23: Clause 11, page 10, line 9, at beginning insert “eliminate,”
Member’s explanatory statement
This amendment would require user to user services to eliminate identified risks to children from their platforms in addition to mitigating and managing them.
My Lords, this large group of 33 amendments is concerned with preventing harm to children, by creating a legal requirement to design the sites and services that children will access in a way that will put their safety first and foremost. I thank my co-sponsors, the noble Baronesses, Lady Kidron and Lady Harding, and the noble Lord, Lord Knight. First of all, I wish to do the most important thing I will do today: to wish the noble Baroness, Lady Kidron, a very happy birthday.
My co-sponsors will deal with some of the more detailed elements of the 30 amendments that we are dealing with. These will include safety duties, functionality and harm, and codes of practice. I am sure that the noble Lords, Lord Stevenson and Lord Knight, and the right reverend Prelate the Bishop of Oxford will speak to their own amendments.
I will provide a brief overview of why we are so convinced of the paramount need for a safety by design approach to protect children and remind digital companies and platforms, forcibly and legally, of their obligation to include the interests and safety of children as a paramount element within their business strategies and operating models. These sites and services are artificial environments. They were designed artificially and can be redesigned artificially.
In her testimony to the US Senate in July 2021, the Facebook whistleblower Frances Haugen put her finger on it rather uncomfortably when talking about her erstwhile employer:
“Facebook know that they are leading young users to anorexia content … Facebook’s internal research is aware that there are a variety of problems facing children on Instagram … they know that severe harm is happening to children”.
She was talking about, probably, three years ago.
On the first day of Committee, the noble Lord, Lord Allan, who is not with us today, used the analogy of the legally mandated and regulated safe design of aeroplanes and automobiles and the different regimes that cover their usage to illustrate some of our choices in dealing with regulation. We know why aeroplanes and cars have to be designed safely; we also know that either form of transportation could be used recklessly and dangerously, which is why we do not allow children to fly or drive them.
It is a great pleasure to follow my noble friend Lord Russell and to thank him for his good wishes. I assure the Committee that there is nowhere I would rather spend my birthday, in spite of some competitive offers. I remind noble Lords of my interests in the register, particularly as the chair of 5Rights Foundation.
As my noble friend has set out, these amendments fall in three places: the risk assessments, the safety duties and the codes of practice. However, together they work on the overarching theme of safety by design. I will restrict my detailed remarks to a number of amendments in the first two categories. This is perhaps a good moment to recall the initial work of Carnegie, which provided the conceptual approach of the Bill several years ago in arguing for a duty of care. The Bill has gone many rounds since then, but I think the principle remains that a regulated service should consider its impact on users before it causes them harm. Safety by design, to which all the amendments in this group refer, is an embodiment of a duty of care. In thinking about these amendments as a group, I remind the Committee that both the proportionality provisions and the fact that this is a systems and processes Bill means that no company can, should or will be penalised for a single piece of content, a single piece of design or, indeed, low-level infringements.
Amendments 24, 31, 77 and 84 would delete “content” from the Government’s description of what is harmful to children, meaning that the duty is to consider harm in the round rather than just harmful content. The definition of “content” is drawn broadly in Clause 207 as
“anything communicated by means of an internet service”,
but the examples in the Bill, including
“written material … music and data of any description”,
once again fail to include design features that are so often the key drivers of harm to children.
My Lords, it is an honour, once again, to follow the noble Baroness, Lady Kidron, and the noble Lord, Lord Russell, in this Committee. I am going to speak in detail to the amendments that seek to change the way the codes of practice are implemented. Before I do, however, I will very briefly add my voice to the general comments that the noble Baroness, Lady Kidron, and the noble Lord, Lord Russell, have just taken us through. Every parent in the country knows that both the benefit and the harm that online platforms can bring our children is not just about the content. It is about the functionality: the way these platforms work; the way they suck us in. They do give us joy but they also drive addiction. It is hugely important that this Bill reflects the functionality that online platforms bring, and not just content in the normal sense of the word “content”.
I will now speak in a bit more detail about the following amendments: Amendments 65, 65ZA, 65AA, 89, 90, 90B, 96A, 106A, 106B, 107A, 114A—I will finish soon, I promise—112, 122ZA, 122ZB and 122ZC.
That list shows your Lordships some of the challenges we all have with the Bill. All these amendments seek to ensure that the codes of practice relating to child safety are binding. Such codes should be principles-based and flexible to allow companies to take the most appropriate route of compliance, but implementing these codes should be mandatory, rather than, as the Bill currently sets out, platforms being allowed to use “alternative measures”. That is what all these amendments do—they do exactly the same thing. That was a clear and firm recommendation from the joint scrutiny committee. The government’s response to that joint scrutiny committee report was really quite weak. Rather than rehearse the joint scrutiny committee’s views, I will rehearse the Government’s response and why it is not good enough to keep the Bill as it stands.
