(2 years, 3 months ago)
Public Bill CommitteesI am sorry, but I must move on. Minister, I am afraid you only have five minutes.
Q
Richard Earley: What information are you referring to?
Data, in particular on the operation of algorithmic promotion of particular kinds of content.
Richard Earley: We already do things like that through the direct opportunity that anyone has to see why a single post has been chosen for them in their feed. You can click on the three dots next to any post and see that. For researcher access and support, as I mentioned, we have contributed to the publishing of more than 400 reports over the last year, and we want to do more of that. In fact, the Bill requires Ofcom to conduct a report on how to unlock those sorts of barriers, which we think should be done as soon as possible. Yes, in general we support that sort of research.
I would like say one thing, though. I have worked at Facebook—now Meta—for almost five years, and nobody at Facebook has any obligation, any moral incentive, to do anything other than provide people with the best, most positive experience on our platform, because we know that if we do not give people a positive experience, through algorithms or anything else, they will leave our platform and will not use it. They tell us that and they do it, and the advertisers who pay for our services do not want to see that harmful content on our platforms either. All of our incentives are aligned with yours, which are to ensure that our users have a safe and positive experience on our platforms.
Q
Richard Earley: I am afraid to say that that is not correct. We have multiple algorithms on our services. Many of them, in fact, do the opposite of what you have just described: they identify posts that might be violent, misleading or harmful and reduce the prevalence of them within our feed products, our recommendation services and other parts of the service.
We optimise the algorithm that shows people things for something called meaningful social interaction. That is not just pure engagement; in fact, its focus—we made a large change to our algorithms in 2018 to focus on this—is on the kinds of activities online that research shows are correlated with positive wellbeing outcomes. Joining a group in your local area or deciding to go to an event that was started by one of your friends—that is what our algorithms are designed to promote. In fact, when we made that switch in 2018, we saw a decrease in more than 50 million hours of Facebook use every day as a result of that change. That is not the action of a company that is just focused on maximising engagement; it is a company that is focused on giving our users a positive experience on our platform.
Q
Richard Earley: No, because as I just said, when we sent the algorithm this instruction to focus on social interaction it actually decreased the amount of time people spent on our platform.
Q
Richard Earley: As I said, it is about ensuring that people who spend time on our platform come away feeling that they have had a positive experience.
Q
Richard Earley: I think that a really valuable part of the Bill that we are here to discuss is the fact that Ofcom will be required, and we in our risk assessments will be required, to consider the impact on the experience of our users of multiple different algorithms, of which we have hundreds. We build those algorithms to ensure that we reduce the prevalence of harmful content and give people the power to connect with those around them and build community. That is what we look forward to demonstrating to Ofcom when this legislation is in place.
Q
Q
Katie O'Donovan: I welcome the opportunity to address the Committee. It is so important that this Bill has parliamentary scrutiny. It is a Bill that the DCMS has spent a lot of time on, getting it right and looking at the systems and the frameworks. However, it will lead to a fundamentally different internet for UK users versus the rest of the world. It is one of the most complicated Bills we are seeing anywhere in the world. I realise that it is very important to have scrutiny of us as platforms to determine what we are doing, but I think it is really important to also look at the substance of the Bill. If we have time, I would welcome the chance to give a little feedback on the substance of the Bill too.
Becky Foreman: I would add that the Committee spent a lot of time talking to Meta, who are obviously a big focus for the Bill, but it is important to remember that there are numerous other networks and services that potentially will be caught by the Bill and that are very different from Meta. It is important to remember that.
While the Bill is proportionate in its measures, it is not designed to impose undue burdens on companies that are not high risk. I have one more question for Richard. I think Katie was saying that she wanted to make a statement?
We are out of time. I am sorry about this; I regard it as woefully unsatisfactory. We have got three witnesses here, a lot of questions that need to be answered, and not enough time to do it. However, we have a raft of witnesses coming in for the rest of the day, so I am going to have to draw a line under this now. I am very grateful to you for taking the trouble to come—the Committee is indebted to you. You must have the opportunity to make your case. Would you be kind enough to put any comments that you wish to make in writing so that the Committee can have them. Feel free to go as broad as you would like because I feel very strongly that you have been short-changed this afternoon. We are indebted to you. Thank you very much indeed.
