Digital, Culture, Media and Sport Committee
Sub-Committee on Online Harms and Disinformation
Oral evidence: Anti-vaccination disinformation, HC 1049
Thursday 17 December 2020
Ordered by the House of Commons to be published on 17 December 2020
Members present: Julian Knight (Chair); Kevin Brennan; Steve Brine; Philip Davies; Alex Davies-Jones; Clive Efford; Julie Elliott; Damian Green; Damian Hinds; John Nicolson; Giles Watling; Mrs Heather Wheeler.
Greg Clark, Chair of the Science and Technology Select Committee, was in attendance.
Questions 1 - 151
I: Theo Bertram, Director, Government Relations and Public Policy EMEA, TikTok; Iain Bundred, Head of Public Policy, UK & Ireland, YouTube; and Rebecca Stimson, UK Head of Public Policy, Facebook.
II: Professor Tom Rodden, Chief Scientific Adviser, Department for Digital, Culture, Media and Sport; Sarah Connolly, Director, Security and Online Harms, Department for Digital, Culture, Media and Sport; and James Sorene, Deputy Director, Covid-19 Vaccine Communications, Department of Health and Social Care.
Witnesses: Theo Bertram, Iain Bundred and Rebecca Stimson.
Q1 Chair: This is the Digital, Culture, Media and Sport Select Committee, and this is a sub-committee hearing into disinformation and fake news. The subject today will be disinformation related to the vaccine. We will be joined by Theo Bertram, Director, Government Relations and Public Policy at TikTok, Iain Bundred, Head of Public Policy at YouTube, and Rebecca Stimson, UK Head of Public Policy at Facebook in our first panel. In the second panel we will have scientific experts and members of the DCMS. We are also joined today by the Rt Hon Greg Clark, the Chair of the Science and Technology Select Committee, who is joining us to pose some questions.
I checked before to see if there are any interests. No members have said they have any interests to declare, but if one pops into your head suddenly, please do say so before you ask your question, if that is possible. Thank you.
My first question relates to all three of you. How are you tackling vaccine disinformation? What is new about your approach? What results can you share with us? Rebecca Stimson of Facebook, can I go to you first?
Rebecca Stimson: Thank you, and thank you very much for the opportunity to be part of this conversation today. We have had misinformation and vaccine-related policies for quite some time, but the particular circumstances of this year have brought new emphasis to those policies and how they have been running on our platform.
As the Committee’s own report found back in July, this is a multi-faceted problem. We have a range of tools at our disposal to try to address it. We broadly think about misinformation in two brackets. One is misinformation that can cause imminent harm and real-world danger, and other types of misinformation. I am sure we will come back to unpacking that definition in the conversation today.
The way we address those two types of misinformation can vary, from removing them, downranking them to make them harder for people to see, and labelling them via our fact-checking programme; there are a host of other tools, including addressing fake accounts. We have a big toolkit. Obviously Covid has been an enormous learning opportunity for us to dial some of this up and think about what works. The one example I would call out that has changed for us is growing our understanding that it is just as important to provide people with access to the right, authoritative information as it is to find and remove the most harmful kind of content. We have learned some big lessons working with the health experts this year, and really key is that the blanket removal of information will not necessarily help the situation. I am happy to talk about that a bit more, but that is basically what we have done pre-Covid, and also during the course of this year.
Q2 Chair: Have you introduced any specific changes in recent weeks and months? Is there a specific area that you can point to where you can say, “Okay, we have changed our approach in this way”?
Rebecca Stimson: There are a couple that I would point to. One is that we have further refined our advertising policy. Already we were not allowed to advertise something that was fake information about a vaccine, but we have expanded that to include adverts that might discourage you from taking a vaccine, adverts that might fall short of that test of “Is it fake?”, but might be scaremongering, or an example that otherwise falls short. We have expanded our policies.
Another example is that we recently revised our Correct The Record tool. I believe that the previous time we spoke to this Committee, we talked about rate back—if a person has shared a piece of information that is subsequently found to be false, they get a notification—but based on the feedback from the session and the Committee’s finding, we have revised that. Now, if it is related to Covid information, it is also if you have clicked on it or liked it; it is an expanded group of people who will get that notification.
Also the notification itself is much better. It shows you a thumbnail of the thing that you shared—because obviously it could have been some time ago—and explains why, and what has happened, and it links you to the source of information that now debunks the thing that you previously shared. As I said, that builds on our approach, making sure that providing information is just as much a part of this as removal.
Q3 Chair: Rowing back a bit, you mentioned how you were looking to ensure that authoritative sources had greater prominence—my word, not yours—in people’s feeds, so that effectively they drown out any sort of disinformation/misinformation. Has that been achieved through specific changes to algorithms on your site?
Rebecca Stimson: We already use our algorithms to amplify the best, most positive content. In a very American expression, we call that “meaningful social interaction”. Harmful content, which can be a range of things from hate speech to the kind of misinformation this Committee is interested in, will be either downranked or removed as effectively as we can, although I think we recognise that that is an ongoing challenge and we are certainly not able to find and remove it all. Our algorithm works so that we amplify the good things that people and their friends and families want to see in the projects they engage with, and either downrank or remove the bad.
The way we have been linking to authoritative sources of information is, for example, as I mentioned, through the Correct The Record tool. If you search for a particular Covid-related term, you get a pop-up that directs you to the NHS and other authoritative sources of information. We have built bots in WhatsApp that send you to them. It is a two-pronged approach. One is about the kind of content we serve, then proactive interventions to direct people to those sources of information.
Q4 Chair: Let me press you on this point though. Are you doing anything algorithmically different in relation to Covid disinformation than you were doing before? You have set out the Correct The Record tool, and you have talked about how you already, through algorithms, manage newsfeeds. Has there been a conscious decision to do something else?
Rebecca Stimson: There has, in the sense that as we learn more and more about Covid and the situation on the ground changes—for example, a vaccine gets approved—there is not a change in our algorithms, because our algorithms were already prioritising the right kind of content and trying to minimise the worst kind, but we have added information about Covid to that process as it has become confirmed by health authorities and the situation has changed.
Q5 Chair: Basically and effectively, you are saying that it is piggybacked on an existing algorithm?
Rebecca Stimson: That is right. We were already addressing this kind of content and obviously as things changed with Covid, we added into that system.
Q6 Chair: I will turn to YouTube now and Iain Bundred. Welcome back, Iain. It is the same question about tackling vaccine disinformation that I put to Facebook: what is new about your approach, and what results can you share with us?
Iain Bundred: Similarly to Facebook, we have looked at two particular things: removing misinformation and conspiracies, and raising authoritative content. Now we are also looking at how we can positively campaign and build public confidence in the vaccine. That is still in progress, as you would expect, but I can talk a little bit about that as well.
On the removal side, between February and November we removed 750,000 videos globally. I think I mentioned this at the last hearing: in October we introduced a new specific Covid anti-vax policy. I do not have a data update yet about how that is going, but machine learning is now addressing it, and we are removing more and more videos as time goes on.
We have done a lot to our product on promoting authoritative sources. We introduced the Covid news shelf. Most of this was done earlier on in the pandemic, I should say, but we have seen an important increase in the amount of watchtimes on authoritative sources. I think watchtimes in the UK increased 6.5 times, and globally we have seen I think a 90% increase in authoritative watchtimes in the first half of the year. That, together with the work that I have already updated the Committee about—
Q7 Chair: I am sorry to cut across you. The connections are not that great, so it is a little bit difficult to hear you. Are you saying, effectively, that in the UK, through YouTube, people have been much more likely to key into authoritative information compared to, for example, globally? You talked about 6.5 times—that is 650%—compared with 90% in the rest of the world.
Iain Bundred: Yes. I just checked and it is 85%. People come to YouTube for lots of reasons; news and politics, frankly, are a low part of that, but as I told the Committee a few weeks ago when we talked about the public service broadcasting, interest in information and the content from news outlet services is growing fast. It is growing fast globally, but I would say that in the UK it is growing faster.
Q8 Chair: I will turn now to Theo Bertram from TikTok to put the same question as I have put to Facebook and YouTube about how you are tackling vaccine disinformation. What is new about your approach, and what results can you share with us?
Theo Bertram: Thank you very much for setting up this panel. I think it is timely and important. Thank you also for including TikTok in particular. As I think I said to you when I was here in September, right at the start of the year, in January, we were putting information from the WHO on the platform. In February we created a Covid hub to try to direct our users. In March we introduced a specific Covid policy before the UK got into lockdown 1.
The others talked about it as a two-pronged approach. We think of it as three-pronged. If I understood Rebecca and Iain, they were saying they tackle disinformation and promote trusted content. We said there is an extra stage, which is tackle misinformation, get the trusted content on the platform and then direct users to it. For us, getting that trusted content on the platform is key.
As for what we have done across those areas more recently—all through the year we have been working on that; we keep trying to iterate, improve and get stronger—in October we began work with Team Halo. I do not know if you are familiar with Team Halo. It is run by the Vaccine Confidence Project at the London School of Hygiene & Tropical Medicine and is about getting reliable, trusted scientists and doctors on to social media to spread trusted information. Some of those have been great on the platform and have helped to reach our community.
We have had that policy around medical misinformation since March, and we started to get people on the platform to talk about the vaccine, but what we are doing this week is redirecting people—directing people to it. At the moment we have a system that is focused around Covid in general. When you create a video and we identify it as a Covid-related video, we put a sticker on it, and that sticker directs people to the Covid hub. This week we are creating a vaccine-specific version; we will direct people to a vaccine-specific hub. From Monday, when you create a video, instead of just trying to identify if it is a Covid video, we will try to identify if it is a Covid vaccine video, and if so, will put a sticker on it directing people to the Covid vaccine hub.
Q9 Chair: Obviously you have seen that Twitter has come up with some new policy in relation to the Covid vaccine. Twitter cannot appear today but has written to us with the details.
You are saying that effectively you have updated your Covid policy to give it this sort of route away for Covid vaccine information.
Theo Bertram: We have had a medical misinformation policy since March, and that policy said that if you are saying something that is false or harmful about Covid, about treatment, about medicines, or about vaccines, we will take it down. That policy has been there all the time. What we have moved to this week is about pushing people towards specific authoritative vaccine sources, rather than just pushing them towards Covid sources. It is a not a new vaccine misinformation policy; that is something we have had in place since March.
Q10 Chair: I did ask all three of you specifically about sharing the data, letting us know precisely what results you have seen from your work against Covid disinformation. I know, Iain, that you mentioned the number of videos taken down, but again they were relatively old figures. Do you have anything new to say to the Committee about your success, in very simple statistical terms? Theo, you are nodding your head. Do you have anything to share with us?
Theo Bertram: Yes. I will briefly recap the numbers I gave you when I was here last time and then I will give you the numbers for October and November, which are the complete months since I was last here. When I was here last time, I told you that we had had 600,000 to 700,000 videos created in the UK stamped with a Covid sticker and we removed 15,000 of those videos over that six-month period for medical misinformation.
In the period of October and November we have had 77,144 videos tagged with Covid. Just to give you a sense of the scale, in the first half of the year it was roughly 100,000 per month; in last couple of months it has been more like 40,000 per month. Those are the Covid videos. The numbers suggest an overall decrease in interest in the issue. The number of medical misinformation video violations was 225, which is probably slightly higher as a proportion of those videos, but it is still relatively low.
Just to give you a sense of how many accounts we took action against, in the first six months of the year we took action against 1,500 accounts for Covid violation, and in the last two months we took action against 1,380 accounts, so you can see the level of action is increasing.
Q11 Chair: That is interesting. Obviously this was before you made your new policy when it comes to shunting certain videos off into a Covid vaccine route, so to speak. In that respect, I think what I have worked out is that you have taken down 0.6% of total videos due to Covid. However, the number of actions against accounts is proportionally far higher. You have done in two months, compared with six months, 1,500 compared with 1,380. I am very conscious of time, but very briefly, why is that there are fewer videos being taken down but more accounts being acted against?
Theo Bertram: I also asked this question when they gave me the data, so you are spot on. We remove accounts for a number of reasons, so it might not just be the video. Some people might create a profile, have an account name, or they might have something else in the content that violates the Covid policy and that is why we would end up with more removals than videos.
Q12 Chair: Thank you. Finally, to Facebook and YouTube at the same time, you have heard what TikTok has said; it has very precise figures for the last couple of months in terms of Covid disinformation. I am not going to ask if you have collated this, but do you have that information to hand? It was the first part of my question, and I noticed that I was getting quite old data. Is there anything empirical that is up to date and that you can tell us about, in terms of the number of pieces of Covid disinformation that you have had to take action against, and what you believe is happening? Do you have anything extra to say beyond what you said before?
Iain Bundred: Yes. I just want to be clear: I gave a number that is right up to date, up to the end of November—750,000 videos removed. The last time I appeared, I gave you the number for October, which was 700,000, so it is 50,000 for the month of November.
Chair: Thank you. Rebecca?
Rebecca Stimson: I have just a selection of our latest figures. On fact-checking, between March and October, 167 million pieces of content have been labelled via our fact-checking programme. In the same period of time 12 million pieces of content were removed from the platform entirely because of present, real-world imminent harm. We have driven 3.5 million UK citizens to NHS data sources through our Covid hub and displayed information to 40 million users, which is almost the entirety of our UK usership. One other statistic: on the warning labels, in terms of effectiveness, we see a 95% rate of people not clicking through to read content that has had a warning placed on it.