The first argument the Government make in their response to the joint scrutiny report is that there is no precedent for mandatory codes of conduct. But actually there are. There is clear precedent in child protection. In the physical world, the SEND code for how we protect some of our most vulnerable children is mandatory. Likewise, in the digital world, the age-appropriate design code, which we have mentioned many a time, is also mandatory. So there is plenty of precedent.
The second concern—this is quite funny—was that stakeholders were concerned about having multiple codes of conduct because it could be quite burdensome on them. Well, forgive me for not crying too much for these enormous tech companies relative to protecting our children. The burden I am worried about is the one on Ofcom. This is an enormous Bill, which places huge amounts of work on a regulator that already has a very wide scope. If you make codes of conduct non-mandatory, you are in fact making the work of the regulator even harder. The Government themselves in their response say that Ofcom has to determine what the minimum standards should be in these non-binding codes of practice. Surely it is much simpler and more straightforward to make these codes mandatory and, yes, to add potentially a small additional burden to these enormous tech companies to ensure that we protect our children.
My Lords, I was not going to speak on this group, but I was provoked into offering some reflections on the speech by the noble Lord, Lord Russell of Liverpool, especially his opening remarks about cars and planes, which he said were designed to be safe. He did not mention trains, about which I know something as well, and which are also designed to be safe. These are a few initial reflective points. They are designed in very different ways. An aeroplane is designed never to fail; a train is designed so that if it fails, it will come to a stop. They are two totally different approaches to safety. Simply saying that something must be designed to be safe does not answer questions; it opens questions about what we actually mean by that. The noble Lord went on to say that we do not allow children to drive cars and fly planes. That is absolutely true, but the thrust of his amendment is that we should design the internet so that it can be driven by children and used by children— so that it is designed for them, not for adults. That is my problem with the general thrust of many of these amendments.
A further reflection that came to mind as the noble Lord spoke was on a book of great interest that I recommend to noble Lords. It is a book by the name of Risk written in 1995 by Professor John Adams, then professor of geography at University College London. He is still an emeritus professor of geography there. It was a most interesting work on risk. First, it reflected how little we actually know of many of the things of which we are trying to assess risk.
More importantly, he went on to say that people have an appetite for risk. That appetite for risk—that risk budget, so to speak—changes over the course of one’s life: one has much less appetite for risk when one gets to a certain age than perhaps one had when one was young. I have never bungee jumped in my life, and I think I can assure noble Lords that the time has come when I can say I never shall, but there might have been a time when I was younger when I might have flung myself off a cliff, attached to a rubber band and so forth—noble Lords may have done so. One has an appetite for risk.
My Lords, I respond to the noble Lord in two ways. First, I ask him to reflect on how the parents of the children who have died through what the parents would undoubtedly view as serious and unbearable harm would feel about his philosophical ruminations. Secondly, as somebody who has the privilege of being a Deputy Speaker in your Lordships’ House, it is incumbent and germane for us all to focus on the amendment in question and stay on it, to save time and get through the business.
Well, I must regard myself as doubly rebuked, and unfairly, because my reflections are very relevant to the amendments, and I have developed them in that direction. In respect of the parents, they have suffered very cruelly and wrongly, but although it may sound harsh, as I have said in this House before on other matters, hard cases make bad law. We are in the business of trying to make good law that applies to the whole population, so I do not think that these are wholly—
If my noble friend could, would he roll back the health and safety regulations for selling toys, in the same way that he seems so happy to have no health and safety regulations for children’s access to digital toys?
20 of 134 shown
First, let us listen to the designers of these platforms and services through some research done by the 5Rights Foundation in July 2021. These are three direct quotes from the designers:
“Companies make their money from attention. Reducing attention will reduce revenue. If you are a designer working in an attention business, you will design for attention … Senior stakeholders like simple KPIs. Not complex arguments about user needs and human values … If a senior person gives a directive, say increase reach, then that’s what designers design for without necessarily thinking about the consequences”.
Companies know exactly what they need to do to grow and to drive profitability. However, they mostly choose not to consider, mitigate and prioritise to avoid some of the potentially harmful consequences. What they design and prioritise are strategies to maximise consumption, activity and profitability. They are very good at it.
Let us hear what the children say, remembering that some recent research indicates that 42% of five to 12 year-olds in this country use social media. The Pathways research project I referred to earlier worked closely with 21 children aged 12 to 18, who said: “We spend more time online than we feel we should, but it’s tough to stop or cut down”. “If we’re not on social media, we feel excluded”. “We like and value the affirmations and validations we receive”. “We create lots of visual content, much of it about ourselves, and we share it widely”. “Many of us are contacted by unknown adults”. “Many of us recognise that, through using social media, we have experienced body image and relationships problems”.