Richard Earley: We will certainly do that and look forward to providing comments in writing.
Examination of Witnesses
Professor Clare McGlynn, Jessica Eagelton and Janaya Walker gave evidence.
Q
“only entitled to conclude that it is not possible for children to access a service…if there are systems or processes in place…that achieve the result that children are not normally able to access the service”.
Ofcom will then interpret in codes of practice what that means practically. Professor McGlynn, do you think that standard set out there—
“the result that children are not normally able to access the service or that part of it”
—is sufficiently high to address the concerns we have been discussing in the last few minutes?
Professor Clare McGlynn: At the moment, the wording with regard to age assurance in part 5—the pornography providers—is slightly different, compared with the other safety duties. That is one technicality that could be amended. As for whether the provision you just talked about is sufficient, in truth I think it comes down, in the end, to exactly what is required, and of course we do not yet know what the nature of the age verification or age assurance requirements will actually be and what that will actually mean.
I do not know what that will actually mean for something like Twitter. What will they have to do to change it? In principle, that terminology is possibly sufficient, but it kind of depends in practice what it actually means in terms of those codes of practice. We do not yet know what it means, because all we have in the Bill is about age assurance or age verification.
Q
Professor Clare McGlynn: My understanding as well is that those terms are, at the moment, being interpreted slightly differently in terms of the requirements that people will be under. I am just making a point about it probably being easier to harmonise those terms.
Q
Professor Clare McGlynn: I read your piece in The Times this morning, which was a robust defence of the legislation, in that it said that it is no threat to freedom of speech, but I hope you read my quote tweet, in which I emphasised that there is a strong case to be made for regulation to free the speech of many others, including women and girls and other marginalised people. For example, the current lack of regulation means that women’s freedom of speech is restricted because we fear going online because of the abuse we might encounter. Regulation frees speech, while your Bill does not unduly limit freedom of speech.
Q
Professor Clare McGlynn: There are many ways in which speech is regulated. The social media companies already make choices about what speech is online and offline. There are strengths in the Bill, such as the ability to challenge when material is taken offline, because that can impact on women and girls as well. They might want to put forward a story about their experiences of abuse, for example. If that gets taken down, they will want to raise a complaint and have it swiftly dealt with, not just left in an inbox.
There are lots of ways in which speech is regulated, and the idea of having a binary choice between free speech and no free speech is inappropriate. Free speech is always regulated, and it is about how we choose to regulate it. I would keep making the point that the speech of women and girls and other marginalised people is minimised at the moment, so we need regulation to free it. The House of Lords and various other reports about free speech and regulation, for example, around extreme pornography, talk about regulation as being human-rights-enhancing. That is the approach we need to take.
Thank you very much indeed. Once again, I am afraid I have to draw the session to a close, and once again we have probably not covered all the ground we would have liked. Professor McGlynn, Ms Walker, Ms Eagleton, thank you very much indeed. As always, if you have further thoughts or comments, please put them in writing and let us know. We are indebted to you.
Examination of Witnesses
Lulu Freemont, Ian Stevenson and Adam Hildreth gave evidence.
Q
Adam Hildreth: I had covid at the time, yes.
Covid struck. I would like to ask Adam and Ian in particular about the opportunities provided by emerging and new technology to deliver the Bill’s objectives. I would like you both to give examples of where you think new tech can help deliver these safety duties. I ask you to comment particularly on what it might do on, first, age assurance—which we debated in our last session—and secondly, scanning for child sexual abuse images in an end-to-end encrypted environment. Adam, do you want to go first?
Adam Hildreth: Well, if Ian goes first, the second question would be great for him to answer, because we worked on it together.
Fair enough. Ian?
Ian Stevenson: Yes, absolutely. The key thing to recognise is that there is a huge and growing cohort of companies, around the world but especially in the UK, that are working on technologies precisely to try to support those kinds of safety measures. Some of those have been supported directly by the UK Government, through the safety tech challenge fund, to explore what can be done around end-to-end encrypted messaging. I cannot speak for all the participants, but I know that many of them are members of the safety tech industry association.
