
Brendan O'Hara
Well, I will leave you to follow that particular stag into the forest, then. Please, feel free.
Brittany Kaiser
I was not a part of those pitches or negotiations. I was based in Washington DC at the time.
Brendan O'Hara
You’ve told me categorically that that’s not true. You then tell me that you have pitched work for the SNP but then tell me that you know nothing about it.
Brittany Kaiser
I believe that there were meetings that took place in London, where individuals came down from Edinburgh to visit us at our Mayfair headquarters. Then further meetings were undertaken in Edinburgh, near the Parliament.
Brendan O'Hara
And do you know the individuals concerned? Do you know what the pitch was?
Brittany Kaiser
I could probably look through old emails and find some names for you and submit that after this inquiry.
Brendan O'Hara
There was no work done at any UK elections for any political party, but you suddenly have a memory of meetings that you were at and meetings with the SNP—is that right?
Brittany Kaiser
Pitches are very different from work undertaken. I would like to correct that by saying that I do believe work was undertaken for a party, and that was the UK Independence party—a small piece of work that was meant to support the Leave.EU campaign during the Brexit election.
Chair
We will come to that shortly.
Brendan O'Hara
Wrapping up, Christopher Wylie said he knew that Cambridge Analytica had pitched for work in the 2014 Scottish independence referendum, but he does not know what came of it. Do you?
Brittany Kaiser
I don’t believe that any work was actually undertaken.
Brendan O'Hara
Do you know what was pitched for and from whom?
Brittany Kaiser
It would have been for research modelling and targeted communications within Scotland and for individuals who had the right to vote.
Brendan O'Hara
Obviously they would have the right to vote, but I am asking, do you know which organisations you pitched for work from?
Brittany Kaiser
I don’t know, but I could try to find that out for you, because I do have a significant amount of emails and documentation from that time.
Brendan O'Hara
I’ll leave it there and maybe return later.
Chair
You have mentioned a wide range of projects and pitches. Who typically is involved in introducing Cambridge Analytica to meet new clients?
Brittany Kaiser
Either it would be somebody internally within the company—we had a lot of people who worked in business development or project management who were specifically hired because of their global network of individuals. We also had, I would say, a kind of roster of consultants to the SCL Group and Cambridge Analytica who were based globally. Some of them would represent us in a part-time capacity and be the local SCL Group office, and some of those individuals would only work for commissions of contracts that they introduced and had no time commitment to the company. So, introductions would usually come from individuals that had contractual agreement with SCL Group and Cambridge Analytica.
Chair
I just want to try and personalise it a little bit more. You have mentioned UKIP and Leave.EU. Who introduced, say, Arron Banks to Cambridge Analytica?
Brittany Kaiser
I believe that introduction came from Steve Bannon, who sat on our board at the time.
Chair
Among some of the other countries you have mentioned, you mentioned France. Which political organisation were you pitching to in France?
Brittany Kaiser
Originally it was to Sarkozy’s campaign. That was before I started at the company, and I remember that Alexander used to tell the story of how he pitched him and they said that they were going to win—and then lost—the election, so we returned again to pitch to the newly branded Les Républicains.
Chair
And in Lithuania, which political party were you working for?
Brittany Kaiser
That was for a mayoral campaign in Vilnius. I can’t remember the pronunciation or the exact name of the party itself.
Chair
And in Mexico?
Brittany Kaiser
In Mexico, it was for the party that is currently in Government.
Paul Farrelly
Colleagues have got lots of questions in a few moments about Leave.EU and the work—unpaid—that was done. But for me, as well as your statement about Emerdata, it is a moving story, where revelations are coming from different sources at all times. When we interviewed Alexander Nix, the line of questioning, because of the knowledge at the time, was focused on Leave.EU, and then we had subsequent revelations about AggregateIQ and its involvement and connections with SCL, Cambridge Analytica—the whole group. I just wanted to ask you a few questions about that, because my abiding impression after interviewing Mr Nix is that he left all that bit out and that we were rather blindsided. Have you watched that interview?
Brittany Kaiser
Yes, I have.
Paul Farrelly
Do you think I am right to feel blindsided a little bit?
Brittany Kaiser
I think that you are right to feel that there was some information that was left out, whether intentionally or unintentionally.
Paul Farrelly
Clearly, there is a difference between what happened, what work was really done, and the bullshit of sales pitches that you have described. You say in your evidence: “I should emphasise that the Kogan/GSR datasets and questionnaires were not the only Facebook-connected questionnaires and datasets which Cambridge Analytica used…But I believe it is almost certain that the number of Facebook users whose data was compromised through routes similar to that used by Kogan is much greater than 87 million; and that both Cambridge Analytica and other unconnected companies and campaigns were involved in these activities.” Can you expand a little on that, please?
Brittany Kaiser
Of course. When I first joined the company, our creative and psychology teams and data science teams worked together to design some of these questionnaires. So I am aware now of what the questionnaire was that Professor Kogan used, although I didn’t know about it when I joined the company. I would see questionnaires. For example, there was one called the “sex compass”, to find out what your personal preferences were privately, and then there was another one on your “music personality”. In my pitches, I used to give examples, even to clients, that if you go on Facebook and see these viral personality quizzes, not all of them would have been designed by Cambridge Analytica, SCL Group or our affiliates, but that these applications were designed specifically to harvest data from individuals using Facebook as the tool. I know at least of those two examples—the music version and the “sex compass”—which were both quizzes that were separate from Aleksandr Kogan’s quiz. Therefore, it can be inferred or implied that there were many additional individuals, as opposed to just the ones through Aleksandr Kogan’s test, whose data may have been compromised.