To test whether the children in this research project were accurately reporting their experiences, the project decided to place a series of child avatars—ghost children, in effect—on the internet, whose profiles very clearly stated that they were children. It did this to test whether these experiences were true.
They found—in many cases within a matter of hours of the profiles going online—proactive contacting by strangers and rapid recommendations to engage more and more. If searches were conducted for eating disorders or self-harm, the avatars were quickly able to access content irrespective of their stated ages and clearly evident status as children. At the same time they were being sent harmful or inappropriate content, they also received age-relevant advertising for school revision and for toys—the social media companies knew that these accounts were registered as children.
This research was done two years ago. Has anything improved since then? It just so happens that 5Rights has produced another piece of research which is about to be released, and which used the exact same technique—creating avatars to see what they would experience online. They used 10 avatars based on real children aged between 10 and 16, so what happened? For an 11 year-old avatar, Instagram was recommending images of knives with the caption “This is what I use to self-harm”; design features were leading children from innocent searches to harmful content very quickly.
I think any grandparents in the Chamber will be aware of an interesting substance known as “Slime”—a form of particularly tactile playdough which one’s grandchildren seem to enjoy. Typing in “Slime” on Reddit was one search, and one click, away from pornography; exactly the same thing happened on Reddit when the avatar typed in “Minecraft”, another very popular game with our children or grandchildren. A 15 year-old female avatar was private-messaged on Instagram by a user that she did not follow—an unknown adult who encouraged her to link on to pornographic content on Telegram, another instant messaging service. On the basis of this evidence, it appears that little or nothing has changed; it may have even got slightly worse.
By an uncomfortable coincidence, last week, Meta, the parent company of Facebook and Instagram, published better than expected results and saw its market value increase by more than $50 billion in after-hours trading. Mark Zuckerberg, the founder of Meta, proudly announced that Meta is pouring investment into artificial intelligence tools to make its platform more engaging and its advertising more effective. Of particular interest and concern given the evidence of the avatars was his announcement that since the introduction of Reels, a short-term video feed designed specifically to respond to competition from TikTok, its AI-driven recommendations had boosted the average time people spend on Instagram by 24%.
To return to the analogy of planes and cars used by the noble Lord, Lord Allan, we are dealing here with planes and cars in the shape of platforms and applications which we know are flawed in their design. They are not adequately designed for safety, and we know that they can put users, particularly children and young people, in the way of great harm, as many grieving families can testify.
In conclusion, our amendments propose that companies must design digital services that cater for the vulnerabilities, needs, and rights of children and young people by default; children’s safety cannot and must not be an afterthought or a casualty of their business models. We are asking for safety by design to protect children to become the mandatory standard. What we have today is unsafe design by default, driven by commercial strategies which can lead to children becoming collateral damage.
Given that it is the noble Baroness’s birthday, I am sure we can feel confident that the Minister will have a positive tone when he replies. I beg to move.
On day three of Committee, the Minister said:
“The Bill will address cumulative risk where it is the result of a combination of high-risk functionality, such as live streaming, or rewards in service … This will initially be identified through Ofcom’s sector risk assessments, and Ofcom’s risk profiles and risk assessment guidance will reflect where a combination of risk in functionalities such as these can drive up the risk of harm to children. Service providers will have to take Ofcom’s risk profiles into account in their own risk assessments for content which is illegal or harmful to children”.—[Official Report, 27/4/23; col. 1385.]
However, in looking at the child safety duties, Clause 11(5) says:
“The duties … in subsections (2) and (3) apply across all areas of a service, including the way it is designed, operated and used”,
but subsection (14) says:
“The duties set out in subsections (3) and (6)”—
which are the duties to operate proportionate systems and processes to prevent and protect children from encountering harmful content and to include them in terms of service—
“are to be taken to extend only to content that is harmful to children where the risk of harm is presented by the nature of the content (rather than the fact of its dissemination)”.
I hesitate to say whether that is contradictory. I am not actually sure, but it is confusing. I am concerned that while we are reassured that “content” means content and activity and that the risk assessment considers functionality, “harm” is then repeatedly expressed only in the form of content.
Over the weekend, I had an email exchange with the renowned psychoanalyst and author, Norman Doidge, whose work on the plasticity of the brain profoundly changed how we think about addiction and compulsion. In the exchange, he said that
“children’s exposures to super doses, of supernormal images and scenes, leaves an imprint that can hijack development”.
Then, he said that
“the direction seems to be that AI would be working out the irresistible image or scenario, and target people with these images, as they target advertising”.