Between us, we have demonstrated a number of different approaches. My own company, Cyacomb, demonstrated technology that could block known child abuse within encrypted messaging environments without compromising the privacy of users’ messages and communications. Other companies in the UK, including DragonflAI and Yoti, demonstrated solutions based on detecting nudity and looking at the ages of the people in those images, which are again hugely valuable in this space. Until we know exactly what the regulation is going to demand, we cannot say exactly what the right technology to solve it is.
However, I think that the fact that that challenge alone produced five different solutions looking at the problem from different angles shows just how vibrant the innovation ecosystem can be. My background in technology is long and mixed, but I have seen a number of sectors emerge—including cyber-security and fintech—where, once the foundations for change have been created, the ability of innovators to come up with answers to difficult questions is enormous. The capacity to do that is enormous.
There are a couple of potential barriers to that. The strength of the regulation is that it is future proof. However, until we start answering the question, “What do we need to do and when? What will platforms need to do and when will they need to do it?” we do not really create in the commercial market the innovation drivers for the technical solutions that will deliver this. We do not create the drivers for investment. It is really important to be as specific as we can about what needs to be done and when.
The other potential barrier is regulation. We have already had a comment about how there should be a prohibition of general monitoring. We have seen what has happened in the EU recently over concerns about safety technologies that are somehow looking at traffic on services. We need to be really clear that, while safety technologies must protect privacy, there needs to be a mechanism so that companies can understand when they can deploy safety technologies. At the moment there are situations where we talk to potential customers for safety technologies and they are unclear as to whether it would be proportionate to deploy those under, for example, data protection law. There are areas, even within the safety tech challenge fund work on end-to-end encrypted messaging, where it was unclear whether some of the technologies—however brilliant they were at preventing child abuse in those encrypted environments —would be deployable under current data protection and privacy of electronic communications regulations.
There are questions there. We need to make sure that when the Online Safety Bill comes through, it makes clear what is required and how it fits together with other regulations to enable that. Innovators can do almost anything if you give them time and space. They need the certainty of knowing what is required, and an environment where solutions can be deployed and delivered.
Q
Adam Hildreth: I agree with Ian that the level of innovation is amazing. If we start talking about age verification and end-to-end encryptions, for me—I am going to say that same risk assessment phrase again—it absolutely depends on the type of service, who is using the service and who is exploiting the service, as to which safety technologies should be employed. I think it is dangerous to say, “We are demanding this type of technology or this specific technology to be deployed in this type of instance,” because that removes the responsibility from the people who are creating it.
Q
Adam Hildreth: Absolutely. Sorry, I was saying that I agree with how it has been worded. We know what is available, but technology changes all the time and solutions change all the time—we can do things in really innovative ways. However, the risk assessment has to bring together freedom of speech versus the types at risk of abuse. Is it children who are at risk, and if so, what are they at risk from? That changes the space massively when compared with some adult gaming communities, where what is harmful to them is very different from what harms other audiences. That should dictate for them what system and technology is deployed. Once we understand what best of breed looks like for those types of companies, we should know what good is.
Q
Adam Hildreth: The technology is there. It exists and it is absolutely deployable in the environments that need it. I am sure Ian would agree; we have seen it and done a lot of testing on it. The technology exists in the environments that need it.
Q
Adam Hildreth: There are ways that can work. Again, it brings in freedom of expression, global businesses and some other areas, so it is more about regulation and consumer concerns about the security of data, rather than whether technological solutions are available.
Ms Freemont, Mr Hildreth and Mr Stevenson, thank you all very much indeed. We have run out of time. As ever, if you have any further observations that you wish to make, please put them in writing and let the Committee have them; we shall welcome them. Thank you for your time this afternoon. We are very grateful to you.
Examination of Witnesses
Jared Sine, Nima Elmi and Dr Rachel O’Connell gave evidence.
Right. For once, we seem to have run out of questions. Minister, do you wish to contribute?
Everything I was going to ask has already been asked by my colleagues, so I will not duplicate that.