Paul Farrelly
It seems blindingly obvious, but we need to search for the evidence. If you have gathered this tremendous data set because you have had access not only to respondees but to their friends and their likes through Facebook, and you have developed this for the US, then if this trick works, you would want to replicate it for other markets that you were pitching for work in, would you not?
Brittany Kaiser
That was the idea, although in Europe it is quite difficult due to the data protection laws.
Paul Farrelly
If you observe them?
Brittany Kaiser
Correct.
Paul Farrelly
Could you describe what knowledge you have about AggregateIQ and how it operated under the SCL umbrella?
Brittany Kaiser
Of course. It was my understanding when I joined the company that AggregateIQ was our exclusive digital and data engineering partner. They would build our software, such as a platform that we designed for Senator Ted Cruz’s campaign. That was meant to collect data for canvassing individuals who would go door to door collecting and hygiening data of individuals in those households. We also had no internal digital capacity at the time, so we did not actually undertake any of our digital campaigns. That was done exclusively through AggregateIQ. Through the rollercoaster of Senator Cruz’s campaign and engagements with the consultants who were working on it, I believed that our relationship with AggregateIQ became strained. They did not turn in work to us on time. Our clients were not always happy with their performance and we therefore stopped working with them. Subsequently, we have heard in the press and from other individuals that some of our board members might have had ownership of AggregateIQ, which I was not aware of at all.
Paul Farrelly
When did you think or consider that the group had stopped working with AggregateIQ?
Brittany Kaiser
Probably late 2015 was when issues became quite strained, when the campaign ramped up and we were near the end of the Republican primaries.
Paul Farrelly
We have had a series of documents and testimonies from Mr Wylie that allege that AggregateIQ was part of the SCL-CA umbrella, did extensive work for Vote Leave and was paid a large sum of money for that.
Brittany Kaiser
I have seen that and his testimony could be exactly correct. I had no knowledge of that type of relationship with AIQ and I actually have never met Chris Wylie. He was no longer with the company when I joined and I believed that he was a consultant, sometimes referred to as SCL Canada, that was based in Canada, working full time for AggregateIQ.
Paul Farrelly
Before and during the referendum campaign here, what was your state of awareness of any work that was being done by the group, including AggregateIQ, for Vote Leave?
Brittany Kaiser
My only interaction with Vote Leave was that I was supposed to have a meeting with Matthew Elliott two days after my first meeting with Arron Banks. That meeting was cancelled after he found out that we had a meeting with executives from Leave.EU, so I never met anybody from Vote Leave and I did not know that AIQ, whom I did not think we were working with at the time anyway—nor did I think we had joint investors or owners of our companies—was pitching for that work or undertaking any of that work.
Paul Farrelly
So while you were there—and you were there until early this year—you were not aware of any talk, chatter or boasts within the company at the time SCL Cambridge Analytica were doing extensive work with AggregateIQ within their umbrella for Vote Leave?
Brittany Kaiser
I wish that I was. It sounds like we probably were working very closely, but I was not involved in US politics at that point.
Paul Farrelly
UK?
Brittany Kaiser
Well—
Paul Farrelly
It was a UK referendum. The United States was not leaving Europe.
Brittany Kaiser
Of course, but Cambridge Analytica and the joint ownership of Aggregate IQ was based out of our US companies and affiliates.
Chair
One of your associates is handing you a piece of paper. We are happy to give you a moment to look at it before Mr Farrelly resumes his questions, if that would be helpful.
Brittany Kaiser
Yes. Of course, I cannot speak for SCL Canada and their relationship with Aggregate IQ, or about documents I am not privy to.
Paul Farrelly
So what were you doing in the group at the time of the referendum in the UK? You were pitching elsewhere, were you?
Brittany Kaiser
Yes. I was between Washington, DC and London. I was undertaking a series of pitches for Leave.EU, the UK Independence party, and Eldon and GoSkippy Insurance.
Paul Farrelly
But you were left out of any work with Vote Leave. That seems odd.
Brittany Kaiser
Well, Vote Leave was technically the competition to Leave.EU. So upon engaging with Leave.EU I was not supposed to be meeting with Vote Leave, because we were in competition with them for the designation from the Electoral Commission.
Paul Farrelly
So it seems there was a Chinese wall operating within the company.
Brittany Kaiser
It sounds like there was.
Paul Farrelly
But you didn’t know about it?
Brittany Kaiser
No, sir.
Chair
I just want to pick up one thing that you mentioned there, which was the other personality surveys that Cambridge Analytica ran on Facebook. We touched on that with Alexander Nix last time around, and he said that those surveys where there was a Facebook login at the end to complete the survey did not mean that your personal data was then being accessed by Cambridge Analytica. But you are saying that that was not the case—that actually the purpose of the survey was to gather this information and that by completing it with your Facebook login, Cambridge Analytica will also get access to your data on Facebook.
Brittany Kaiser
I believe that was the point of the quizzes in the first place, yes.
Chair
That does sound like the point of it, but that is totally different from what Alexander Nix said to us.
Ian C. Lucas
Did you know at the time that you were pitching to Leave.EU that Cambridge Analytica was also pitching to Vote Leave?
Brittany Kaiser
I knew that I had a meeting that was meant to take place after Leave.EU, so that I could also explore work with Vote Leave, but, as I said, that meeting was cancelled and I didn’t know that there were any further engagements with Vote Leave until Mr Arron Banks made it clear that he was not interested in paying us for any of the work that we had either already undertaken or proposed to undertake. So I do believe that some of my executives re-engaged in talks with Vote Leave, but at the time I was no longer involved in that.
Ian C. Lucas
Did you know that Cambridge Analytica had an agreement with AIQ to transfer data?