Chair: Thank you, Rebecca. I notice that those numbers are from March to October. We are now concerned with Covid vaccine disinformation, which obviously has started to spike during the last month or so. Before I call on John Nicolson, I would ask if you could go away and write to the Committee as quickly as you can, please, to get that information to us, because we think it is very important that we have up to date data, such as TikTok has obviously provided. If you could do that, it would be helpful.
Q13 John Nicolson: Mr Bertram of TikTok, can I begin with you? We are in the middle of a worldwide pandemic; 1.7 million people have died, 66,000 in the UK alone. At last we have a vaccine, yet you allow anti-vaccination fanatics to spread lies on your platform. Why?
Theo Bertram: That is not accurate. We do not allow anti-vax misinformation on the platform.
Q14 John Nicolson: Let me give you an example. Do you know who Olivia Madison is?
Theo Bertram: No, I don’t.
Q15 John Nicolson: She has 606,000 followers on your platform. I looked her up last night just to see, because I suspected you would tell us you were taking down all these videos. She is very beautiful and what she does is utterly wicked. The first thing I came across that she had to say was, “Vaccinations contain foetuses”, and—I quote her—“I don’t believe in injecting a baby with another baby”. She has 666,000 followers and her videos are all still up on your platform.
Theo Bertram: I would be happy to refer that to our teams so that they can review. That certainly sounds like it is violating our rules. It is very—
Q16 John Nicolson: There are a lot of them. It just took me minutes to find lots and lots of people saying similar things, with tens and hundreds of thousands of followers, up on your platform.
Theo Bertram: Our policy is that we will remove any falsehood—any false vaccine information—from the platform. It is very simple to flag if you hold your thumb on the screen. If you have that account, you can just hold it down and you will see that a report comes up and vaccine flagging is the first thing that you can see, so you can flag that and we will look at it.
Q17 John Nicolson: The point I am making is that if you cannot sort out somebody with 606,000 followers, what chances are there that you are going to get rid of the smaller fry? This woman is just screaming lies as publicly as she possibly can in very professionally produced videos.
Theo Bertram: I can’t comment on these specific videos. I said that I have not seen them, so I am not going to make a judgment about it, but we have a team that is expert and will look at it. If that is flagged to them, they will remove it. I can’t make a judgment for you on that specific video, but we will review it and I am happy to write to you about it. What I can say now, as I have said to the Committee, is that we have a clear policy against vaccine misinformation and we have had it in place since March.
John Nicolson: It is not working.
Theo Bertram: We have a policy in place. We remove thousands of videos and we are banning accounts where they breach our rules.
Q18 John Nicolson: I found scores of them last night just looking at TikTok. It did not take me very long. I have just picked up my iPhone to double-check what you said about authoritative sources and warnings. I looked at Olivia Madison again today to see if there was an authoritative sources redirection—absolutely nothing.
Theo Bertram: We are launching the redirection on Monday, to be clear.
Q19 John Nicolson: All right, so it is not up and running yet. How many people do you employ? If I can find this stuff so quickly, and such egregious breaches of the code, why aren’t your staff finding it? How many people do you employ to look for disinformation videos?
Theo Bertram: We have more than 10,000 content moderators globally, and they all will flag to a team of specialists who are working on Covid misinformation, who work with fact-checkers like SciVerify and others. We are constantly learning what we can from SciVerify about new trends and how we can tackle Covid and vaccine misinformation. We have more than 10,000 moderators currently working on it.
Q20 John Nicolson: I am a journalist by profession, so I research. That is what my profession does when we write stories. If I was writing a story on this, the headline would be, “TikTok’s system not working: anti-vax fanatics everywhere”. This person, Olivia Madison, even highlights her anti-vaccination credentials, because she puts in #antivax, #antivaccine. It is the easiest thing in the world for your team, if they are serious about doing this, to look at people who are putting anti-vaccination hashtags, because they are shouting at you; they are doing their very best to draw your attention to them, and you are not finding them. Your system is not working.
Theo Bertram: I think it is worth saying that it is a bit more complicated than simply removing every video that says “anti-vax” or “anti-vaccine”.
John Nicolson: It would be a starting point.
Theo Bertram: The challenge that we have as a country is that we need to reach, I think, 70% of us having taken the vaccine. The real challenge is those people in the middle, who have real concerns. Probably about 50% of us, like me, would take the vaccine straight away, as soon as it is offered. There is probably a smaller group—about 9% to 15%—who do not want to take the vaccine whatever. Then there is that middle group. That middle group have real concerns and questions, and sometimes they post videos that have #antivaccine or something like that, and they do have reasonable concerns. If we were to simply delete them all from the platform—
Q21 John Nicolson: That is not what I suggested. I just suggested that it is a good route to go down, if you are trying to find folk who are spreading disinformation.
Let me move on to Ms Stimson from Facebook. After the measles outbreak in 2019, most of the main social media companies acknowledged their role in spreading anti-vaccination propaganda, but anti-vaccination propaganda has been increasing. Isn’t one of the reasons for that the fact that anti-vaxxers pay you?
Rebecca Stimson: I am not sure exactly what you are referring to, in terms of paying us.
Q22 John Nicolson: Facebook accepts paid-for anti-vax adverts.
Rebecca Stimson: It has always been in our advertising quality policy that we do not allow fake claims around vaccines and treatments, so we would not. As I said at the beginning, we recently refined our advertising quality to now also exclude any advertising that would cause caution or any concerns about taking the vaccine.
Q23 John Nicolson: So you don’t accept any advertisements of any kind from anti-vax campaigners?
Rebecca Stimson: Our policy is to not allow any adverts that make fake claims about vaccines or might discourage people from getting the vaccine.
Q24 John Nicolson: One of the problems is that social media companies like yours, it seems to me, are anxious not to alienate the anti-vax user base. Estimates, particularly from the Centre for Countering Digital Hate, show that they are worth a lot of money. I think that is what some people do not realise. The estimate is that these anti-vaxxers, especially those who are operating a business, are worth up to £1 billion a year. That is powerful motivation for companies like yours, isn’t it, not to upset them too much?
Rebecca Stimson: There are two things. We work with the Centre for Countering Digital Hate. I think in that report, their sample size was pretty small and did not, in our view, reflect quite accurately what is going on. One thing I would add is that there is no business incentive for us to carry this kind of content. One of the policies we do have is if we find a group or a page that continually violates our policies for this type of misinformation, we downrank them very heavily, so that very few people are able to find that content, and that can reduce engagement by 80%.
Q25 John Nicolson: Why don’t you take them off?
Rebecca Stimson: I will just finish my point. If it keeps happening, we also remove their access to any form of ability to raise money or fundraise through the platform. The reason—
John Nicolson: As the publisher, why don’t you take them down?
Rebecca Stimson: The reason we do not take them off our platform is very much to Theo’s point, which is the information we have had from projects like the Vaccine Competence Project, run by Professor Heidi Larson and others, is that you might compound these conspiracy theories if you simply remove them entirely, because obviously they exist outside of our platform as well as on other parts of the internet. What you want to try to do is reduce engagement very heavily with that kind of content, but where someone is looking for it and finding it, you then use that to redirect them to the right sources of information, the authoritative sources of information, so you hopefully begin trying that process of education. As we all sit around this Christmas talking about what is happening, it is much better if people are armed with that information, rather than just knowing that we had it all deleted, which may compound their belief that it is true.
Q26 John Nicolson: But again, the problem is that it does not seem to be working, because anti-vaxxers keep increasing in number. Estimates are of 8 million additional followers since 2019, and the estimate continues that anti-vaxxers now enjoy a following of some 58 million, with particular hotbeds in the United Kingdom and the United States. Whatever it is that you are doing is not working. They are increasing in number, and they are threatening people’s lives. It is not a game anymore. People will die if they do not take this vaccination.
Rebecca Stimson: We absolutely agree. I agree with you on a personal level as well. I am sure we are all keen that this vaccine rollout is as successful as it needs to be, for all of us to get back to the way life was before. I am not familiar with the figures you are quoting, and I would be very happy to look at them. I don’t know if they relate specifically to Facebook’s platform. We are taking this incredibly seriously. We are investing very heavily in tackling it. We have long-standing policies. This is an absolute priority for the company through this year. However, I think you are right. No one can claim this is working perfectly. This content is still there. Our thinking about it—
John Nicolson: It is not working at all, let alone perfectly.
Rebecca Stimson: Wherever there is vaccine misinformation that presents some imminent, real-world harm, we do find and remove it. But where there are other types of misinformation, including some conspiracy theories where there is not an imminent risk, as I have said, our aim is not to find and remove it all. Our aim is to use it as a moment to try to educate people.
Q27 Damian Hinds: This is not a level playing field, is it, between truth and untruth? Just presenting a debunk does not achieve what you need to achieve. It is a very well-documented psychological point that if we present people with something that scares them and something that reassures them, the scary and the fearful has more resonance and lasts longer than the reassurance. Unlike the earlier phases of Covid, we are dealing with people having to make an active decision. They do not need to believe something they see online; they just need to have the seeds of doubt sown for vaccination to be undermined.
I get the point that a couple of you have made, that you need to get that balance between reassuring people but not fanning the flames of the conspiracy theory. You have a lot of tools now at your disposal: labelling, takedowns, promotion of trusted sources, warning accounts, taking accounts down. My question—and I will put it to Rebecca—is a simple one, although the answer may be a bit complex. The question is simply this: what works?
Rebecca Stimson: Two things have struck us in the last few months. One is the importance of redirecting to information. I think it is a very complicated and confusing picture—
Q28 Damian Hinds: I am sorry to talk across you when you have only just started. Do you know for a fact that that has efficacy in reassuring people and changing their real-world attitudes?
Rebecca Stimson: It is hard for me. We do not have empirical evidence of that. Obviously this is an ongoing situation. I have pointed out the 3.5 million people we have driven to the NHS to read information there. It is very difficult for us to draw from that whether there has been a material behaviour change. What we can draw on a bit, and it partly speaks to your point, is the statistic I already mentioned, which is that where we are labelling things that are subsequently found, for example, by our fact-checking programme to be untrue, 95% of people do not then click through. Has that changed the way they think about that issue? It is very difficult to say, and this is a hugely interesting new area of behavioural science to think about. Personally, I think that is quite a good sign that that is quite an effective intervention, that people have not read that piece of content in the first place.
Q29 Damian Hinds: I would say, Rebecca, that it goes beyond being hugely interesting. In this particular case, it is also a matter of life and death in some cases. Who is doing that research? At the end of the day, it is great to know that somebody had read some other post, but if it has not removed from their minds the thing that is going to stop them doing what is needed to protect them and protect others around them, they can read all the posts in the world and it is worthless to us. What we need to know is how attitudes change.
Rebecca Stimson: One of the things that we are doing to use our platform to help in that regard is working with the NHS on the flu vaccine—the normal annual flu vaccine programme. It is obviously up and running at the moment. We have given the NHS $350,000, I believe it is, of free credit—free advertising—on our platform. We are also, besides that, going to do an insights report. Through the flu vaccine programme, we can test some of these behaviours, what messages work, what do people respond to or not respond to. It is obviously not a perfect proxy for Covid, but hopefully this is an important partnership for us, with the NHS, where we will be able to draw the kinds of lessons you are talking about to show what changes behaviour in the real world.
Q30 Damian Hinds: Can you, or either of the other two panellists, tell us anything about relative reach, read rates and dwell times of Correct The Record postings, relative to the original postings, particularly on reach?
Rebecca Stimson: I think we may be the only platform that has Correct The Record. Can you just explain to me? Do you mean how many—sorry, could you clarify your question?
Q31 Damian Hinds: If we go back to the pre-vaccine phase, we know there were some postings that—I hesitate to use the phrase “go viral” in this circumstance—had extremely wide reach across social media, and many, many people saw them. Friends of ours, and family members of everybody on this call, saw some of these messages; and some of those people, who are intelligent and well-meaning, were taken in by them and passed them on to others. How many of them then saw and read some debunking of that claim, and how effective was the debunking compared to the original message?
Rebecca Stimson: Correct The Record is one of the things that has evolved through this year. Previously, it was only if someone shared a piece of content that we would go back and give them, at the time, a relatively limited amount of information, but that would have been everybody who shared it. Now, in relation to Covid, we have expanded that tool so that it includes people who may like it, click on it, comment on it or leave an emoji. A much larger group of people who have interacted with that piece of content will now get a much more in-depth notification when they open Facebook that tells them, for example, “Two months ago you liked this thing and now it has been debunked. Here is the thing that you liked. Here is the source of information about why it has been debunked”. That goes to everyone.
Q32 Damian Hinds: What do you know about dwell-time engagement on those debunks, relative to the original postings?
Rebecca Stimson: I would need to come back to you on that. I don’t have numbers. I am not sure if we have those numbers, but I am very happy to look into—
Q33 Damian Hinds: Please do. I am sure they can be derived. Tell us about WhatsApp.
Rebecca Stimson: WhatsApp obviously is a private encrypted messaging service between people who have each other’s phone numbers. We have a number of tools that run on the unencrypted parts of WhatsApp—profile photos, status and so on. One of the ways that we have tried to address the spread of harmful misinformation in WhatsApp is around reducing forwarding limits and labelling. We previously had policies where if you forward something several times—and I am sure some of you will have seen this—it says “forwarded” on it. If it happens multiple times, it says “frequently forwarded”. What we have also done to restrict, so you are more aware—
Q34 Damian Hinds: Do people take “frequently forwarded” as being a signal of poor quality or high quality?