Q
Jared Sine: I would just make one brief comment. I think it has been mentioned by everyone here. Everyone has a role to play. Clearly, the Government have a role in proposing and pushing forward the legislation. The platforms that have the content have an obligation and a responsibility to try to make sure that their users are safe. One of the things that Dr O’Connell mentioned is age verification and trying to make sure that we keep young kids off platforms where they should not be.
I think there is a big role to play for the big tech platforms—the Apples and Googles—who distribute our apps. Over the years, we have said again and again to both of those companies, “We have age-gated our apps at 18, yet you will allow a user you know is 15, 14, 16—whatever it is—to download that app. That person has entered that information and yet you still allow that app to be downloaded.” We have begged and pleaded with them to stop and they will not stop. I am not sure that that can be included in the Bill, but if it could be, it would be powerful.
If Apple and Google could not distribute any of our apps—Hinge, Match, Tinder—to anyone under the age of 18, that solves it right there. It is the same methodology that has been used at clubs with bouncers—you have a bouncer at the door who makes sure you are 21 before you go in and have a drink. It should be the same thing with these technology platforms. If they are going to distribute and have these app stores, the store should then have rules that show age-gated apps—“This is for 17-plus or 18-plus”—and should also enforce that. It is very unfortunate that our calls on this front have gone unanswered. If the Bill could be modified to include that, it would really help to address the issue.
Dr Rachel O'Connell: Absolutely. I 100% support that. There is a tendency for people to say, “It is very complex. We need a huge amount of further consultation.” I started my PhD in 1996. This stuff has been going on for all that time. In 2008, there was a huge push by the Attorneys General, which I mentioned already, which brought all of the industry together. That was 2008. We are in 2022 now. 2017 was the Internet Safety Strategy Green Paper. We know what the risks are. They are known; we understand what they are. We understand the systems and processes that facilitate them. We understand what needs to be done to mitigate those risks and harms. Let’s keep on the track that we are going on.
Regarding industry’s concerns, a lot of them will be ironed out when companies are required to conduct risk assessments and impact assessments. They might ask, what are the age bands of your users? What are the risks associated with the product features that you are making available? What are the behaviour modification techniques that you are using, like endless scroll and loot boxes that get kids completely addicted? Are those appropriate for those ages? Then you surface the decision making within the business that results in harms and also the mitigations.
I urge you to keep going on this; do not be deterred from it. Keep the timeframe within which it comes into law fairly tight, because there are children out there who are suffering. As for the harassment—I have experienced it myself, it is horrible.
Those would be my final words.
Q
Rhiannon-Faye McDonald: It is incredibly important that we have this education piece. Like Susie said, we cannot rely on technology or any single part of this to solve child sexual abuse, and we cannot rely on the police to arrest their way out of the problem. Education really is the key. That is education in all areas—educating the child in an appropriate way and educating parents. We hold parenting workshops. Parents are terrified; they do not know what to do, what platforms are doing what, or what to do when things go wrong. They do not even know how to talk to children about the issue; it is embarrassing for them and they cannot bring it up. Educating parents is a huge thing. Companies have a big responsibility there. They should have key strategies in place on how they are going to improve education.
Q
I would like to pick up on a point that has arisen in the discussion so far—the point that Susie raised about the risks posed by Meta introducing end-to-end encryption, particularly on the Facebook Messenger service. You have referenced the fact that huge numbers of child sexual exploitation images are identified by scanning those communications, leading to the arrests of thousands of paedophiles each year. You also referenced the fact that when this was temporarily turned off in Europe owing to the privacy laws there—briefly, thankfully—there was a huge loss of information. We will come on to the Bill in a minute, but as technology stands now, if Meta did proceed with end-to-end encryption, would that scanning ability be lost?
Susie Hargreaves: Yes. It would not affect the Internet Watch Foundation, but it would affect the National Centre for Missing and Exploited Children. Facebook, as a US company, has a responsibility to do mandatory reporting to NCMEC, which will be brought in with the Bill in this country. Those millions of images would be lost, as of today, if they brought end-to-end encryption in now.