Brittany Kaiser
We would have had that, because that would have been essential to our work on Senator Ted Cruz’s presidential campaign. But I was under the impression that AIQ did not fulfil its contractual obligations to our client and therefore we were no longer working with them. I remember many angry phone calls back and forth between AIQ and our internal staff, that tensions were very tight, and that we decided instead to hire our own data engineers and our own internal digital team, and build out internal capacity, so that we no longer had to rely on them.
Ian C. Lucas
Can you confirm that Cambridge Analytica transferred data to AIQ?
Brittany Kaiser
What I can say is that the essential relationship between AIQ and Cambridge Analytica—or the SCL Group—was that AIQ had built platforms that we used to collect data, so we would have had to transfer data between ourselves for us to access that; and that they also ran our digital campaigns for a while, not just for Senator Ted Cruz, but for other clients—political, commercial and social. So we would have had to transfer data back to them for them to use those lists for digital targeting. I believe that by the essential nature of that relationship, that would have been a requirement.
Ian C. Lucas
Of course, once data is transferred to AIQ, they have the data from Cambridge Analytica. So they could use Cambridge Analytica data to pursue their relationship with Vote Leave.
Brittany Kaiser
They could have done, but likely in contravention to the contracts that we originally signed with them—although I am not privy to those contracts, so I can’t be sure.
Chair
On a point of clarification, you mentioned Chris Wylie worked for SCL Canada.
Brittany Kaiser
That was the name used for—I don’t know him as an individual or the AIQ office. That was considered SCL Canada. Our company tended to have a business model where we would partner with another company, and that company would represent us as SCL Germany or SCL USA. That was the model.
Chair
So, as far as you are concerned, SCL Canada and AIQ are the same thing.
Brittany Kaiser
I believe so, yes.
Chair
Do you believe Chris Wylie was working for AIQ if he was also working for SCL Canada?
Brittany Kaiser
Yes, I believe that SCL Canada was just a glorified title, but he was full time engaged with AIQ.
Simon Hart
Sticking with the referendum theme for a moment and dealing more specifically with Leave.EU, once the pitch was successful, what was the actual level of work involved between Cambridge Analytica and Leave.EU? What was the scope of the work?
Brittany Kaiser
When I first began engagement with Leave.EU we had parallel discussions going on on what became—
Simon Hart
You pitched for the work yourself?
Brittany Kaiser
Yes, myself and our CFO and COO, Julian Wheatland, who is the current acting CEO, and a senior data scientist and usually a psychologist would be in the room as well, or one of our researchers. We were undertaking parallel pitches for Leave.EU, the UK Independence party and Eldon and GoSkippy Insurance. Those proposals at first were all preliminary. The one for Leave.EU and the UK Independence party included taking receipt of data and matching that data with researcher survey data and then modelling it so that we could find the different personality types of individuals that would be most likely to engage with the messaging for Leave.EU. That was specifically for individuals that would be interested in voting to leave in the referendum, and for the UK Independence party it was obviously similar, as that is the nature of the party itself. When we first started talking to Leave.EU they did not have much data to speak of because it was very much in the beginning of the campaign. So, in parallel, we were in talks with the UK Independence party where we believed that their membership data would be quite indicative of individuals that were interested in the UK leaving the European Union. It was decided that, while Leave.EU did not possess much data, we could begin with an engagement for the UK Independence party, which we did. We took receipt of their membership data and survey data. There was a survey among their members of different reasons why people might be interested in the UK leaving the European Union and what the benefits or consequences might be. That survey data was matched to the membership data and used to make a model. Out of that analysis we found five different personality types.
Simon Hart
How much data are we talking about? How many people? I don’t know what the membership of UKIP was at the time.
Brittany Kaiser
I would like to follow up with documentation on that. I never took receipt of the data, so I don’t know what the final count was, but I probably have an email somewhere that states the number. I can revert with that. The analysis was drawn up into briefing material. Those briefs were used in order to talk to the executives of the Leave.EU campaign and other consultants about the insights that we had from the UK Independence party data, and that was used to give insights for their social media messaging, their campaign messaging, events, canvassing and so forth. While we were undertaking this work we were also being consulted by Philip Coppel, QC on how this data could be used and modelled.
Simon Hart
Who was he working for? Who was paying him?
Brittany Kaiser
He was contracted by Cambridge Analytica through Matthew Richardson, a barrister who I believe was secretary of the UK Independence party at the time. I believe I have turned those documents in to the Committee as well, where we were given advice on what we could and could not do. That was preliminary while we were planning on working with national data sets and undertaking a much larger survey than the one that we were provided with by the party itself. This piece of work was, in the proposal and in the contract and invoice that we submitted to them, a chargeable piece of work—not very significant in terms of our other pricing structures, but £41,500, which is something that should have been paid. I understand that in the press two days ago it came to light that Arron Banks made a donation to the UK Independence party a week before the referendum vote that was meant to cover that piece of work, but the UK Independence party decided not pass on those funds to Cambridge Analytica and pay the invoice that we had submitted.
Simon Hart
On the question of money, what was the total value—this may be somewhere in the written evidence; forgive me if it is—of the contract between Cambridge Analytica and Leave.EU?
Brittany Kaiser
We never had a contract with Leave.EU. The contract was with the UK Independence party for the analysis of this data, but it was meant to benefit Leave.EU.
Simon Hart
Okay, so who paid who how much money, and when did they pay it?
Brittany Kaiser
We submitted a contract for £41,500, and that was never paid to us.
Simon Hart
That was for the whole thing—for the company’s entire involvement in this process?