Rebecca Stimson: The signal it sends to us in an encrypted world where we cannot see the content is that kind of mass bulk sharing tends to be a sign, among many other signals, that we may want to prevent that from happening—that that may not be the most helpful thing. What we have done is reduce it. Now you can only forward something to five groups at a time, to greatly slow it. On “frequently forwarded”, if it looks to us like something is going viral, to use that word, you can only share it to one group at a time. What we have seen is that that has led to a 70% reduction in forwarding. Like you say, without the access to content, obviously we are having to rely heavily on signals to work out what is happening, but those tools have been effective. We also ban about 2 million fake accounts a month from WhatsApp.
Q35 Damian Hinds: To Rebecca again, does it remain true that if one takes some of the highest volume fake postings from earlier in the pandemic, we don’t know how far they travelled on WhatsApp?
Rebecca Stimson: It is not possible for us to say how many people may have shared those in WhatsApp, but as I said, we have seen a 70% reduction in the rate of forwarding because of the measures we have instigated in the last year.
Q36 Damian Hinds: Can I ask what we have learned from the early Covid phases, particularly on the motivations of originators of false messages? I am not talking now specifically about anti-vaccination, but in general. If you take something like the “classmate’s uncle and nephew” message, which spread on Facebook, WhatsApp and plenty of other platforms, we know lots of reasons why people would forward that. What do we know about who originated it, and why and what have we learned from that for this anti-vaccination space?
Rebecca Stimson: Is that directed to me as well?
Damian Hinds: Yes, please.
Rebecca Stimson: Intent is always a very interesting one. As Theo pointed out—we have seen the same research—there are hardcore anti-vaxxers, and people who don’t believe Covid is real and are spreading very dangerous conspiracy theories about Covid and the vaccine. That is the kind of content we try to remove as quickly as we can. Then there are people who are completely fine with it; and then there is this big cohort in the middle, where there is great variety in what they are talking about. Lots of the time—John referenced this in his questions—they are innocently sharing something that they might think has some merit that they do not understand, right the way the way through to—
Q37 Damian Hinds: Sorry, Rebecca; forgive me, but I specifically am not talking about sharing. I am talking about originators. We know hostile foreign states will sometimes take advantage of all manner of situations, including the aftermath of terrorist outrages, to create doubt and confusion and undermine faith in institutions. Is there any reason to believe that foreign states were involved in some of the early-phase Covid misinformation and disinformation?
Rebecca Stimson: No. We have long-running teams within Facebook who look for what we call co-ordinated inauthentic behaviour, and we have been very, very alive to that risk. We have not seen that kind of activity that you have just described in relation to Covid during 2020.
Q38 Damian Hinds: Just be very, very clear, you have not found any reason to believe there may have been seeding of multiple false postings, some of which then went viral?
Rebecca Stimson: That is correct. We have not seen evidence of that.
Q39 Damian Hinds: Thank you. I have just a couple more questions. Can I draw this out to the rest of the panel? The 2020 US presidential election seemed to be, relatively speaking, a success for social media companies compared with the 2016 election, partly because you had unprecedented levels of resource working on some of these aspects. How does the level of resource that you now have looking at anti-vax misinformation and disinformation compare with what you had running at the height of the US presidential election? Can I go to Iain Bundred?
Iain Bundred: We are fully resourced and ready for this challenge at a global level. It is a global pandemic, and it needs absolutely the full resource of the company towards it. We are spending about $1 billion on content moderation globally, and that will continue.
Q40 Damian Hinds: Is that higher or lower than it was a few months ago—the numbers in the total workforce working on misinformation and disinformation?
Iain Bundred: I would say that it is unchanged. There were shift patterns around the days of the election itself, where we saw particular things. We had a big change in terms of our work from home policy at that start of the pandemic, trying to find ways to work through that, bearing in mind that particularly people were watching videos with some very sensitive materials and that can be difficult, but we have addressed all that and we have put in the full resources of the company towards this.
Q41 Damian Hinds: I realise I am very short on time, so can I get just a very quick answer from Theo and Rebecca? Is your level of resourcing at least as high as it was a few months ago at the peak of the US presidential election, looking at misinformation and disinformation globally?
Theo Bertram: For us, yes, because we are growing so fast. Obviously every new month means more resource and more people.
Damian Hinds: Thank you. Rebecca? Or you can write to us about the level of resourcing over time. My very final question to you. Clearly there are going to be personal tragedies for people shortly after taking the vaccine. There will be old people who, very sadly, will die shortly after taking the vaccine. It will not be as a result of the vaccine, but from some other cause. Clearly people will seize upon that and use your platforms to spread falsehoods. What are your contingency plans for dealing with those situations? Rebecca.
Rebecca Stimson: As you know, we are part of the policy forum set up by the Government, with a number of the other companies. That is a very important mechanism for us to keep absolutely in lockstep with all the latest developing information. We have recently seen, obviously this week, the variant of Covid found in the UK. That is a great example of how quickly we work with the Government, and with all the health experts, to understand at least the facts as we have them at that point, so that we can start to figure out whether our policies are in the right place or not. That develops.
We have already had a number of discussions with the Department of Health and the DCMS about how complex—the exact example you have given—the rollout of the vaccine is, and exactly how it is going to work in different groups. All I can tell you is that what we have done is build very open and very rapid channels of communication with the right health experts, so that we can move as quickly as possible to know what is true and what is not true and can start to react to it.
Q42 Mrs Heather Wheeler: Your last answer, Rebecca, was very helpful because what I want to pose questions about is how the three of you work together. Are you allowed to talk outside of that government body? Could you give us concrete examples of working with each other on the panel—of how you are getting more effective at getting rid of the anti-vaccine propaganda, please?
Iain Bundred: It is worth saying that we all have to work together. It is not just the platforms. We are working together. There is the Government counter-disinformation task force. We are meeting regularly. There is also work that we do through Full Fact—Twitter, Facebook and ourselves—and we are working closely, sharing learnings and that kind of stuff. It is very important as well that we learn from academia and from NGOs and campaigners on this issue who have a different reflection on it. They can tell us information about, for example, the particular messages that are resonating best. We are working with the NHS as well, of course, and raising authoritative content.
Going forwards, as we try to use our platform as a way of driving confidence in the vaccines, we are working with bodies like the Royal Society for Public Health and doctor creators on the platform who can give reassurance messages. It is a massive group effort, because we all need to put our shoulders to the wheel on that.
Theo Bertram: I agree with what Iain said. Obviously this is not just a UK thing. We are all signed up to the European Code of Practice on Disinformation, and there is an important part of that now that is specifically around Covid, so we are working together a lot.
I would also agree with what Iain says. I think there is a crucial role for the Government to pull this all together; it is not just the companies here. If you think about great campaigns in the past, when we had HIV as the big issue that the country wanted to tackle and to change public opinion on, it was not just that you put the Minister on the “Today” programme. You also made sure that there was a plot on “Coronation Street”. It is that kind of collective pull that we need to do together. I know the Government are working on that.
Rebecca Stimson: I would just echo those comments. There are a number of different forums and ways that we are all working together. We have different academic partnerships to make sure that our data can be used, anonymised and aggregated to help build things like disease prevention maps, to help the Government understand where people are moving around and things like that.
We have other tools, like a dashboard that helps people track the big themes of conversations that are happening online, to help people track something that is emerging. We saw that was very helpful back in the spring, when the 5G conspiracy theory emerged. It was picked up that something was happening; there was a conversation beginning out there on the platform. Yes, we all work very closely together. It is extremely serious, and something we are all very committed to.
Q43 Mrs Heather Wheeler: A last couple of questions from me on this. Is there a particular barrier at all—I am not sure that it sounds like there is—about you sharing information on this, trying to deal with the anti-vaxxers?
Theo Bertram: I do not think so. We all have slightly different platforms in the way they work. We have different profiles of users as well. The challenge as it hits us will be slightly different, but I do not think there is a barrier to us working together. It is something we do across all areas.
Mrs Heather Wheeler: I suppose I am pleased to hear that. I am trying to work out whether my takeaway from your evidence is that I need to phone up my mates that do “The Archers” and get them to have this as a line coming up. That should be interesting—to see next Sunday if they have fitted it in or not. Thank you very much, all.
Chair: You are muted, Clive.
Q44 Clive Efford: Apologies, I clicked to another screen rather than clicking off my microphone. Rebecca, can I just ask you about something you answered to John Nicolson earlier on? I get that you cannot take down all content, and there is a fine line to be trod, but you said that you would rather use the content that is left up there as an opportunity to educate. Could you give us an example of what you do in those circumstances? What is your source of the information with which you try to educate? How do you measure whether people are engaging with that positive information that you are putting out?
Rebecca Stimson: Thank you for the question. We do this in a number of different ways. We have the largest fact-checking programme of any of the platforms, so that can be a way that pieces of content end up being labelled, and that we redirect anyone that comes across it to the accurate sources of information. As I said, that involves also going back and correcting the record when people have interacted with pieces of content previously.
We also have links on our site to all kinds of authoritative sources of health information, and we put warnings on different things, so there is a big toolkit. The reason we do this is based on the information and advice we have been given from health experts. This is what they recommend is the best course of action when you are dealing with, just to be very clear, misinformation that does not present any imminent real-world harm, but other forms of misinformation. For example, we have seen a few things around the distribution of the vaccine—that certain parts of the country will not get it, or that it is running out—where although there is not an imminent, real-world risk of someone doing something harmful to themselves, it is clearly something about which we would want to give people accurate information, so we link to third parties where that kind of content is found.
Q45 Clive Efford: Is there any way of monitoring how successful you are and how much impact the positive messages you try to put out have?
Rebecca Stimson: Yes. Like I said in previous answers, we have some empirical evidence of that around, for example, people not clicking through where we screen a piece of content and say, “This has been fact-checked and debunked”, which is a good sign that people find that helpful and are then deterred from reading that sort of content. Like I mentioned, we are doing this interesting work with the NHS on the flu vaccine now to test out what kinds of messages and ways of working influence the way people think about that, so that we can take forward that learning into future things we may do in relation to Covid in particular.
Q46 Clive Efford: Is this more effective than just taking stuff down? Are those people influenced by the good as much as the bad, so if you use that as an opportunity to put out positive messaging, it has more impact than just taking stuff down? Is that what you are saying?
Rebecca Stimson: That is the policy that we have, based on the approach that health experts have told us is most effective.
Q47 Damian Green: I want to move on to what you can do. Helpfully, this is picking up on something that Clive just said. One of the problems is clearly a distrust of traditional authority figures, which is likely to be very strong among people who are tempted by anti-vax propaganda. Is there a problem? Are we short of trusted voices in this area?
Iain Bundred: Certainly on YouTube we have some great creators who are medical experts. We have, for example, a doctor called John Campbell, who has an older audience. He is a retired former nurse and has a nice following, and that is someone you can look to promote. We have a bunch of other doctor creators who also speak to some of the younger audiences. The NHS tells us that right now, the number one public health message is just to reassure all users that the vaccine is safe, rather than specific targeting, which I think will come in the New Year, around audiences who need to take it up.
We are also talking to various partners about how we can drive segmented messages, and how we can seek to normalise things. We saw that research yesterday about BAME concerns. We want to see, for example, nurses and doctors who are eligible for it from the BAME community telling their stories. For example, the Mayor of London’s office has been talking to us about doing something in this area. There is a massive opportunity. The magic of YouTube is very much our creators and these engaged audiences. If we can help them understand the situation and help them normalise some of the vaccine points, but also of course help them spread a very important public health message, then we can hopefully play our part and support this the best we can.
Theo Bertram: I would say something very similar to Iain, but first of all I would give a perspective: the UK is not in a bad position globally in terms of the level of trust. If you look at the disincentives, looking at some of the survey data and the research in this area, the challenge in the UK is less about mistrust of authority and more that there are perhaps reasonable concerns that people are expressing around the speed of the introduction of the vaccine and whether there are long-term effects. We have great answers to both those points. It is about who we get to carry that message to the particular audiences.
All the Members of the House here have all won elections. You all know how important it is to not just have the right message, but to have the right way of carrying that to particular audiences. I think that is probably what we need to do more over the New Year. We need to work out how we get this message to this audience that has this concern. On our platform, we might reach a different part of the electorate than “The Archers”, but we can play a very important role in that, and we would be happy to do that.
Q48 Damian Green: That is the question I am asking. To reduce it to specifics, if you had to pick three people on TikTok who would influence most of your people to get vaccinated if they were doubtful, who would they be?
Theo Bertram: There are two I would pick off the top of my head. One would be Dr Anna Blakney, who is great and is working on the virus, but is able to communicate in a very simple way in 30 seconds, which is an art form in itself. Another is—I am trying to remember his name—Dr Raj. I have just been watching his videos this morning. He is great. He would never do a video the way that you might expect a Minister to. I do not know; maybe I have the wrong idea about Ministers. He videos himself lying on his bed, telling you about how he is responding to Covid questions. There is a relaxedness to it that resonates with our audience, and he has hundreds of thousands of views. It is about identifying those key voices, but I think celebrities will also have a very important role to play as well.
Q49 Damian Green: I was going to ask about that. You mentioned doctors and Iain mentioned medical experts as well. Who is most effective: celebrities or experts?