Q
Susie Hargreaves: Because they are scanning Facebook—sorry, I am just trying to unpack the way it works. It will affect us, actually. Basically, when we provide our hash list to Facebook, it uses that to scan Messenger, but the actual images that are found—the matches—are not reported to us; they are reported into NCMEC. Facebook does take our hash list. For those of you who do not know about hashing, it is a list of digital fingerprints—unique images of child sexual abuse. We currently have about 1.3 million unique images of child sexual abuse. Facebook does use our hash list, so yes it does affect us, because it would still take our hash list to use on other platforms, but it would not use it on Messenger. The actual matches would go into NCMEC. We do not know how many matches it gets against our hash list, because it goes into NCMEC.
Q
Susie Hargreaves: Yes, sorry—I was unclear about that. Yes, it would on Messenger.
Q
Susie Hargreaves: As I said before, it is essential that we do not demonise end-to-end encryption. It is really important. There are lots of reasons why, from a security and privacy point of view, people want to be able to use end-to-end encryption.
In terms of whether the technology is there, we all know that there are things on the horizon. As Ian said in the previous session, the technology is there and is about to be tried out. I cannot give any update at this meeting, but in terms of what we would do if end-to-end encryption is introduced and there is no ability to scan, we could look at on-device scanning, which I believe you mentioned before, Minister.
Yes.
Susie Hargreaves: That is an option. That could be a backstop position. I think that, at the moment, we should stand our ground on this and say, “No, we need to ensure that we have some form of scanning in place if end-to-end encryption is introduced.”
Q
Susie Hargreaves: I agree 100%.
Thank you very much indeed, Ms McDonald and Ms Hargreaves. We are most grateful to you; thank you for your help.
Examination of Witnesses
Ellen Judson and Kyle Taylor gave evidence.
Q
Ellen Judson: At the moment, no. The rights that are discussed in the Bill at the minute are quite limited: primarily, it is about freedom of expression and privacy, and the way that protections around privacy have been drafted is less strong than for those around freedom of expression. Picking up on the question about setting precedents, if we have a Bill that is likely to lead to more content moderation and things like age verification and user identity verification, and if we do not have strong protections for privacy and anonymity online, we are absolutely setting a bad precedent. We would want to see much more integration with existing human rights legislation in the Bill.
Kyle Taylor: All I would add is that if you look at the exception for content of democratic importance, and the idea of “active political issue”, right now, conversion therapy for trans people—that has been described by UN experts as torture—is an active political issue. Currently, the human rights of trans people are effectively set aside because we are actively debating their lives. That is another example of how minority and marginalised people can be negatively impacted by this Bill if it is not more human rights-centred.
Q
Ellen Judson: I accept that that is what the Bill currently says. Our point was thinking about how it will be implemented in practice. If platforms are expected to prove to a regulator that they are taking certain steps to protect content of democratic importance—in the explanatory notes, that is content related to Government policy and political parties—and they are expected to prove that they are taking a special consideration of journalistic content, the most straightforward way for them to do that will be in relation to journalists and politicians. Given that it is such a broad category and definition, that seems to be the most likely effect of the regime.
Kyle Taylor: It is potentially—
Q
Is it not true that a member of the public or anyone debating a legitimate political topic would also benefit from these measures? It is likely that MPs would automatically benefit—near automatically—but a member of the public might equally benefit if the topic they are talking about is of democratic or journalistic importance.
Ellen Judson: Our concern is that defining what is a legitimate political debate is itself already privileging. As you said, an MP is very likely automatically to benefit.
Well, it is likely; I would not say it is guaranteed.
Ellen Judson: A member of the public may be discussing something—for example, an active political debate that is not about the United Kingdom, which I believe would be out of scope of that protection. They would be engaged in political discussion and exercising freedom of expression, and if they were not doing so in a way that met the threshold for action based on harm, their speech should also come under those protections.
Kyle Taylor: I would add that the way in which you have described it would be so broad as to effectively be meaningless in the context of the Bill, and that instead we should be looking for universal free expression protections in that part of the Bill, and removing this provision. Because what is not, in a liberal democracy, speech of democratic importance? Really, that is everything. When does it reach the threshold where it is an active political debate? Is it when enough people speak about it or enough politicians bring it up? It is so subjective and so broad effectively to mean that everything could qualify. Again, this is not taking a harms-based approach to online safety, because the question is not “Who is saying it?” or “In what context?”; the question is, “Does this have the propensity to cause harm at scale?”