Brittany Kaiser
That was meant to be phase 1. I had two choices of next phases for engagements, which were to be either almost £500,000—it was £496,500—or, if we were to license them further data, undertake further database engineering and give them access to our software development platform Connect, which was for data management and canvassing purposes, it would have been £808,100.
Simon Hart
It may be a slightly unfair question, but how much of that, if you can remember, came in during the reporting period—the period they would have been legally required to declare that?
Brittany Kaiser
None of what was proposed in this document was actually undertaken. It was only phase 1 of my first preliminary proposal that was undertaken, and that was a total of £41,500.
Simon Hart
I suppose I am trying to put words in your mouth slightly. As far as you were aware, all the necessary declarations were made by—I thought it was going to be just Leave.EU, but you are suggesting that the relationship could have been more directly with UKIP. As far as you are concerned, the legal requirements in terms of reporting were fulfilled.
Brittany Kaiser
No, sir. I do not believe that our work was reported by anyone—not by Leave.EU, not by UKIP, and not by Cambridge Analytica.
Ian C. Lucas
You mentioned that Steve Bannon facilitated the introduction. Was that to Leave.EU or to UKIP?
Brittany Kaiser
I don’t recall. I was not copied into those communications, but I was pretty sure that he was involved in the setting up of the first meetings. That could also have been true of our meeting with Vote Leave that was then cancelled. I am not sure of the exact relationships there.
Ian C. Lucas
So when you were pitching, did you think that you were pitching to UKIP or Leave.EU?
Brittany Kaiser
I knew that I was pitching to Leave.EU, although Matthew Richardson, the party secretary, was part of those meetings almost every time. I knew that UKIP was a public supporter of the Leave.EU campaign, and that much of the work that we were being asked to design in our proposals was going to be either parallel or joint workstreams.
Ian C. Lucas
I am concerned about the transfer of data between UKIP and Leave.EU. Did you say that Mr Richardson was the secretary of UKIP?
Brittany Kaiser
I believe that he was. I don’t know if he still is.
Ian C. Lucas
He appears to have been concerned about it as well, because he commissioned a legal opinion on the issue on 17 November 2015. Did he commission it?
Brittany Kaiser
He commissioned it alongside ourselves, as Cambridge Analytica, because we wanted to understand the implications of the usage of the data that UKIP was transferring to us—what we could do with it, not only for that first project but for the remainder of the Brexit campaign engagements.
Ian C. Lucas
That’s very odd, if Mr Richardson commissioned the opinion, because his name is at the end of it, giving the opinion. So he has commissioned an opinion from himself, and then given the opinion that he perhaps wanted to hear, which was presented to Cambridge Analytica. Is that right?
Brittany Kaiser
I believe it was technically paid for and commissioned by Cambridge Analytica to Philip Coppel, QC. Matthew Richardson was a part of it.
Ian C. Lucas
Just for clarification, Matthew Richardson and Philip Coppel, QC have signed this opinion.
Brittany Kaiser
Yes.
Ian C. Lucas
Now, I am a lawyer, but I am not an expensive lawyer—I’m not a QC. My understanding is that, for example, the Labour party wouldn’t be permitted to transfer data to Britain Stronger in Europe, because they are different organisations. Is that your understanding, too?
Brittany Kaiser
Yes. I would say that it’s probably not normal professional practice to be one of the people giving a legal opinion to your own organisation.
Ian C. Lucas
Let’s just talk about transferring data. Do you think it is lawful—you work in this field—to transfer data from a political party to a separate campaign in connection with leaving the EU?
Brittany Kaiser
Not without the specific opt-in of the individuals whose data is concerned.
Ian C. Lucas
Do you know what this opinion was used for?
Brittany Kaiser
This opinion was used for the phase 1 small piece of work that we undertook.
Ian C. Lucas
So it wasn’t presented, for example, to Arron Banks.
Brittany Kaiser
I’m not sure. I do know that Matthew Richardson sent it to me and Julian Wheatland for us to understand, or at least to give it to our data scientists and our legal team to understand what could be undertaken with that data. There are quite a few more implications to that opinion. During our analysis of the UKIP data set, we were asked by Arron Banks to consider combining the three work streams that we were talking to him and his colleagues about. Those work streams were for Leave.EU, the UK Independence party and Eldon or GoSkippy insurance. We made a fourth proposal—
Ian C. Lucas
That’s a private company.
Brittany Kaiser
Yes, it is.
Ian C. Lucas
Which is owned by Mr Banks, just for clarity.
Brittany Kaiser
I believe so, yes.
Ian C. Lucas
So it appears that he was transferring data between a private company, a political party and a campaign to leave the EU.
Brittany Kaiser
I have two points on that. One is that he asked us to design a strategy, where we could work with Leave.UK, UKIP and Eldon insurance data together. You will see that the fourth proposal that I turned in to Arron Banks shows a diagram: complementary work streams for UKIP, Leave.EU and Eldon insurance. We were asked whether savings could be achieved by running these three programmes together, instead of separately. I believe that that legal opinion is one of many that would have had to be undertaken if we were going to be contracted to do this piece of work. We were never contracted to do this piece of work, although from other evidence that has been given by individuals such as Dr Emma Briant, and by Arron Banks and Andy Wigmore themselves to the press and others, upon knowledge of recording, I believe that this work was carried out, just not by Cambridge Analytica. It is of note that within the legal opinion that I provided to you, in paragraph 64, Mr Philip Coppel specifically says that the data under no circumstances should be transferred abroad. That is specifically because Cambridge Analytica is a foreign company. That would need to be written out in the contract and all this data would have to be analysed within the United Kingdom. In regard to this proposal and work that I believe was undertaken with or without us, there are a few specific concerns of legality. I believe those concerns are also concerns of the Committee. The first is in relation to breaches of electoral law for chargeable work, some of which I did, which was never paid for and was unreported to the Electoral Commission. Secondly, I have evidence from my own eyes of possible breaches of the Data Protection Act concerning the commercial and personal data of individuals in the Eldon Insurance database, and possibly the UKIP database, being used for the benefit of the Leave.EU campaign. I am just going to expand on that slightly. I visited the Eldon Insurance and Leave.EU headquarters, which was the same building, with the same staff. When a senior data scientist and myself spent time with their phone bank, I was told by the people using the phone bank that the individuals they were calling were out of the insurance database. Thirdly, there is the question of whether the personal data of British citizens was sent abroad, specifically to Mississippi for processing by Arron Banks’ firm, Big Data Dolphins, contrary to the Data Protection Act and to the advice of Mr Philip Coppel, QC.