Theo Bertram: If I said to you, “Who is the most effective person at convincing someone to vote for your party?” you would not say, “It is these three people, and that is it”. You would say, “It depends what the circumstance of that person is, and what the message is we are carrying,” but you would have teams of people, I am sure, working on that, and working out what is the best way to run that campaign. That is where we will move to in the New Year. We have to see this as a massive public information campaign, in which we are all on the same side.
Q50 Damian Green: Sure. Are any of your companies doing anything active to say, “Okay, we are part of the battle against anti-vax propaganda, so we are spending money and effort and time in doing something”? If so, what?
Theo Bertram: Yes. There is getting stuff off the platform, and we are all doing that. We have talked about that. It is about things like Project Halo, which is the programme that we are running with the Vaccine Confidence Project. How do we get those voices on the platform? How do we find creators on our platform and link them together with those trusted sources? At the start of the pandemic, it was not just about making sure we had Matt Hancock on the platform to give a message, but asking, “How do we also help connect our creators to the NHS, so that the people that are already being followed by hundreds of thousands of people on our platform give the same message?”. I think there are things that we can all do, and I am sure we would be happy to help.
Iain Bundred: From a YouTube perspective, if you think about raising authoritative content, we are informing news outlets all the time about the searches that are happening on our platform, getting people to fill the content void on some of these issues and some of these concerns. We are also backing campaigns over Christmas and in the new year, and trying to find the right partners and creators, including services—the ones that academics, the Government and others are telling us will have the most credible impact with the target audiences that are important at each individual stage. That is a massive job for us all to do in the New Year, I am sure, and over Christmas as well.
Q51 Damian Green: A related question. You talked about other organisations you could work with. One of them is the Government’s own counter-disinformation unit. Have you found them useful in any way?
Iain Bundred: You are listening to representatives of DCMS later. You also have James Sorene, who has a new role in the Department of Health, who has been great for us, helping to connect us to the latest information. I think that is a very important part. As we saw the vaccine development coming closer, our policies got updated. We have driven 400 billion impressions to information panels globally from local health authorities, from the equivalents of the NHS around the world. That is only possible if we have the NHS information to link to. It is critical that we work together to see what the information is that is being asked, the questions that are being searched and the public health messages that drive forward.
I have been very pleasantly surprised—“surprised” is the wrong word. I have enjoyed working with the Government on this. I think they are doing their best. In the vaccine rollout in particular, they are driving it forward to try to focus this as best they can. Of course there is more to do—there is always more to do—but I am happy with the process so far.
Q52 Damian Green: If there is one thing they are not doing that you would like them to do, what is it?
Iain Bundred: More videos—just putting out more video content. We can drive people towards more videos. We have heard already that different audiences respond to different messages and different message carriers. JVT seems to be doing well—more of him as well. I think lots of people will agree with that.
Damian Green: We need a JVT channel. Thank you. Back to you, Chair.
Q53 Chair: Just to follow up on that, Iain, if I decided to post a video on YouTube with anti-vax theories, would I still be paid for it? Would I still earn money from it?
Iain Bundred: No. That video would be removed. It would be against our Covid misinformation policy.
Q54 Chair: Some of the conversation we have had so far is that there are gradients of disinformation. If I were putting a video out there that argued coherently—which would be a rarity for myself, obviously—about anti-vax disinformation, would I receive money for that?
Iain Bundred: Our policy is now pretty simple and straightforward. It is a flexible Covid medical misinformation policy that removes any content that contradicts the local health authority—in our case, NHS or WHO guidelines. If what you were saying was all factual and did not contradict NHS guidelines, for example, then the video would stay online. We may treat it as borderline, perhaps, if there were certain things in there; and that would mean that you could be demonetised and so on. I cannot win against the hypothetical, but I am pretty sure it would be removed, given the sort of thing you are saying.
Q55 Chair: It is just interesting, because there still seems to be room there for some blurring. I know that you talk about wanting to have engagement with people in order basically to reach that 10% or 15%. I think Theo was talking about that, in terms of that middle ground. I can understand that, but it seems to me that there is still the monetary aspect of this, in which you could effectively be in quite a lot of grey area, but at the same time there is a freedom of speech issue as well.
Iain, you mention contradicting NHS guidelines; in what respect? Do you mean, for example, me saying that you should be able to do this or do that, or do you mean basically any anti-vax disinformation—any anti-vax going off-piste, if you like—could lead to you being taken off?
Iain Bundred: To give a concrete example, I mentioned James earlier from the Department of Health and Social Care. He came to us a few weeks ago to tell us that they were worried about false claims that the Pfizer vaccine involved animal products. By giving us that information and giving us clear guidance from the NHS, we could make the view that any claims such as that are prohibited from our platform.
Chair: You are muted, Kevin.
Q56 Kevin Brennan: I thought I had unmuted, but there we are. Theo, there has been a lot of talk about the disinformation unit in the Government. Is TikTok included in that?
Theo Bertram: We have been working with the disinformation unit at the start of the year. There is also a Government ministerial roundtable, which we are not currently part of. That is not to say that we do not work closely with the Government. We work closely with DCMS, DH and NHS. The message we got from the folks running the roundtable is that the priority is on another set of companies, which are the ones—
Q57 Kevin Brennan: As I recall, yesterday, or whenever it was when the response to the Government White Paper was issued, you were in one of those category 1 companies of the main social media companies that the Government were concerned about. Does that mean they should be involving you in this roundtable or not?
Theo Bertram: As I said, we work closely with the Government. I think the idea of the roundtables is to focus on what their priority is at the time. I think we are a little further down. If you think of the tiers of priority of people who are getting the vaccine, our age group is probably a little lower than some of the other platforms. The suggestion from the DCMS is that we might be more of their priority in the New Year, rather than just at the moment.
Q58 Kevin Brennan: Can I ask you one question about something John Nicolson asked earlier on about that account that had 606,000 followers? You said if it was reviewed and the video was found to be as John described, it would be removed. Would you under those circumstances remove the whole account, or just that one item?
Theo Bertram: I cannot pronounce on a video or an account that I have never seen, as I have said, so—
Q59 Kevin Brennan: I accept that, but if it was as he described, a video that bragged about its anti-vax credentials—I am not asking you to rule on something you have not seen, just on principle—would the decision be taken to take down just the one item of information or one video, or would it be to remove the account?
Theo Bertram: One item we would not necessarily remove an account for. There are lots of people who perhaps unwittingly post misinformation. They have doubts, or they are repeating concerns that they have had. Often when you remove that video and they understand what it has been removed for, that is enough of a corrective in itself. If someone was repeatedly breaching our rules—and we would look at it in a holistic way—then we would remove it.
Q60 Kevin Brennan: Would it be something like three strikes and you are out, or would it be a second offence and you would be gone? How would it work?
Theo Bertram: It is not as simple as that. It could be one strike and you are out, depending on what the content is. We would review the account as well as each video and then take a decision, but it could well be, if there are repeated violations, that we remove the account
Q61 Kevin Brennan: Can I turn to Rebecca and Facebook? Earlier this year you promised to remove harmful misinformation about coronavirus, yet I have also looked some of the stuff from the Centre for Countering Digital Hate, which was mentioned earlier on. It did a study with a youth group called Restless Development, which found that Facebook removed just one in every 10 such posts that had been reported to it by users, and these are posts that included false claims that zinc, vitamin C and vitamin D cure Covid, that viruses cannot be transmitted by air, that the only way to contract Covid is for it to be injected into one’s body. In that sense, you have not lived up to those promises, so why should we, as a Committee, trust what you now say you will do on misinformation about the Covid vaccine? What has changed that would mean that we could have more trust, given those facts?
Rebecca Stimson: That particular report was a very small sample size, and we have taken action on some of the content that that report flagged up to us, but not all—
Q62 Kevin Brennan: What happened was that there was a follow-up because it was not removed, and that resulted in 17% of the posts being acted on. Did you not accept that the rest of the posts deserved acting upon?
Rebecca Stimson: As I have said, our misinformation policies are based, particularly in this particular subject area, on what the medical experts tell us and the health experts tell us is the right approach to these subjects. Part of—
Q63 Kevin Brennan: Let me stop you there for a moment. I am sorry to interrupt, but let me give an example, so that the Committee has this information and it is on the record. One example of a post that you did not remove following user reports said, “Vitamin C and D are important, and zinc. Covid doesn’t survive if these are in your system. Drink hot drinks also regularly through the day”. Another one said, “Viruses cannot be transmitted through the air. They are not contagious. A virus has never been proven to kill anyone. A flu is a body’s immune response to flushing the toxins out” and so on. “Viruses are good. We have been lied to about everything.” Another one said, “Covid-19 is just a false flag to trumpet in the vaccine mandates” and so on and so on. These are ones you chose not to remove, so how do you explain that?
Rebecca Stimson: It is hard for me to comment without seeing the actual pieces of content. As others have said, it is difficult to speak about these things in the abstract. Probably the answer—
Q64 Kevin Brennan: With all due respect, you were aware of the report. You were aware of the fact that these had been reported. You were aware of the fact that one in 10 had been removed. You were aware of the fact that 17% were eventually removed. You said that you knew that it was a small study, so you should be completely familiar with these postings, as the company reviewed them all when they were reported and decided not to reject them. I do not understand how you can say it is not possible to comment.
Rebecca Stimson: To answer your question, my guess would be the reason those posts were still allowed was that they would fall short of causing imminent, real-world physical harm to anybody. For example, if someone believed that vitamin C might prevent it, that would fall into the category where, based on the experts who advise us on our policies, we would label that, and try to bring better information to that person, rather than someone who said, for example, “Drink bleach,” where clearly there is an imminent, real-world problem.
Q65 Kevin Brennan: What about the statement, “Viruses cannot be transmitted through the air”? That is the exact transmission mechanism that is most likely to be the way that this particular virus is transmitted—via aerosols, principally. Why wasn’t that removed?
Rebecca Stimson: As I said, I would need to look in detail at that exact post to give you a specific answer. My suspicion is that probably it falls short of imminent, real-world physical harm if somebody believes that, but clearly that is misinformation. It would count as our misinformation policy. Any post like that, if not removed, would be very significantly downranked, so it would have 80% less engagement with it, and it would very likely end up being labelled in some way as well, so that it directs people to the right source of information, should they still find it.
Q66 Kevin Brennan: There is a headline today in The New York Times. I know that executives from Facebook do not always read The New York Times, because the last time one appeared before the Committee and I quoted a story about Facebook from The New York Times, they claimed not to have read it. Today’s headline says, “Facebook reviews post-election algorithm changes that boosted news from authoritative sources”. In other words, the company has taken an active decision to de-boost, if you like—to change its algorithms, to relate this to a question that the Chair asked about the virus—so that authoritative sources are not boosted in the way they have been boosted during the election count period in the United States.
First of all, why would Facebook want to reduce the boosting of authoritative news sources to its users? Isn’t there a danger that the company is going to do exactly the same thing with regard to viruses and public health when attention is taken away from this subject?
Rebecca Stimson: As you can imagine, during an election period, it is extremely important to make sure people have access to the right kind of information. There were some particular challenges with the last US election, which took place in the context of a pandemic. Our priority was to ensure that people had access to the best, most accurate sources of information during that time period, including information around voting and voter suppression. Obviously it was a complicated period around the election and the process of that election. Our priority is to make sure people see the right information. However, we are not able to—
Q67 Kevin Brennan: My point is: why isn’t that always your priority? The decision was during that period to amplify news ecosystem quality. We are doing an inquiry at the moment into public service broadcasting in this country. It seems to me that any responsible player here would want to always amplify news ecosystem quality, whether it is accurate information about our democracy, or about a dangerous virus. Why is it only a priority at that time to amplify news ecosystem quality? At other times, presumably you do not care about news ecosystem quality.
Rebecca Stimson: We agree with you that it is important that people see the best quality news journalism, and that is why we have rolled out a new product called Facebook News tab. It is coming to the UK next month, and it will collate together the sources of news information that are independently verified to be high quality—of the highest journalistic standards—from a whole spectrum of political views. We will put that in one place, so that users have access to all that quality news and journalism in one place and easily.
Q68 Kevin Brennan: Let me move on to Iain from YouTube, which of course is part of Google, which is effectively just a huge advertising company with a turnover of $161 billion annually, making staggering margins of profits out of its advertising. There is obviously a very strong need for the company to make sure it is being accurate in what it is placing its advertising next to.
In October you announced that you had removed videos containing false claims about the Covid vaccine, and yet there are still videos, one of which I saw just before this Committee, with over a million views, which claim Covid was planned as part of a conspiracy involving the Moderna vaccine, Bill Gates, George Soros and even Jeffrey Epstein. The video is published on a channel called Be Inspired, which regularly promotes false claims about vaccines, and yet it is still permitted by your company to post videos to their 7 million followers. Why is this channel still on your platform?
Iain Bundred: Sorry, that is on YouTube?
Kevin Brennan: Yes. Be Inspired on YouTube. I will send you the link after this.
Iain Bundred: Yes, please do. I have no idea. That content, just the headline alone, is violative, so it should not be on our platform.
Q69 Kevin Brennan: This is a channel that produces very professional videos featuring conspiracy theorists like David Icke, most of which are monetised and most of which feature adverts placed by Google. Is that acceptable?