Q
Kyle Taylor: Can I respond to that?
Yes, sure.
Kyle Taylor: My point is that if there is a provision in the Bill about freedom of expression, it should be robust enough that this protection does not have to be in the Bill. To me, this is saying, “Actually, our free expression bit isn’t strong enough, so we’re going to reiterate it here in a very specific context, using very select language”. That may mean that platforms decide not to act for fear of reprisal, as opposed to pursuing online safety. I suggest strengthening the freedom of expression section so that it hits all the points that the Government intend to hit, and removing those qualifiers that create loopholes and uncertainty for a regime that, if it is systems-based, does not have loopholes.
Q
Ellen Judson: We absolutely recognise that. There is discussion in terms of meeting certain standards of responsible journalism in relation to those protections. Our concern is very much that the people and actors who would most benefit from the journalistic protections specifically would be people who do not meet those standards and cannot prove that they meet those standards, because the standards are very broad. If you intend your content to be journalistic, you are in scope, and that could apply to extremists as much as to people meeting standards of responsible journalism.
Q
Kyle Taylor: Remove the exemption.
Q
Kyle Taylor: Well, I am struggling to understand how we can look at the Bill and say, “If this entity says it, it is somehow less harmful than if this entity says it.” That is a two-tiered system and that will not lead to online safety, especially when those entities that are being given privilege are the most likely and largest sources and amplifiers of harmful content online. We sit on the frontlines of this every day, looking at social media, and we can point to countless examples from around the world that will show that, with these exemptions, exceptions and exclusions, you will actually empower those actors, because you explicitly say that they are special. You explicitly say that if they cause harm, it is somehow not as bad as if a normal user with six followers on Twitter causes harm. That is the inconsistency and incoherency in the Bill.
We are talking here about the press, not about politicians—
Kyle Taylor: Yes, but the press and media entities spread a lot of disinformation—
Q
Kyle Taylor: Except that that is inconsistent in the Bill, because you are saying that for broadcast, they must have a licence, but for print press, they do not have to subscribe to an independent standards authority or code. Even within the media, there is this inconsistency within the Bill.
That is a point that applies regardless of the Bill. The fact is that UK broadcast is regulated whereas UK newspapers are not regulated, and that has been the case for half a century. You can debate whether that is right or wrong, but—
Kyle Taylor: We are accepting that newspapers are not regulated then.
Q
Kyle Taylor: I am not suggesting that the freedom of the press is not sacrosanct. Actually, I am expressing the opposite, which is that I believe that it is so sacrosanct that it should be essential to the freedom-of-expression portion of the Bill, and that the press should be set to a standard that meets international human rights and journalistic standards. I want to be really clear that I absolutely believe in freedom of the press, and it is really important that we don’t leave here suggesting that we don’t think that the press should be free—
Q
Ellen Judson: To the media exemption—
To clause 50, “Recognised news publisher”.
Ellen Judson: One of the changes that the Government have indicated that they are minded to make—please correct me if I misunderstood—is to introduce a right to appeal.
Correct.
Ellen Judson: Content having to stay online while the appeal was taking place I would very much urge not to be introduced, on the grounds that the content staying online might then be found to be incredibly harmful, and by the time you have got through an appeals process, it will already have done the damage it was going to do. So, if there is a right to appeal—I would urge there not to be a particular right to appeal beyond what is already in the Bill, but if that is to be included, not having the restriction that the platforms must carry the content while the appeal process is ongoing would be important.
Kyle Taylor: You could require an independent standards code as a benchmark at least.
Order. I am afraid that brings us to the end of the time allotted for the Committee to ask questions. It also brings us to the end of the day’s sitting. On behalf of the Committee, I thank the witnesses for your evidence. As you ran out of time and the opportunity to frame answers, if you want to put them in writing and offer them to the Minister, I am sure they will be most welcome. The Committee will meet again on Thursday at 11.30 am in this room to hear further evidence on the Bill.
Ordered, That further consideration be now adjourned. —(Steve Double.)