Ian C. Lucas
We also know that Cambridge Analytica shared data with AIQ. There was a contract, but AIQ is a Canadian company, so presumably that was in contradiction to this provision as well.
Brittany Kaiser
I can infer that from the evidence that has been submitted by others, but because I do not have any knowledge of that myself, I couldn’t expand that far.
Ian C. Lucas
Do you know whether any other private businesses shared data with Cambridge Analytica through UKIP?
Brittany Kaiser
No, although we were exploring matching data from Experian, which is a big data vendor that we work with in the United States. We were exploring licensing that in the United Kingdom for matching to the UKIP database and to any database of individuals that would have been collected by Leave.EU.
Ian C. Lucas
Do you know whether Wetherspoons ever shared any information with UKIP?
Brittany Kaiser
No, I don’t know. If there is evidence of that, it would be interesting.
Ian C. Lucas
Yes, it would be very interesting. You talked about the SCL Group, including Cambridge Analytica, very much as one organisation. The picture that I have is of data being shared freely between all of the companies within that group. Is that accurate?
Brittany Kaiser
I am not sure which entity actually owned the database, but most of the staff within Cambridge Analytica and the SCL Group had two business cards and two emails. We worked for SCL Group and Cambridge Analytica in parallel.
Ian C. Lucas
When I spoke to Mr Nix, he said that SCL could transfer data to Cambridge Analytica but, for security reasons, Cambridge Analytica could not transfer data to SCL, because SCL had security clearance. Does that ring any bells?
Brittany Kaiser
Ah, yes. It is a very complicated structure.
Ian C. Lucas
Some might say deliberately so.
Brittany Kaiser
Possibly. Originally, the SCL Group was the parent company for all these other divisions, including Cambridge Analytica. Now, once Cambridge Analytica became a very popular global brand, Cambridge Analytica then subsumed most of those companies and divisions and the SCL Group became just our defence company, SCL Group or SCL.gov, based in Arlington.
Ian C. Lucas
But Cambridge Analytica had access to the SCL database?
Brittany Kaiser
I believe once the company was rebranded that there should have been, at least, no access between those two companies except for the contractual obligations that you are discussing, although I do not believe up until that point that there was a difference between the SCL Group and the Cambridge Analytica database. That would be slightly concerning when we are talking about the difference between commercial and political and defence work. I have had those concerns in the past because when I joined the company—I didn’t remember this, but upon reflection—I found documents from Nigel Oakes, the co-founder of the SCL Group, who was in charge of our defence division, stating that the target audience analysis methodology, TAA, used to be export controlled by the British Government. That would mean that the methodology was considered a weapon—weapons grade communications tactics—which means that we had to tell the British Government if that was going to be deployed in another country outside the United Kingdom. I understand that designation was removed in 2015.
Ian C. Lucas
Can I ask a question about the invoice issued by Cambridge Analytica to UKIP? Do you know why that was issued to UKIP as opposed to Leave.EU?
Brittany Kaiser
Yes, Leave.EU did not have any data yet at the beginning, because it was just a campaign, so we went to UKIP—they were participating in our meetings jointly with Leave.EU executives—in order to find some data to begin working on. That was the UKIP membership data and the survey data provided to us for the phase 1 piece of work.
Ian C. Lucas
But the intention was—the work that you were doing was relating to a pitch to work for Leave.EU?
Brittany Kaiser
Yes, sir.
Chair
Just to clarify, what you said about the target audience analysis work—that was part of the proposal put together for Leave.EU. That is correct?
Brittany Kaiser
Yes.
Chair
So what you are saying is that the proposal was for Leave.EU to use what you call “weapons grade communications techniques” against the UK population?
Brittany Kaiser
I believe that in 2015 target audience analysis no longer had an export designation. I don’t know at what point that was dropped off, but I believe that it still existed in 2014 when I joined the company.
Chair
What do you mean by “export designation”?
Brittany Kaiser
I don’t actually know the complexities of this, but I believe that every year or every few years something that is export controlled has to be renewed, and that was not renewed. I don’t know what you need to qualify for that or what the implications of that are.
Chair
Basically, these were techniques where previously you would have needed to inform the authorities that you intend to use these in the UK—would that be correct?
Brittany Kaiser
That is what I was told by Nigel Oakes, yes.
Chair
And when that designation was removed, it could be done without informing the authorities, but nevertheless, it is still the same product. A product that was considered so dangerous that its use was restricted: nevertheless the intention was that that strategy would be used in the referendum campaign?
Brittany Kaiser
The interesting thing is that the target audience analysis uses a lot less data than psychographic micro-targeting, which is what we use in the United States.
Jo Stevens
Can I ask a few questions about the Eldon Insurance and the Leave.EU stuff? But first, this invoice for the work that you did on phase 1 which you have described, can you remember the dates between which you did the work—when it started roughly and when it finished roughly?