Iain Bundred: For example, if it was featuring David Icke, it would probably be removed anyway because of the actions we have taken around that. Like I said, that video should not be on our platform. I am sorry that it is. Machine learning is improving all the time. We are working on this, but frankly that headline should be pretty easy to find.
Kevin Brennan: The Centre for Countering Digital Hate has found there are 5.3 million UK-based followers of anti-vax accounts across social media, so there is a lot of work still to be done.
Q70 Chair: Mr Bertram, have you been on your phone while we have been doing the evidence? We appear to have had a development. The account of Olivia Madison on your platform has mysteriously disappeared, including not just the video in question but the account itself. Did you let them know about that or have they been watching?
Theo Bertram: I assume that they have been watching. I have not been checking my phone.
Q71 Chair: No. It is a pity it takes a parliamentary Select Committee hearing though, isn’t it, in terms of getting rid of this stuff? 606,000 users. Frankly, we cannot do this every single time, can we?
Theo Bertram: No, but I assure you that we respond to all flags. It is very simple, and I encourage anyone who finds any content on our platform about which they are not sure to press their finger on the screen; in two clicks, you can report it as Covid misinformation, and our teams will review it. If it is violating, we will remove it.
Chair: Or alternatively, get elected to Parliament and get you dragged in here and then basically put the points to you.
Q72 Giles Watling: Immediately following on from that and from Kevin Brennan’s questioning, a previous incarnation of this Committee went to George Washington University. We saw there how bots from what they call “bad actors” can kick in. It is a piece of information that hits the social media sites, and then somebody somewhere, some algorithm, clues into it and says, “This is useful for our purposes. Let’s let this out”. We have seen bots from Russia and from Venezuela, and we saw real-time examples there of how immediately this goes from one little dot, two little dots, three little dots, then many thousands right across the country. If you tell a lie often enough, it becomes the truth. Is it not the case—and I think I put this to you, Rebecca—that you need to up your game? If one piece of information gets out that helps a bad actor, your platform is drawn into a conspiracy theory.
Rebecca Stimson: As I said, we have a co-ordinated inauthentic behaviour team that looks at the foreign state actor bot problem that you are describing. I have checked with them before the hearing today, and to date we have not seen anything related to Covid specifically of that nature. As you say rightly, however, we do recognise, of course, that there are individuals, including in some cases public figures, who are promoting things that are untrue and conspiracy theories. It is a big challenge. I have outlined the policies that we take, but there is definitely more to do as this process rolls forward to find and remove this kind of harmful content when it is there, and then also, as I have described, try to educate people away from it and provide them with an alternative viewpoint.
Q73 Giles Watling: Thank you, Rebecca. I understand that you have already outlined all this. What I want to know is what you are going to do to sharpen it up. If one piece of information gets out, if you are hitting one in 10, it clearly is not good enough.
Rebecca Stimson: We have found and removed many, many millions of pieces of content. I think I referenced earlier 167 million pieces of content that we have found. We are doing this at scale and have, I think, delivered a very impactful response to the Covid crisis globally, but there is absolutely more to do.
We have already demonstrated through this year that we continually refine and update our policies, often very quickly when different conspiracy theories—5G being a good example—suddenly emerge. We have ways of knowing that that is coming. Before the first vaccine was approved, we had worked very closely with health authorities to understand the kinds of conspiracies that may be around it, and the kinds of complexities—for example, how long that vaccine may be effective—which open the door to some of those conspiracy theories. We can be as ready as we can, but it is a live situation on the ground, so it will not always be perfect.
Q74 Giles Watling: What I take from that—and it is good news, I will give you that—is that you are actively working to expand your activity, to close these things down as quickly as possible. Is that true of you, Dr Bertram?
Theo Bertram: Yes, we are constantly looking for ways to improve, and we will learn from each other. We have iterated our processes during the year. We are working with more and more external researchers to learn what we can do. What we call “co-ordinated inauthentic behaviour”, which is where you are seeing state-sponsored or other bad actors trying to drive up conflict, is not something that we have seen on our platform in this area yet, but we remain vigilant for it.
Q75 Giles Watling: So you are expanding your operations; you are closing it down quicker and quicker every day, you think?
Theo Bertram: Yes. We continue to grow as a platform, so all of our trust and safety, as well as our area of security, are continually growing as well.
Q76 Giles Watling: Thank you. You are all saying the right things, all three of you. You are being positive about the posting of proper scientific research and so on. Of course many, many hundreds of thousands of people like to listen to uninformed conspiracy theorists: Kennedy, UFOs, the death of Princess Diana and so on. All this is great stuff and it is sexier than boring parliamentarians telling the truth. Therefore this is good, I would imagine, for your businesses. Do you not find that you do not want to kill the golden goose? In all three of your titles is the word “policy”—your policy. Are you not being pressured by senior management to keep these lines of conspiracy theories running? I know, Iain, on YouTube we get quite a lot of that, and a lot of it is quite harmless, but when we are talking about this pandemic, it can be quite harmful. Iain, are you getting pressure from senior management?
Iain Bundred: Absolutely not. From the very top of our business, we are 100% committed to both removing the Covid misinfo, but also playing our part in raising authoritative content, driving quality information and, as YouTube, we are working with our creators to 100% commit to help. This is the number one public health challenge of our time. That has never been a problem in my time.
Q77 Giles Watling: Even at the expense of some business to your company?
Iain Bundred: I do not believe there is an expense. The monetisation is tiny on these sorts of things anyway, and we are committed to making not a single dollar of profit or raising anything around that, so absolutely not. Not challenged.
Theo Bertram: TikTok does not function the way that YouTube and Facebook do. You are thinking of where, if you post a piece of content, there is an advert attached to it, and that is not how our platform works. If you are the kind of huckster that is looking to come and put anti-vax content and hope you are going to get advertising on it, it will not work on our platform. That is not the way our platform works, so I am absolutely under no pressure from anyone to do anything other than get rid of anti-vaccine content on the platform.
Q78 Giles Watling: Yes, and I should imagine it would be far harder on Twitter, but we do not have them before us today. Rebecca, famously, Mark Zuckerberg did not come and appear before us, and clearly he still has his hands very much on the reins. Are you getting pressure from senior management, would you say?
Rebecca Stimson: The only pressure is to play our role, as we have from the beginning, in addressing this pandemic, in supporting health authorities in Governments around the world to tackle it, and in ensuring that the best information is made available to people. We are working very hard. We have 35,000 people working, as I say, on safety and security. We invest heavily in these processes. We are part of all the forums that have been mentioned. The only pressure is to continually do better in what is a complicated situation, but one where we have delivered real results through this year.
Giles Watling: It might be rather nice if you could say to Mr Zuckerberg he ought to come along and have a cosy chat with us. Thank you.
Q79 Greg Clark: Thank you for allowing me to participate in today’s meeting. Can I just follow up on some of the questions that Heather Wheeler asked? All of you said in response to Heather that there are no barriers to stopping you working together. This is a big national moment. The vaccines are being rolled out and we want them rolled out in a short space of time. Your organisations are famously creative. You have illustrated some of the ways in which you are finding methods to take down content and to detect it. Why don’t you come together and form a group, a task force that meets perhaps on a daily basis, to make sure that we are absolutely on top of it, so that anything that one of you discovers is shared and spread across social media platforms? That is to Dr Bertram.
Theo Bertram: I think it is a good proposal, and I think we are always looking for ways that we can work more together. We are the people that do the Government and public policy work. I know that our trust and safety teams, who are the people that work behind the scenes removing that content, already work together a great deal, but I am sure we can do more to make sure that we work together, particularly to have that focus on the UK, given that the UK is the first country to put out the vaccine. Hopefully we will stay at the forefront of that rollout.
Q80 Greg Clark: Clearly, so I hope this is not something that is going to be considered in your report back after Christmas. Perhaps you might write to the Chair of this Committee before Christmas to say what you are doing about that. The implication of the answer to Heather was that there is no barrier in competition law to you talking to each other. The Chair, I am sure, would want to know if that were a problem. If you would do that, I think we would be grateful.
One final point, perhaps to Iain Bundred. Clearly your platforms are the principal way in which miscommunication is made—that false information from the anti-vax people is conveyed. That would be fair, would it not?
Iain Bundred: I am not sure I understand the question. That our platforms are the principal way of—
Greg Clark: The means of conveying this anti-vax information is principally through social media, is it not?
Iain Bundred: I think anti-vax concerns go back to the very first vaccination, and we have seen word of mouth growing over time. I do not see large amounts of it on the platform. If you think about those figures I was quoting to the Chair earlier—750,000 videos—given the global amount of videos on YouTube, I do not think we are the primary thing for that at all. The least you can expect is for us to remove those ones that are violative, and to make sure that, where we can, we are working with the creators, who are the lifeblood of our platform, to drive scientific reassurance messages.
Q81 Greg Clark: It is not in the public press, is it? Are you saying it is mostly word of mouth, rather than on online platforms?
Iain Bundred: I was saying that for YouTube, I do not see it as being—
Greg Clark: Generally, for the industry, as a collective.
Iain Bundred: I have seen some data on this. There was something from the Tony Blair Institute released this morning, I think, which is quite interesting, showing that heavy users of social media are less likely to share conspiracy theories. That is one piece of data, obviously. We are looking to learn from this all the time. We are absolutely recognising our responsibility, which is why you have heard from all three companies today how much we are trying to address and contribute to this massive issue.
Q82 Greg Clark: Thank you; that has been clear. Obviously some messages do get out, and we have heard that you take them down as soon as you can and suppress them, but public money has to be spent counteracting the messages on your various sites. At least for the period in which they are up, you do derive revenue, directly or indirectly, from that. Would it not be reasonable to refund to the public health authorities the benefits that you are making from these communications before they are taken down, to pay for the counteracting of them?
Iain Bundred: One thing I would say is that our machine learning tools catch most of this stuff before it gets to a large level. I think 96% of the videos were removed automatically. Of those, more than 98% were—
Q83 Greg Clark: For the ones that get through, it would be reasonable, would it not, for the money you make from that to be refunded to the authorities to counteract the effect of them?
Iain Bundred: My point is that we are talking a tiny amount of money here. The way that we support—
Greg Clark: Every little helps, to coin a phrase.
Iain Bundred: I think the millions of ad grants we have given the Government to drive public health messaging is probably more important, in terms of the way we are supporting the public health authorities.
Greg Clark: I cannot speak for the public health authorities, but I think they would be grateful for any contribution, from what you make from hosting, however inadvertently, anti-vax material, to the cost of combatting it, but I will leave it there.
Q84 Chair: Can I pick up on that point? It is not so much the amount of money, but the fact that it can be monetised, and there is therefore an incentive to put out false information. That is one of the key things. I am wondering whether or not you should perhaps, in the circumstances—or right the way through—put this money in escrow, and then be able to reclaim it, so that if people are found to have breached the terms and conditions of this in a reasonable period of time—say, 30 days—that money is effectively brought back in, so they do not make any money from it at all. That would disincentivise disinformation.
Iain Bundred: Yes, our objective is to have zero transactions. It is not just about creators who put up this bad content; we also do not want the advertiser who has put an ad on that video to be out of pocket for an ad that should never have been served. As it happens, because of payment terms and stuff like that, we do stop a number of videos within 30 days. We have stopped those transactions, but there have been transactions that go through sometimes. It is a very small amount, if you think about the process.
It is worth saying that on YouTube, to be part of the YouTube partner programme, you have to be a certain eligible creator. You have heard evidence from our company before around this. You need to have a certain number of watch hours; you need to have over 10,000 people working; and so on. I think you’ll find that not many are able to monetise this. For example, on the question around actors, bots cannot monetise on a platform, because they do not have the credibility to be able to benefit from advertising.
Q85 Chair: That still does not go against my point about putting this in escrow. You could say to people who are putting content out, “Okay, you can do this, but it will be taken down, and you also you won’t make a penny from it. We can promise you that, because we will keep hold of that cash for a month, and then you will make nothing from it”. That would seem to me a massive disincentive. It would not mess around with your business model too much, either. It would seem to be a very sensible means of stopping this at source.
Iain Bundred: I worry slightly, because if it is a Covid medical misinformation-rated subject, it is a very small group of videos. Our filters should have seen it. I am happy to consider that and—
Q86 Chair: It may only be a small group of videos, but you are hardly like a church noticeboard. You are YouTube. You are the biggest platform on the planet in terms of video-sharing. You made the point to Greg earlier that word of mouth is the issue; you are responsible. You are effectively responsible for ensuring that this Covid disinformation does not get out there. I do not think you ought to abrogate responsibility. You ought to look at every single means of disincentivising those who would cause online harm and disinformation. Do you accept that point?
Iain Bundred: I absolutely accept we have huge responsibility here, but the only way I could perhaps think about applying a rollout would be to escrow every single monetisation on our platform. If you think about the £1.4 billion that we are investing in the ecosystem across the UK into UK creators, that would have a massive effect on that. Specifically to Covid, any references to Covid, perhaps we could look at that. I just think that the truth is that—
Q87 Chair: Sorry, Iain, if you buy something on eBay—that is quite a big marketplace, is it not?—you are given time. When you pay through PayPal, there is a time between the transaction and you receiving the payment as the seller. That seems to be very logical.
Iain Bundred: As I said, that is also the case here. It is not necessarily 30 days specifically, but we have a regular payment process. There are times that we do remove that video. If a video is removed during that period, the creator is not paid.