Brittany Kaiser
Yes, I believe we took receipt of the data in late October, although I wasn’t present in the office; I was in DC when a member of the UK Independence Party showed up to SCL Group’s HQ in Mayfair with a big computer stack. That was their form of transferring data—they actually brought an old desktop into the office physically. We took the membership data and the survey data off that computer and began analysing it. Usually that type of matching, modelling and analysis takes us two to three weeks. I don’t know how many days our data scientists actually spent on that, but we estimated that the work was worth £41,500 based on the time spent—
Jo Stevens
Sorry to interrupt you. Is that on market rates or mates’ rates?
Brittany Kaiser
No, that was generally what we were charging at the time. That was in preparation for the public press conference that we did on 18 November. There would have been a couple of weeks of work leading up to that. I spent 18 November, the day before and the day after in Millbank Tower, the Leave.EU London offices, briefing everybody who was working at those offices on my findings, and working with the communications and social media teams on how you could use that insight to craft different communications campaigns.
Jo Stevens
When did the work that you were doing on that phase stop?
Brittany Kaiser
It would have been within a couple of days after that. The rest of it was putting together this joint proposal, which took quite a few weeks to do, to understand the complexities of how the data could possibly be combined, although we would probably have had to have a few more legal opinions drawn up to understand that we could do this, or have undertaken opt-in campaigns with the individuals whose data were contained on those databases. The last time I actually engaged with any individuals in Leave.EU myself was at the end of November, although in January this proposal was presented by Julian Wheatland and Alexander Nix to, I believe, Arron Banks, Liz Bilney, Andy Wigmore and Matthew Richardson.
Jo Stevens
Thank you. I think you said earlier regarding this proposal that although you did not undertake this work for Arron Banks, it has subsequently been done. Do you know who has done that work?
Brittany Kaiser
Both Arron Banks and Andy Wigmore have told multiple individuals that they took my proposal and copied it, and that they created their own Cambridge Analytica which they called Big Data Dolphins, in partnership with the data science department at Ole Miss University—the University of Mississippi.
Jo Stevens
So presumably—stop me if I am wrong—implementing this proposal involves, based on what you have said, taking data from UK customers of Eldon Insurance, GoSkippy Insurance, UKIP members and Vote Leave data.
Brittany Kaiser
Leave.EU data.
Jo Stevens
Sorry—Leave.EU data. And that goes over into the United States?
Brittany Kaiser
Yes.
Jo Stevens
Do you know whether or not Cambridge Analytica have a continuing relationship, other than what you have already described—I am particularly interested in Julian Wheatland’s involvement in this—with Arron Banks?
Brittany Kaiser
I don’t believe so. He never paid us for any of our work, never accepted any of our subsequent proposals, and has bragged about copying our company and recreating it for himself, so I do not believe that there would be a relationship left to have.
Jo Stevens
Okay, thank you. I know in your written evidence you mention the fact that there was a discussion, or something was talked about with Arron Banks, about the use of the Eldon and GoSkippy data. To quote from your written evidence: “Arron Banks said, ‘It’s my data’”, so he is using customer data from GoSkippy and Eldon Insurance for other purposes.
Brittany Kaiser
That is what I saw with my own eyes when I spent a day at the Bristol headquarters. I was brought there under the understanding that I was going to Leave.EU’s headquarters. It became very quickly obvious that this was actually an Eldon Insurance headquarters, which is also the parent company of GoSkippy—maybe that is one of their brands. All the staff there were employed by an insurance company and almost none knew anything about political campaigning. A lot of what I said to them and what I presented to them was novel. They had never heard of any of these tactics. They had never participated in politics before and openly said, “I work for an insurance company, so—help!”
Jo Stevens
So why were you going to see the people in the call centre? Were they going to be working on the Leave.EU campaign?
Brittany Kaiser
They were actively working on the Leave.EU campaign when I was there.
Jo Stevens
And how many people were working in the call centre?
Brittany Kaiser
The call centre was about the size of this room—maybe a little bit smaller—with at least, probably, 20 active callers. They all had a computer and a headset, making calls that I assume were normally sales calls or calls for customer assistance or advice, and instead they were calling people to undertake a survey. They were talking to those people about their interest in leaving the European Union, and issues around Brexit, and Leave or Remain. We were sitting with those individuals specifically because we said it was unlikely that the survey that they were undertaking was up to scratch with what we would need for modelling. So, we sat there to understand what data they were using, what surveys they were undertaking and how we could improve on that, so that that data could be used for nationwide modelling. I was under the impression, by what they told me, that every single individual that they were pulling up to call was actually a lead or a current customer of Eldon Insurance or GoSkippy.
Jo Stevens
And do you know for how long they’d been making those calls? I think you visited in November, didn’t you?
Brittany Kaiser
I believe I visited in November, yes; I would have to double-check that. But it would have been end of October/early November. And they had probably already been doing that for a month or two, because we started having these conversations with them in September.
Jo Stevens
Right. And this was a full-time call centre?
Brittany Kaiser
Yes.
Jo Stevens
With 20 people in it?
Brittany Kaiser
At least.
Chair
I just want to clarify the dates. So, the date of the legal opinion, which you presented as written evidence, the date that was signed off was the day before the press launch for Leave.EU. Are those two things connected? Was this commissioned just to check that the methodology or the way of working that would be at the press launch was in fact legal?
Brittany Kaiser
Yes, before I presented it to members of Leave.EU.
Chair
So there was a concern, clearly, within the organisation that that opinion needed to be sought before the official launch, because there was some ambiguity?