Q88 Chair: We have seen examples before where disinformation is effectively monetised. We have seen that on countless occasions in this Committee, and you know that is the case, so I do not accept that point.
Iain Bundred: It has never gone past 30 days, though. That is the point.
Q89 Chair: Rebecca, to touch on your point regarding demoting rather than removing content, it is an interesting point, and I can see how that works on newsfeeds and with our algorithms. How does that work for Facebook groups?
Rebecca Stimson: It is exactly the same. All our machine learning, policies and moderation that apply to groups are the same. If there is violating content in a group, it will be removed. If a group consistently has violating content, that whole group will be demoted. We also remove it from things like search and recommendations. As I have said, that can very significantly impact who sees that group; there can be up to 80% reduction in engagement. If it continues to violate, to your last point, we will remove it entirely, and we also have ways of demonetising anyone who violates our misinformation policy.
Q90 Chair: If someone posts Covid disinformation on a Facebook group, are you saying you have algorithms that will effectively demote that, or are you just relying on monitors?
Rebecca Stimson: We have a mixture of human review, reporting and machine learning, so it will depend a little bit on what it is. Where there are posts in that kind of group that violate our policies, it will either be removed or downranked; then the whole group can have the same treatment as well. Also, people are searching for or looking for groups related to Covid-19, many of which are really good. We have seen millions of people join groups to support their communities and their neighbours, so I would not want to give you the impression that Covid groups are all just factories for misinformation. Many of them are fantastic in the work they are doing. When you search for Covid-related groups, you immediately get directed to accurate sources of information as well as an extra layer of the policies we have.
Q91 Chair: One final question, as this is a theme of today. We have had the online harms response from the Government this week, in which they mentioned a greater degree of age verification to prevent harmful content around children and pornography. How does that work for you, given the reluctance in the past to have very robust age verification? How are you seeing that playing out, and how robust do you think it will be?
Rebecca Stimson: Age verification, as we have seen a number of times, is genuinely quite difficult. The way we do it at the moment is we ask people to input their age. If they get it wrong, they cannot join, and they are not allowed multiple attempts, and so on and so forth, but that clearly is not absolutely watertight. We will find and remove underage accounts when they are on the platform. If you are younger—for example, if you are 15—and you are allowed to be on Facebook, there are a number of features we do not allow, such as your location and a few other things like that. That is to protect younger users who are, perfectly validly, using our platform; it just reflects that they are young.
In terms of the online harms proposals, the term “age assurance” is in the online harms proposal, and we are interested in unpacking that a bit more as these conversations go forward. Verification is difficult. Obviously it is effectively like a form of ID for everyone, and when you come to younger users, that is particularly complicated. One of the things we are working on at Facebook is machine learning that can accurately identify and tell the difference between people who are too young to be on our services and those who are not, and we will be able to find and remove those people without requiring everyone to input possible private data and identity. We are interested in working through the online harm process on how the Government are thinking about that.
Q92 Chair: Your answer is an algorithm, is that right?
Rebecca Stimson: There are lots of advantages. If you can train machine learning to be accurate about finding people, it removes the need to collect a lot of private data about users to verify their age, so there are advantages to that system.
Q93 Chair: There must be some merit in proper age verification. If you want to go into a pub, you have to show your ID to get a pint. Why should you not show ID to prove that you are over 16 or over 18 when accessing platforms and content?
Rebecca Stimson: You could do it that way. There are some interesting questions about privacy and data processing if we are to have that kind of information about people in order to deliver our service. There is also some interesting research about the socioeconomic impact of that; large sections of society do not have ready access to things like credit cards or passports, and you would potentially be excluding them from access to these services. Definitely both are up for debate, but what we are thinking about is whether algorithms and a machine-learning approach might be more effective and not have some of those problems.
Q94 Chair: Theo, this is a very pressing question for TikTok, given its much younger demographic. What is your interpretation of age verification and age assurance, and what do you think will happen? Where will we be?
Theo Bertram: We discussed this a little when I was here last time, so forgive me if I repeat myself. You have to be over 13 to use the platform. You have to enter your date of birth. We do not say, “Are you 13 or over?” we mention a date of birth. If you are not, then you do not have access to the app. That is not where it stops. Once you are on the app, another user can report that they think a user is underage. Any time a piece of content is reviewed by one of our moderators, they are also looking to see whether that user is under 13.
Q95 Chair: We get where you are now, and clearly across the whole of social media it is inadequate. That is part of online harms. They are telling you it is inadequate; therefore we have to have a new system. Precisely what do you think that will mean in the future?
Theo Bertram: I think the new age assurance means you need to think carefully around the risk and the age of your users, and you need to manage those two things. We have already started to take steps around things like direct messaging, which we will not allow for those under 16. It is much more that we need to think about how we frame our product in a way that is safe for 13 year-olds. That is a key part of it.
When we have run family focus groups in the UK, which we have done with independent groups, one of the very strong feedbacks that we get from parents is they do not want to have to give their password details or whatever. I do not know how it is we are going to verify that a 13 year-old is a 13 year-old. They do not want to have to give that to every app. For us, as a challenger to the two other companies that are here, if you have a rule that every app must hand over data to be used, there will be far fewer apps that are used and much less data that is handed over in that way. I think that will benefit the other players, rather than those that might challenge them. We need to try to find solutions that work for everyone, and not just for the very biggest.
Chair: I am going to wind it up there, because we are over on time, but thank you for your evidence today. It has been very interesting to see precisely where we are at. Could you write on the matters that we have discussed, in terms of what Greg Clark said about co-operative working and your plan to do that?
I would also like to know precisely the interactions you are having with Government on these matters, in private. It will not be for publication. We are keen to know, during this absolutely crucial period, that the right messaging is getting out there, and that you are coming down on disinformation. We would also like the latest details, particularly from Facebook, not to the end of October but beyond that, in terms of what is happening on the ground when it comes to disinformation.
Thank you again for your evidence today. We are going to have an adjournment while we set up the second panel.
Examination of witnesses
Witnesses: Professor Tom Rodden, Sarah Connolly and James Sorene.
Q96 Chair: This is the Digital, Culture, Media and Sport Select Committee. This is our second panel to appear before the sub-committee on the subject of disinformation relating to the Covid vaccine. We are by Thomas Rodden, Chief Scientific Adviser, Department for Digital, Culture, Media and Sport, Sarah Connolly, Director of Security and Online Harms at DCMS, and James Sorene, Deputy Director, Covid-19 Vaccine Communications, Department of Health and Social Care.
My first question is to all of you. How do we know that the disinformation unit is having an effect in relation to Covid-19 vaccine disinformation, and what data are you analysing in order to gauge how effective you are in countering vaccine disinformation? Who would like to take that question?
Sarah Connolly: Perhaps I could start and then I can hand over to Tom, who will know much more about the measuring. The short answer is this is quite difficult. It is difficult to understand in any great detail exactly the volumes, and exactly what is coming down and what is not. We do have some indicators. Ofcom published some data recently that said that fewer people were reporting having seen false or misleading information online, so we have some indicators, but there is no denying that it is a challenge to understand exactly, in terms of specific numbers and specific data.
What we also know, and you have just heard from the companies, is they are leaning into this. We are working very hard across Government to understand the problem, and to take action as quickly as we can. We have a system that is better geared than it was nine months ago to do this and, as I said, we have some indicators. Perhaps Tom might want to talk a little bit more about the specifics.
Q97 Chair: One second there. On what you are saying about companies working together with Government, they were not that specific. “Come and get us,” was the line they had for us on this particular issue; it was, “We are here to help”. It does not feel as if there is a huge amount of day-to-day involvement there, or that they are being utilised to the extent they need to be. Is that a fair impression or not?
Sarah Connolly: I think not. We talk to them. Would it help if I talk a little bit about the cell—the structures and how it works?
Sarah Connolly: The cell itself is effectively a hub that sits in DCMS. It has two functions. The first is it pulls together analysis from all across Government; lots of different Departments, and lots of agencies, feed information in on different aspects of disinformation and misinformation. We can come back to the difference between those if that is helpful in a bit. It pulls it all together and it analyses. It takes a judgment as to what is gaining prevalence and what is out there; then it helps co-ordinate the response to that. If it is the sort of thing that needs rebuttal, that happens quite quickly from Cabinet Office. If it is something that needs more authoritative information, certainly anti-vax goes to DHSC and James’s team. It does a bit of that, but we try not to be sand in the machine. Where it is obvious where it should go, it goes there.
The other big function it has is talking to social media platforms and passing information over. It gets information back from them, and encourages that swift takedown—the swift dealing with the platforms. The cell has daily interactions with almost all the platforms.
Q98 Chair: To be clear, what happens? You see a piece of this, and then send an e-mail and then do an act on it—is that the process?
Sarah Connolly: It depends. Each platform is a slightly different set-up. For some of them, we have what is called trusted flagger status. If somebody from the cell says, “We are worried about this,” that goes immediately to the top of the pile. Whoever it is in whatever company then acts on it. It is the same system they have across Government for things like terrorist content.
Q99 Chair: Is there any data available? I am going to go to Thomas on data, what disinformation you see, and how effective you are. Are you having any real, major impact on the number of phony videos and posts?
Professor Rodden: The first caveat is that the metrication measurement in terms of state of the art is patchy. You have no standard means of measuring this, and quite a plethora of APIs and different techniques for accessing this information. What you end up with is different metrics about how to reconcile them. What we do track are the volumes and levels at a broad trend level. What we have seen is that misinformation tends to go with the rhythm of news. We have seen a decline in Covid misinformation, but what we are anticipating would happen is an increase as the vaccine rolls out and that hits the main newsfeed. You see those general trends. Specific pulldowns at a very granular level are quite hard to measure.
Q100 Chair: What you are saying is there is not what you would call very reliable data, something that you can get a hold of. It is a shifting picture?
Professor Rodden: That is right; it is a shifting picture. There is nothing at the moment. There are a number of research programmes that are seeking to improve that.
Q101 Chair: How can we judge your success, then?
Professor Rodden: The success and the judgments of the success come from the impact and effect, rather than the specifics of the data. You can see the effect in the general trend, but in specific very fine-grained data, that is going to be quite hard to do with the current state of the art.
Q102 Mrs Heather Wheeler: Perfect timing, Chair, because my questions absolutely follow on from how we measure success. The vaccine is only useful if enough people take it. Let’s try to get this granular stuff here. How many meetings have you had to discuss how you get that message out?
Professor Rodden: In terms of the message and messenger, James is probably the best person to respond to that, given that he leads in the comms meetings. From a science perspective, and from the SAGE and science side, what we have done is structure a series of meetings where we explicitly meet with James and his counterpart and talk about the behavioural science inputs and the means by which they are messaging. I would like to hand over to James on the specifics of the campaign.
James Sorene: Yes, I am happy to pick that up. I am leading the communications strategy across Government. I should say at the outset that the most important thing to tackle this problem is to get as much clear, quality and proactive information out there as possible. What we are driven by is using our best spokespeople to build a partnership, not just in Government, but across all the health partner agencies, stakeholders and partners, right across from health charities and royal colleges to local government. There is a very, very big coalition working together to try to get these messages out. In terms of how many meetings we are having, we are having a very large number of meetings every week to brief all the people I have just mentioned. This is very dynamic and is changing all the time.
It is probably useful to orientate you on where we are in the process. We had a phase where there was no vaccine that had been approved and we were talking about the vaccine in the abstract. Then we got to a point where the vaccine was approved, and now it is being delivered, but it is being delivered in a very phased way across the country. These are difficult messages to be doing in parallel in this way, but the most important thing that we are trying to get across is that the vaccine is simply the most effective way to combat coronavirus and protect people. That is our number one message.
In terms of what we are doing with the companies, which is what you were alluding to, perhaps, with the meetings, our most important job with them, my role, as you heard in the earlier evidence, is to get information to them around giving them the tools, so that they do what they are supposed to be doing in terms of fact-checking. You heard the example, which was quite simple, of animal products in vaccines. Through a good flow of information, they are able to remove content that is simply inaccurate. Together with what the counter-disinformation unit is doing, we are looking very closely at the themes that are coming through in this disinformation, and how they end up moving across to a lot of other audiences—the more vaccine-hesitant, rather than the actual campaigners who are spending all their time trying to campaign against vaccines.
Q103 Mrs Heather Wheeler: That is an interesting answer, and I can understand why you, as a comms professional, need a message on the anti-vaxxers, a message on the hesitance, and the message that the vaccine works and is not full of stuff that should not be there. Could you give me an example of where there is counter-propaganda from the Government that you have deployed against the anti-vaxxers, please?
James Sorene: I would not want to picture it as counter-propaganda. We are in a pandemic, and there is unprecedented interest in, and concerns and questions about, the spread of coronavirus and the treatments. The vaccine now is the most effective tool that we have to protect people. What I am interested in—this is what our strategy tries to do—is answering people’s questions. There are people out there with a lot of questions about how the vaccine is going to work, who is going to get it, the side-effects and how it is tested. That is the position we are in: trying to listen to those questions and answer them.
My advice to people locally, and those in communities that are being targeted by anti-vaccination propaganda, is always not to take that on directly, but to look at the themes being raised, and to see if that is resonating with that particular audience and community. Why is that audience and community being targeted, and how can we, with our proactive communications, address the concerns that those themes are trying to target? I can go through some of those themes if it is useful.
Q104 Mrs Heather Wheeler: I wonder whether that is a bit too esoteric. We have a real problem now; one anti-vaxxer alone has 600,000 followers, and you are talking about, “Maybe if I drill down into this, I can talk about something else”. I am not sure that is strong enough, frankly.
James Sorene: I am not saying we should talk about something else. I am saying that we are trying to have a very useful, concrete, proactive conversation with people across the country to answer their questions about this particular vaccine. If you want to get more into the detail of how the Government and their different Departments address directly some of the anti-vaccination activities, then we probably need to go back to some of the detail of what the counter-disinformation unit is doing.
Mrs Heather Wheeler: To be honest, that is what we were talking about, so perhaps you need to write to us on that.
Q105 Kevin Brennan: Sarah, do you have a good measure of how much anti-vax material is in circulation on social media and how many people are reading it or seeing it?
Sarah Connolly: It is a very good question. In terms of numbers, I am not sure that we do. We are reliant on the companies having that conversation with us and working with us to try to understand. This is a world of grey, but it is quite challenging to understand the difference between somebody seeing it and engaging with it, somebody clicking on it and clicking past it, and somebody liking it, thinking it is something else. It is quite difficult to understand not just the volume but the levels of interaction with it. What we can do is have a sense of the sorts of themes; James alluded to that. We know broadly the kinds of themes that are coming out, but that is not quite the question that you asked.
Q106 Kevin Brennan: On the information you do know about, do you go looking for that, or do you have agreements with the social media companies to report to you this sort of thing?
Sarah Connolly: The answer is we do a bit of both. We definitely go looking for it ourselves, and that is partly what I was talking about a moment ago with different parts of Government looking at different aspects of it. We definitely go looking for it ourselves. We also have a very strong partnership with lots of academics, lots of civil society groups that also go looking for it and will tell us. We also have a live conversation with the platforms about what they are seeing. I do think it is one of the areas that we could get better at with them, that conversation around greater transparency about what they are seeing and feeding it back. It is one of the areas that we want to keep working on, but we pull from as wide a group as we can.
Q107 Kevin Brennan: From what you have seen, what kind of anti-vax material causes you the most concern?
Sarah Connolly: You will forgive me for not giving specific examples, because there are a lot of variations. There are broadly three themes, with a fourth. The three that we worry the most about, there is material around speed, the vaccine was done too quickly, it is not safe, those kind of narratives. The second one is around side-effects. The third is around monetary and big business and links to pharma. The fourth is there is, as the Committee pointed out in the previous session, a long history of anti-vax and in some communities it is quite embedded. From those sometimes you get the slightly outlandish 5G-type conspiracies. Those are broadly the themes that we see.
Q108 Kevin Brennan: You said you did not want to get specific on it, but how are we to judge what kind of information we should be concerned about as a Select Committee if you as a Government are not able to tell us which ones you are most concerned about?
Sarah Connolly: We are worried about all of them.
Q109 Kevin Brennan: Are there specific examples that you have come across that you have evidence of as genuinely influencing public opinion against vaccines?
Sarah Connolly: Not strong evidence that I can think of, not a single case.
Q110 Kevin Brennan: Which platforms are of concern to you in how they respond to this in terms of their policies and their approach?
Sarah Connolly: Again, that is not a straightforward question. The platforms are all different, at the risk of stating the obvious.
Q111 Kevin Brennan: Who is the best and who is the worst?
Sarah Connolly: I am not sure I wish to get drawn on who is the best and who is the worst. They each have strengths and weaknesses.
Kevin Brennan: You must have an opinion after working with them.
Sarah Connolly: They are all different and they have all adapted in slightly different ways. The thing that we would like to see is more of the transparency that I talked about. The other thing, and again you touched on it in your previous session, there is probably more we can do about talking and sharing information. Forgive me, I cannot remember which member asked the questions about sharing information and the degree to which they can talk to each other. They do talk to each other; I think there is something about making sure that they really understand and we understand from them how things move between them.
Q112 Kevin Brennan: Did you say that they could all be more transparent? What we are looking to do is to make some sensible recommendations here around this. It is quite useful for us to know what you think about their current practices and whether we should be making recommendations either aimed at all of them, some of them, or one or two in particular to improve their practices. If you were helping us to write this report, what would you recommend?
Sarah Connolly: The two things for us are transparency, so more information about what they are seeing, what kind of volumes and have they taken things down. More light shone gives us all a better understanding about what to target and where. The second thing is the themes and narratives that they are seeing between them. You know yourselves if you see something from Twitter you might WhatsApp it to a friend, so things move between platforms and the more understanding that we can get of those and how they move the better we can all tackle it. I think those are the two things.
Q113 Clive Efford: Can I go back to an answer you gave to the Chair earlier on about the hub or the cell that you work in being in daily contact with the media platforms? How far does that go back? How long has that arrangement been in place? Has that just been to do with anti-vax or does it predate that?
Sarah Connolly: It predates it. We first set up the cell for the European elections last May and I think that was the first time we did it. It was relatively low-key; we wanted to see how it worked. The major test that we used the cell for was the general election last year. We stood it up, it looked quite different, we stood it back down again and then we stood it up this time on 5 March and it has been stood up ever since. In any case, colleagues working day in and day out on disinformation are in contact with colleagues in social media platforms regardless of whether or not the cell is stood up. Certainly as soon as you go to that footing then it is daily—on occasion hourly—contact.
Q114 Clive Efford: There was a meeting on 8 December, a roundtable with the Ministers and the media platforms. How did that move things on? Was that a staging post that got the platforms to be more responsive? What was achieved at that roundtable, given that you had this relationship ongoing from previously?
Sarah Connolly: I think that meeting was a very helpful first iteration of a policy group. It was always intended to be the start of a more strategic conversation; that is perhaps the best way of putting it. We have had quite frequent—as I mentioned, daily, sometimes hourly—contact on specific incidents. What we wanted to do was raise it to the next level. That conversation started with Secretaries of State meeting some of the platforms, I believe twice. This policy working group is intended to make sure there is a rhythm of conversations on things like transparency and the commonality of what people are seeing. It is to try to tackle those slightly bigger issues.
Q115 Clive Efford: Are there any specific asks of Government that the media companies have come forward with? Is there anything that they specifically need?
Sarah Connolly: I know that they are keen to keep understanding from James and colleagues what we are seeing, in terms of vaccine rollout, vaccine hesitancy, and the numbers, so that they can respond in kind. That is a thing that they ask for, and that I think they are getting from us.
Q116 Clive Efford: In terms of the overall attitudes of the public towards vaccines, how important are these social media platforms?
Sarah Connolly: In terms of people getting their news from them?
Q117 Clive Efford: In terms of attitudes towards having the vaccine. Are they the main source, for instance, of people being reluctant to want to take the vaccine? How important are they in that discussion?
Sarah Connolly: I am not sure if Tom or James wants to come in.
Professor Rodden: They are one of a set of forums. If we look at the behavioural barriers to vaccination, a number of them are practical barriers based on access, but in terms of knowledge barriers, people have concerns about effectiveness and about understanding side-effects, and there they turn to a range of trusted sources. When people have been asked, they say they turn to their local trusted sources, local communities, GPs, scientists and the media generally.
Obviously, social media is quite an important part of the media side of that tapestry, but generally speaking, they draw on a broad portfolio of sources in making their judgments. That kind of behavioural knowledge feeds directly into James’s campaigns, when it comes to resourcing different channels and sources of information.
Q118 Clive Efford: What information are people using to make choices about whether to have a vaccination? Is that coming chiefly from social media, or are there other sources that they are picking up that negative information from? I am trying to get a feel for how influential in this debate social media platforms are.
Professor Rodden: I understand your question. The one thing to say is that individual judgments that people make are a tension between the sources of information—the positive sources, and things like misinformation. The role of misinformation is one factor in their judgment call. Social media is one of the sources of misinformation. When you look at surveys, you will often see a gap; people report that as little as 3% or 4% of their judgment is based on misinformation. Again, James would have a significant amount of this information from his regular surveys and links with people. We track that to see what proportion of that judgment is happening.
James Sorene: I think the advice from your GP or health professionals is hugely important in the decision to take a vaccine, as are authoritative sources like the NHS website. That is why it is so important that the social media companies are signposting users to the NHS website—to NHS.uk/conditions/coronavirus-covid-19/coronavirus-vaccination/ coronavirus-vaccine/, which is where all the information is around the current vaccination programme. That is hugely important.
Also, I know we are talking about social media companies, but we must not forget that traditional TV news is still overwhelmingly the source of information for people. We have absolutely been using that as much as we can, and using all the opportunities that have been given to us on this journey of approving vaccines and making them available around the UK. We are constantly reinforcing the messages around who it is available for, why it is so important—because it protects people from coronavirus—the safety profile of the vaccine, and the fact that it is approved by bodies such as the Medicines and Healthcare products Regulatory Agency. We ensure they have the opportunity to talk about their process and how this has been approved.
We know that people, if they do feel more hesitant, want to hear a lot more about the approval process. They want to hear more about the process of clinical trials and the detail of that. It is important that the MHRA and its leadership are able to have opportunities to explain more of that. That is what has been happening over the last few weeks. We have been using Government and NHS social media channels to do exactly what you heard YouTube requesting: that the Government should use a lot more video content, and not just TV news. We should also make sure that when our spokespeople, local clinicians and experts are talking on TV, we badge that up, package it up into smaller clips, and share it across a whole range of social media platforms.
Some of our social media clips like that are reaching millions of people. The interesting thing about the level of engagement and discussion, and the retweets and the comments, is that you are seeing a live discussion; users are talking about the issues and raising questions, and then other people are answering them. That is exactly the type of discussion you want to be having in this space.
Q119 Clive Efford: You may have heard Rebecca Stimson talk about preferring not to take some content down, because it provided the opportunity to educate people. I do not know who would be best placed to answer this, but have you seen evidence of that taking place, and how effective that has been?
James Sorene: I am happy to start the answer and then I will let my colleagues contribute. I think we all need to be aware that some parts of this topic are an easy way into talking about the vaccine. There is fascination with it in some sectors of the media, and in others, they want to talk about vaccine misinformation, sometimes in a way that is out of proportion to its scale. I do not think that is helpful.
What I do think is helpful is where there are questions—people have a lot of questions; this is a new virus, we are in a pandemic and this is a new vaccine—they are answered. Certainly the debates and arguments going on in some of our social media channels are incredibly useful. If it generates interest in local communities, and that gives directors of public health, NHS leaders and well-known GPs forums with local communities, particularly communities that are more hesitant and want to discuss this, and they get the authoritative information and advice, I think that is very constructive.
Q120 Clive Efford: Do the social media platforms co-operate with you and use your sources of information when they are responding to anti-vax misinformation on their platforms?
James Sorene: My DCMS colleagues can contribute with more detail, but where you are searching around coronavirus, particularly around coronavirus vaccines, yes. The top of a Facebook page, for instance, will directed you to the NHS vaccination page. They are made aware by us of the content that we are putting on, in terms of a key video and infographics, and they can place them within their platforms in the way that they see fit.
Q121 Clive Efford: Sarah and Tom, do you have anything to add? Do you think that the social media platforms are taking this seriously enough, and respond robustly and quickly enough?
Professor Rodden: I certainly think they are taking it seriously enough. I think the speed of response and how they respond is a bit of a journey for them, but they are working with us on that, from my observations. I do want to stress also that the social media platforms provide a communication channel for information as well as misinformation. A large number of our scientists have direct access to large populations of people to explain the virus and there are a range of initiatives globally through the WHO and nationally that look to do that. One that was previously mentioned was the #TeamHalo on TikTok, which is a collection of scientists that provide detailed information, often rebutting some of the claims and taking some of the claims and unpacking them as well. They do provide that channel as well.
Q122 Clive Efford: What do you make of the argument that it is better sometimes to have the misinformation there so that you can challenge it? Is that an argument that you subscribe to?
Professor Rodden: I think I would answer that in a broader way and my apologies for doing that. There is a danger of becoming myopically focused on misinformation and not seeing misinformation in a broader landscape. The reports from the academies have highlighted the need to think about this, to think about that misinformation and the information around it. Behaviourally, there is always that line of taking misinformation and treating it in such a manner that you inadvertently legitimise it through taking it down, or through breaking it at such a point that it becomes the story in itself and legitimises the information. There is a fine balance, which I know is a broader answer to your question.
Q123 Clive Efford: Is there anything like a wish list of what you wish the social media platforms would do, or respond to, that would improve the situation?
Professor Rodden: The main thing to improve the situation—this was the focus of the first policy forum—is to think about transparency, and access to data, including for researchers, so they can understand more clearly the phenomenon of misinformation and how it spreads. Also, stronger data-sharing partnerships would allow us to understand what is happening scientifically, and to understand the appropriateness and effects of different mitigations.
Q124 Chair: To follow up on that, Professor Rodden, you talked about the need for transparency—the need to share data with academics and those who have an interest in this area. How much data are the platforms sharing with you?
Professor Rodden: Sharing with academics generally?
Chair: I know what the issues are with academics, when you have Facebook, for example—
Professor Rodden: I presume not me personally. I presume you mean with the Government?
Chair: Yes, the Government.
Professor Rodden: Generally speaking, I am not sure a volume of data is going to be the correct answer for you here. I think it is probably best to give you the channels of data. They make available a range of APIs that allow us to access data and track through that data. I think they, as always, could do more, and more consistently across all the platforms. In some parts we have a range of APIs that give us programmatic interfaces, so that programmes do this, rather than people, which gives us a footprint of the data. Some others provide less information through that computational means, which means that you have to use techniques that are much more cumbersome. So we need less volume but more access.