Brittany Kaiser
The idea was that directly after the press launch we would begin phase 2 of the proposal and we needed to understand what the legalities around beginning that second phase of work would be.
Chair
But as Mr Lucas mentioned earlier on, the existence of that opinion would seem to suggest that there clearly was a working relationship, which was being tested here.
Brittany Kaiser
Of course. We knew, though, that with the data that was given to us by UKIP for phase 1, we were allowed to model internal data for UKIP, but I wouldn’t have been able to present that to Leave.EU without that legal opinion nor suggest that I could undertake phase 2 of the proposal, or begin contractual negotiations over that.
Chair
There were other organisations, obviously, involved in campaigning on the referendum question, aside from those that got the official designation. That includes organisations like Grassroots Out. Were they involved? Was data from Grassroots Out accessible to the work you were doing, with the intention to use that, too?
Brittany Kaiser
I was no longer in conversation with any of these individuals when they became Grassroots Out. That was way later, after Arron Banks made it very clear that none of our contracts were going to be signed or our bills paid.
Chair
From the email evidence you’ve supplied to us, and you refer to it in your statement as well, you clearly believe that Arron Banks had approved the budget that was presented?
Brittany Kaiser
Yes, or at least Liz Bilney, who was the CEO of Leave.EU at the time and his right hand—she travelled with him everywhere and was in charge of multiple of his businesses, or business engagements, all over the world. So, she had given us the green light, in writing and verbally, and therefore I asked Julian—at the time, our CFO—if we could have the go-ahead to begin work. He was in charge of our finances and if he said, you know, on a gentlemen’s agreement, “Go ahead, while we wait for the legal opinion and the contract”, then that was my permission to go ahead.
Chair
The preparatory work that was done was analysing the data available at Eldon and through GoSkippy to see what information they had and how that could be used. I imagine that that data in and of itself—it needs to be processed in some way to make it useful to a political campaign.
Brittany Kaiser
It would have to have been, yes. I was just looking at what was available. I had seen that a data scientist who was there with me was looking to see what was there, so that we understood how we could process it.
Chair
Just for the record and the Committee, it might be helpful if you could walk us through what the recommendation was. How do you process that data to make it useful to a political campaign? What do you need to add to it to make it work?
Brittany Kaiser
So, by nature of data science, the more that you know about an individual, the more accurate an algorithm you can build to predict their behaviour, their opinions, their personality, or their emotions on a certain issue. We would have taken the data that we had on their commercial and political activities, and the most data that we could pin back to their personally identifiable information. In the US we would use first name, last name, zip code or post code, and a birthdate to ensure that we were talking about the right person. Then you continue to append that data to get as deep a data profile on an individual as possible. On top of that, once you have data profiles on as many people across the country as possible, you would design a quantitative survey. I believe we said that we were going to do two surveys of 40,000 individuals, and that is used as a training set. So when we know what 40,000 people would answer on specific questions about political ideology and issues, then we can build a model that gives us predictions about what everybody else would say.
Chair
And then, presumably, would you then use tools like Facebook to help identify people from the training set, and then invite other people like them, who Facebook can help you identify?
Brittany Kaiser
Exactly correct.
Jo Stevens
The Eldon and GoSkippy insurance data—did they give you all the customers’ data? Do you know how many people’s data you received?
Brittany Kaiser
I never received any data from the insurance company. We were only privy to what data points existed in its database. It probably told us how many individuals were in its database, and I can look through my emails if you are interested.
Jo Stevens
It would be really helpful to know that.
Brittany Kaiser
Of course.
Chair
How was it planned that Facebook data would be used? Would that be gathered during the course of the campaign and added to the training set, or was there Facebook data that was accessible already?
Brittany Kaiser
When I sat with their social media team we looked at undertaking surveys on Facebook—it was similar to what Cambridge Analytica or the SCL Group had done in the past. I believe that they began to do those surveys on Facebook without us, from what I understand from what they have said to the media, but usually Facebook would just be used in terms of their targeting tools. You upload a certain database and that matches individuals within Facebook, and then you can target a lookalike audience to those people, and find even more individuals who are as similar as possible to the individuals that you already know are likely to be interested in leaving the European Union.
Chair
I believe Leave.EU said that a lot of the work it did after it didn’t get the designation was through Facebook pages, and getting people to sign up and register interest on the Leave.EU website. Presumably, it was getting contact emails and addresses from Facebook users that they could then match against their data set to identify people in the data set against their Facebook profile and use that as the basis of targeting them and other Facebook users who had similar favourable opinions.
Brittany Kaiser
That should have been the way that they were doing it, but it was an undertaking by Cambridge Analytica so I can’t say for sure what the exact methodology was.
Chair
For our benefit, to be effective in this space, how big a working set do you need to use that to create the basis for targeting the whole country in terms of voting?
Brittany Kaiser
I believe that the algorithms that we built for Cambridge Analytica used at least 400 variables, and it would be very difficult to have thousands of data points on individuals in the United Kingdom. Therefore the models would be scientifically less accurate than what you can produce in the United States, for instance. I am not a data scientist, so I would not be able to say the minimum number of data points that you would require, but I do know that, for instance, there are other data science consultancies that will build an algorithm with 50 variables.
Chair
But how many individuals do you need to have profiled as part of your working set in order to use that as a model to help you to target effectively others who are like them, and through them reach most of the electorate in an election? How big a sample set of individuals do you need to work from in order to achieve a successful campaign?
Brittany Kaiser
You would want millions in a country the size of the United Kingdom, at minimum, and to be talking to tens of thousands of people to complete surveys. But it depends what level of accuracy you would be interested in having. Some people are fine with less than 95% confidence rate, for example. It depends how exact you need to be.