Q125 Chair: Yes. To think of this politically, and as a layman in this respect, it does seem to be quite astounding that we have to beg these companies to share data in the middle of a public health crisis. Should it not be beholden on them to share as much data as you want, not as much data as they feel free to hand over?
Professor Rodden: I think there is a complex set of things to unpack, not the least of which are issues to do with privacy and access to various forms of personal information, the de-anonymisation of information and also how the process works. I think Sarah can probably talk a bit more operationally about what the particular programmes have been. I sense less of a reluctance to share information on Covid, but more of a concern about how much to share and how they share it without impinging on other parts and other issues, such as privacy.
Sarah Connolly: I think Tom has said exactly the right thing. It is less of a reluctance and more “How do they make it work? how do they make it happen?” There is also a volume question. We have very much focused our efforts collectively on taking down things that we are worried about and, as I said a moment ago, we are starting to pivot now to the bigger strategic questions about how we make things work in the longer term. Then obviously there is online harms—
Q126 Chair: Sorry to cut across you. He has told us that the companies are not stepping up to the plate as much as he would like them to do. That is effectively it, in terms of the quality, if you like, and the transparency. You have mentioned transparency yourself. Do you not think that it is genuinely beholden on them in this emergency to do that no matter what? Also GDPR does not even count in the Covid crisis in that respect. There are get-outs of GDPR in Covid because of the nature of the crisis we face.
Professor Rodden: As a way of balancing it, the scale issue and the technical issues are significant here, in terms of them stepping forward. We want to see some of those technical mechanisms develop. In terms of scale, something like 300 to 500 hours of video are put on YouTube each minute, and over half a million comments per minute are generated on social media platforms. That is a tremendous volume of information, and the technical challenges to do with how you provide access to it, how you sample that and how you do that so that it can be properly ingressed are quite substantial. I would not want to put those aside.
Q127 Chair: So it is not just a matter of not wanting to, although we have noted as a Committee on many occasions that the platforms do not like to be transparent. In Facebook’s case, it only uses partner researchers, thereby retaining control—something this Committee has been quite critical of. In that respect, you are saying there is a huge volume of this stuff, but the truth of the matter is you are suggesting that it may be process, rather than will—or is it a combination of both?
Professor Rodden: I suspect it is a combination of resourcing, process and will. I think it is a combination of those.
Chair: Would you do me a favour? You can write to me confidentially on this if you like. Could you please let us know precisely the areas where you are not at present receiving the information in the way you wish to receive it. We can take that up with the social media companies in writing and we will see whether or not we can get any movement there. I think that in a crisis like this you need to move on from something like that.
Q128 Alex Davies-Jones: Thank you to the witnesses for joining us this afternoon. I am quite conscious of the time, so I will keep my questions brief. Sarah, how much experience from coronavirus and the anti-vaccine misinformation will feed into the upcoming and long-awaited online harms legislation?
Sarah Connolly: A lot.
Q129 Alex Davies-Jones: Perfect. The next question for you as well, have you considered how to tackle the problem of niche social media platforms that are disproportionately spreading anti-vaccine misinformation? For example, we have heard of sites such as Gab, 4Chan and Telegram and these are currently exempt from the online harms regime, despite being notable for spreading anti-Semitic conspiracies about coronavirus.
Sarah Connolly: We have thought a lot about it. I should fess up to being the director in charge of online harms, and the full Government response coming out this week was my team’s doing. When I said glibly “a lot”, I did mean that we are working very closely on both the online harms and the disinformation side. You are absolutely right. There are platforms out there that are niche, that are small and are deeply problematic, both in the way that they behave and in the content that they have. To be crystal clear, anything that has user-generated content is in scope, so they will not be exempt. If it is available in the UK it is also in scope, so even if they are based elsewhere.
You are quite right that they are problematic and some of them take a view that they will simply not talk to Governments about anything. Others will take the view that they will talk to Governments or they will seek to remove things that are clearly illegal, but not issues that are legal but harmful, and I would include disinformation in that space.
We are working in the online harms framework to attempt to give the regulator some teeth to be able to engage with platforms and to tackle them head-on if they do not comply, but it remains a challenge. We will continue as we work up the online harms proposals with Ofcom, as the regulator, ways in which we can tackle this. You are absolutely right; it is a challenge.
Q130 Alex Davies-Jones: Yes, absolutely. A challenge hopefully that you will be able to tackle. I am grateful for your work on the online harms paper that was finally brought forward this week, but we do know that the Government have been significantly delayed in introducing this online harms legislation that this Committee and many other MPs have been calling for for years. Do you think that if this legislation was in place now we would be better placed to tackle the misinformation around the Covid-19 vaccine?
Sarah Connolly: It is very difficult to tell, not least because we do not have a completed Bill. That Bill has not passed through the House and it is difficult to know exactly where it will end up. I think some of what we are trying to do with the companies and elsewhere around disinformation does read across to what we have said in the full Government response that we would like to see, but that is a slightly tricky question to answer.
Q131 Chair: It is extraordinary. You are suggesting that the online harms itself would not make much difference or you would not know? That is hardly a vote of confidence, is it, in what has been brought forward?
Sarah Connolly: I do not think that is quite what I said. What I said was I do not know what the final shape of the Online Harms Bill will be. I do not know what the Online Safety Act will look like and so it is a bit difficult to know exactly how that might tackle it.
Q132 Chair: It would be a grave error if it was not able to deal with something like this. One question before I bring in Greg Clark, our final questioner. On online harms, age verification, age assurance, you will have seen the first panel. They were not exactly bowled over with the idea. They were not exactly keen on putting robust safeguards in place to prevent children being subjected to online harm. What is your view on that? Where does this eventually land?
Sarah Connolly: There are three things that the full Government response on online harms says. All companies must tackle illegal content, the most high-risk companies must tackle legal but harmful, but all companies must do something about children and what is appropriate for children is different from what is appropriate for adults. We and the Ministers would say that it is very important that proper systems and processes are in place to safeguard children online. We think age assurance certainly plays a part in that and in fact Government produced about six weeks ago a report on verification of children online and tech is changing in this space all the time. We are also doing quite a lot of work to support the safety tech industry that does things like age-assure. There is a difference of course between age assurance and age verification, the latter being the gold standard. It is quite hard to do.
Q133 Chair: Why can we not have the gold standard when it comes to children?
Sarah Connolly: I think for certain things you probably do want the gold standard. What I was going to say was there are different risk profiles and that probably needs to be taken into account. Age verification is quite challenging for under-18s, so at 18 you have things like credit cards, bank details, something that will allow a higher threshold to happen more easily. That does not mean that there aren’t technologies that can help identify younger than that, 13, 14, things like linguistic identification; the way that a 13 year-old speaks or types is different to the way a 35 year-old does. There are a range of these things, but we think it makes sense to match the kind of technology with the kind of risk that a platform or a company and scope are posing.
Q134 Chair: So it is more than just algorithms?
Sarah Connolly: Yes.
Q135 Chair: It will be designed according to the potential risks of that platform?
Sarah Connolly: Exactly.
Q136 Chair: So there is a possibility there that should it be deemed a high-risk environment then the requirement could be much closer to verification rather than assurance? Is that fair?
Sarah Connolly: Yes, that is exactly right and it would be a conversation for the regulator to determine, but you are exactly right.
Q137 Greg Clark: Perhaps I can turn to James Sorene. Mr Sorene, you are at the Department of Health and Social Care and we are told that you lead all Covid-19 vaccine communications across Government. Is that right?
James Sorene: Yes.
Q138 Greg Clark: What is the monthly budget that you have for communications?
James Sorene: That would involve adding up all the different budgets for the different communications teams among the Departments involved. It is not a figure that I can give you now.
Q139 Greg Clark: You lead it all. You must have an idea. Are we talking millions, hundreds of thousands, tens of millions?
James Sorene: If I can provide some detail about who is involved in this, we have the Department of Health and Social Care communication teams, we have the NHS England and Public Health England. We are working with the MHCLG; we are working with the Cabinet Office. Those are a mixture of full-time and temporary staff, so that would involve a calculation of all their salaries and how many hours they are working on it. If you are asking me what is the paid spend of the paid-for content then that is reported on by the Cabinet Office on a rolling basis.
Q140 Greg Clark: What is the paid spend for communication around the vaccines that you have to deploy as the lead cross-Government for this?
James Sorene: I do not have a pot of money. The way that it is working within the coronavirus response is that it is managed centrally by the Cabinet Office communications. As I said, the reporting on that is on a rolling basis.
Q141 Greg Clark: You are the lead we are told and you just said.
James Sorene: Yes, that is what I said, but I cannot give you a figure for how much is being spent on vaccines right now.
Q142 Greg Clark: Okay, but you do use paid media?
James Sorene: Yes, there is a paid-for element to the campaign around Covid-19 vaccines.
Q143 Greg Clark: Do you use paid social media?
James Sorene: Yes, that is part of the campaign.
Q144 Greg Clark: Do you pay for advertising on Facebook, for example?
James Sorene: There is work that is undertaken with the social media companies, but like I said, I cannot give you a figure for that. It is part of the ongoing campaign.
Q145 Greg Clark: You said you use paid media, you use paid social media. Do you pay for advertisements? I am not asking for the figure—you might write to us—but do you pay for adverts on Facebook?
James Sorene: There is some boosted content on Facebook. Almost all of the work we have done so far on Covid-19 vaccine communications is what we would call organic content, it is just placed there and it reaches the audience without boosting and without advertising. I think the important thing to say is this is material that has huge public interest. Also the other—
Q146 Greg Clark: I am just trying to get at whether you spend money with the social media platforms.
James Sorene: I think I would have to write to provide some more detail on that, but the point I wanted to make was that without spending money—
Q147 Greg Clark: No, no. Surely, Mr Sorene, you are the lead for all Covid-19 vaccine communications across Government. You must know whether you pay for paid advertisements, paid publicity on social media, surely?
James Sorene: I think I answered the question, that the vaccine communications so far, the social media work has not involved advertising. It has been organic.
Q148 Greg Clark: This is a very important point. You do not pay any money to social media platforms?
James Sorene: For the Covid-19 vaccine communications work that we have been doing so far we have not been paying them to boost content. It has been what I call organic content.
Q149 Greg Clark: So why is that, given that we know that the purpose of your work is to communicate important information? We remember over time very important communication campaigns by your predecessors. We think of the publicity around AIDS-HIV at the time, very important hard-hitting messages that were paid for. There was advertising in different media. Are you saying you are not spending money on promoting on social media these very important messages?
James Sorene: There are two answers to that question. Number one is as part of our work with the social media companies they are committed to prioritising and positioning content from authoritative health sources, so they are making that available and they are boosting its exposure in a way that is not routine. We are not in a routine campaign like the ones you have mentioned. They are doing a lot of that through their own procedures.
The other thing is that, as one of the witnesses mentioned, Facebook, for instance, have made credits available to do some of that advertising for free, so that is the reason why so far we have not. There is a larger campaign that will probably launch in the New Year as we make the vaccine available to much larger numbers of people and I would expect there will be some paid element there and we can report back on that.
Q150 Greg Clark: What is behind my questions here is that we know that some of the anti-vax propaganda is hosted by social media platforms; that is the essence of today’s session. We know that. We have talked about the efforts that the companies make to suppress it and take it down, but nevertheless it is there and we know that however inadvertently they do benefit financially from it. I am interested to know whether the Government are paying social media companies in effect to counteract the messages that those social media companies are, through their platforms, communicating.
I will say behind that question I think it would be wrong to do so and I think there is a case for the social media companies refunding the advantage that they gain from that for you to spend. I would be even more concerned if you were paying the social media companies for this, although I reiterate my point that I think it is important to advertise and promote and I hope that they would be going out of their way at this time, when the whole country is pulling together to make available their platforms for your messages free of charge.
James Sorene: As I think I explained, that is largely what they are doing. For users out there who are trying to access vaccine information on social media, what is important for the Government is to make sure that where people are trying to access information that they are seeing prominently Government content and they are being signposted to, for instance, the NHS website. They are committed to doing that and they are putting a lot of resources into doing that. I do not know if Sarah wants to come in on some of the detail of that.
Q151 Greg Clark: We are running out of time, Mr Sorene, but on this you very kindly offered to write to the Committee. I would be grateful to the Chair of the Committee if you could do that and specify how much money central Government are paying, have paid over the last three months, say, and on a continuing basis, if you have that information, to social media platforms for vaccine purposes, and for that matter anything else relating to Covid. That is the first thing, how much money is being paid there on that?
Secondly, what is the paid media spend across Departments that you anticipate making and that you have made? Perhaps finally, a question rather than a request for information, presumably if it were the case, happily, that the social media companies were willing to refund some of the revenues that they have made, you would be a grateful recipient of that, I would imagine.
James Sorene: I think that would be a very interesting idea. I think that a precedent has already been set for that in the example that I talked about, where Facebook is already providing a significant amount of advertising credits. As we have said, we had the meeting with the social media companies and they have set out more proposals and they are on a journey.
Greg Clark: There is no time to be lost on this, so again, as with the previous set of witnesses, if you would write with all of this information before Christmas we would be very grateful.
Chair: Thank you to Tom Rodden, Sarah Connolly and James Sorene.