Chair
But you still need a big group.
Brittany Kaiser
Yes, but that can be achieved by licensing data sets from a company like Experian, where that is based on the entire country’s credit profile. In the United States, that gives us basically everybody above the age of 18—240 million Americans—just from going to one source.
Chair
I would like to follow on from that and your experience of the Ted Cruz campaign. There was a lot of talk in the referendum about the value of door-to-door canvassing techniques and the ability to feed that data back into the system, but is the purpose of that not just to support voter turnout on election day, with people going around knocking on doors—historically what canvassing has been there for—but to enrich the overall data set that is being used during the campaign to target people with digital messages?
Brittany Kaiser
I would say that these days it is much more important for data collection than turnout specifically. Turnout messages can be targeted to those individuals without a person going to speak to them in person.
Ian C. Lucas
We know that there was contact between Facebook and Cambridge Analytica about the use of data. I think it was in 2015, from memory. Did you know about that at the time?
Brittany Kaiser
Yes. The first time that I heard anything about Facebook data, in writing or even in a personal conversation, which didn’t have to do with those personality quizzes, was late April 2015. Facebook had announced to all of its clients, and therefore to its clients’ clients, that it was going to close its personal data API, which allowed apps and their clients to have access to this data. Before the closing of this access, my chief data officer gave me and two of my colleagues a list of thousands of different groups on Facebook and asked us to choose 500 of them that we thought having information of the individuals who like those groups would be useful in our modelling for our new commercial business, which I was helping grow. So we chose 500 of those groups and turned that into what I suppose is now GSR, although I wasn’t aware of who we were getting them from at the time, in order to get the data of those individuals who liked the groups that we chose. We received that data and turned in that request on 5 May, 2015. If Facebook actually closed that API in April, either a company contracted with Facebook was contravening their legal obligations or they were selling us old data—I am not sure. Seven months later, in December 2015, an individual from Facebook contacted Cambridge Analytica to ask if we still had any Facebook data sets that we had received from Professor Kogan or GSR. We replied that yes, we had, but we had an agreement with GSR to have that data. Now, Facebook’s agreement with GSR apparently said something different, that it wasn’t meant to have the data and therefore any of its clients were no longer meant to have the data, so we agreed to delete it, but our chief data officer specifically said, “We haven’t contravened any contractual obligations, but at your request we are happy to delete it.” That deletion was apparently made in January 2016, which is quite a long time after no one was supposed to have access to those data sets any more. I find it incredibly irresponsible that a company with as much money as Facebook—until a couple of weeks ago it was worth $500 billion—had no due diligence mechanisms in place for protecting the data of British citizens, American citizens or their users in general. Something else of note is that in March 2016 I was copied into a back and forth conversation—
Ian C. Lucas
Before you go on, how do you know that in January 2016 the data was deleted?
Brittany Kaiser
I don’t know that it was deleted. I know that our chief data officer said it was deleted—to Facebook.
Ian C. Lucas
Okay, thank you.
Brittany Kaiser
That was forwarded to me willingly in order to help deal with the questions from our clients around PR issues, because The Guardian had reported on this issue. In March 2016, six or eight weeks after our chief data officer said that those data sets were deleted, I had an email from one of our senior data scientists responding to a question that said that we were actually using Facebook-like data in our modelling. So that seems quite strange to me: if we had deleted all of the Facebook data sets, how we were still using that for modelling in March.
Ian C. Lucas
So your suspicion is that someone might have been continuing to use the Facebook data in March 2016.
Brittany Kaiser
Yes. That is only upon reflection.
Ian C. Lucas
Okay. I understand that. Furthermore, on Facebook taking steps to ensure that the assurance they were given by Cambridge Analytica that the information had been deleted, you are not aware of them actually checking that?
Brittany Kaiser
They did not come and check that at all. That was the final communication, I believe, on the topic.
Chair
I know Paul Farrelly wants to come in, but I want to finish off on Facebook, if that’s okay. Facebook said that it believed that Aleksandr Kogan had access to data profiles on up to 1 million people in the UK. Did you ever get any sense of how many UK profiles were part of the data set that Aleksandr Kogan had collected?
Brittany Kaiser
No. I didn’t even know who Professor Kogan was when I joined the company. Once he was mentioned in an email, where he was referred to as Alex Kogan, a psychologist that we work with. I actually did not know much about him or GSR at all, so a lot of this is new information to me, unfortunately.
Chair
But with Cambridge Analytica, did you get any sense of how much Facebook profile data it had related to UK users of the platform?
Brittany Kaiser
I was told that we never had any data on UK citizens, because I was constantly pitching for commercial contracts here and I did not really have much to back up my proposals that said that we could undertake similar kinds of work as we did in the United States.
Chair
You said that you were asked to look through lists of Facebook groups and choose 500. Why groups as opposed to pages or profiles?
Brittany Kaiser
I don’t know. I suppose there was just a good source of data. If people liked this particular group, it was a topic that they were interested in. That ranged from restaurants to pop stars to anything you can imagine. It was personal interest.
Chair
Are groups more open? Is it easier to see who has joined various groups if it is an open group rather than on a page?
Brittany Kaiser
That’s possible, yes.
Chair
It’s interesting. It’s an issue we’ve raised with Facebook before. Next week I am sure we will do the significance of Facebook groups as a means of analysing user behaviour and disseminating information. In their official responses to us, they always downplay the significance of groups and focus more on pages and on profiles. I just find it interesting that that was particularly highlighted in your case.
Ian C. Lucas
Can I follow up on the groups question? Does the data relating to the groups only relate to people who like the group or is it people who have visited the group as well?