His argument is that it is not just the image but the dissemination and tailoring of that image that maximises the impact. The volume and frequency of those images create habits in children that take a lifetime to change—if they change at all. Amendments 32 and 85 would remove this language to ensure that content that is harmful by virtue of its dissemination is accounted for.
I turn now to Amendments 28 and 82, which cut the reference to the
“size and capacity of the provider of the service”
in deeming what measures are proportionate. We have already discussed that small is not safe. Such platforms such as Yubo, Clapper and Discord have all been found to harm children and, as both the noble Baroness, Lady Harding, and the noble Lord, Lord Clement-Jones, told us, small can become big very quickly. It is far easier to build to a set of rules than it is to retrofit them after the event. Again, I point out that Ofcom already has duties of proportionality; adding size and capacity is unnecessary and may tip the scale to creating loopholes for smaller services.
Amendment 138 seeks to reverse the exemption in Clause 54 of financial harms. More than half of the 100 top-grossing mobile phone apps contain loot boxes, which are well established as unfair and unhealthy, priming young children to gamble and leading to immediate hardship for parents landed with extraordinary bills.
By rights, Amendments 291 and 292 could fit in the future-proof set of amendments. The way that the Bill in Clause 204 separates out functionalities in terms of search and user-to-user is in direct opposition to the direction of travel in the tech sector. TikTok does shopping, Instagram does video, Amazon does search; autocomplete is an issue across the full gamut of services, and so on and so forth. This amendment simply combines the list of functionalities that must be risk-assessed and makes them apply on any regulated service. I cannot see a single argument against this amendment: it cannot be the Government’s intention that a child can be protected, on search services such as Google, from predictive search or autocomplete, but not on TikTok.
Finally, Amendment 295 will embed the understanding that most harm is cumulative. If the Bereaved Parents for Online Safety were in the Chamber, or any child caught up in self-harm, depression sites, gambling, gaming, bullying, fear of exposure, or the inexorable feeling of losing their childhood to an endless scroll, they would say at the top of their voices that it is not any individual piece of content, or any one moment or incident, but the way in which they are nudged, pushed, enticed and goaded into a toxic, harmful or dangerous place. Adding the simple words
“the volume of the content and the frequency with which the content is accessed”
to the interpretation of what can constitute harm in Clause 205 is one of the most important things that we can do in this Chamber. This Bill comes too late for a whole generation of parents and children but, if these safety by design amendments can protect the next generation of children, I will certainly be very glad.
The third challenge is that non-statutory guidance already looks as if it is causing problems in this space. On the video-sharing platform regime, which is non-mandatory, Ofcom has already said that in its first year of operation it has
“seen a large variation in platforms’ readiness to engage with Ofcom”.
All that will simply make it harder and harder, so the burden will lie on this regulator—which I think all of us in this House are already worried is being asked to do an awful lot—if we do not make it very clear what is mandatory and what is not. The Secretary of State said of the Bill that she is
“determined to put these vital protections for … children … into law as quickly as possible”.
A law that puts in place a non-mandatory code of conduct is not what parents across the country would expect from that statement from the Secretary of State. People out there—parents and grandparents across the land—would expect Ofcom to be setting some rules and companies to be required to follow them. That is exactly what we do in the physical world, and I do not understand why we would not want to do it in the digital world.
Finally—I apologise for having gone on for quite a long time—I will very briefly talk specifically to Amendment 32A, in the name of the noble Lord, Lord Knight, which is also in this group. It is a probing amendment which looks at how the Bill will address and require Ofcom and participants to take due regard of VPNs: the ability for our savvy children—I am the mother of two teenage girls—to get round all this by using a VPN to access the content they want. This is an important amendment and I am keen to hear what my noble friend Minister will say in response. Last week, I spoke about my attempts to find out how easy it would be for my 17 year-old daughter to access pornography on her iPhone. I spoke about how I searched in the App Store on her phone and found that immediately a whole series of 17-plus-rated apps came up that were pornography sites. What I did not mention then is that with that—in fact, at the top of the list—came a whole series of VPN apps. Just in case my daughter was naive enough to think that she could just click through and watch it, and Apple was right that 17 year-olds were allowed to watch pornography, which obviously they are not, the App Store was also offering her an easy route to access it through a VPN. That is not about content but functionality, and we need to properly understand why this bundle of amendments is so important.
The interesting thing that he went on to develop from that was the notion of risk compensation: that if you have an appetite for risk and your opportunities to take risks are taken away, all you do is compensate by taking risks elsewhere. So a country such as New Zealand, which has some of the strictest cycling safety laws, also has a very high incidence of bungee jumping among the young; as they cannot take risks on their bicycles, they will find ways to go and do it elsewhere.
Although these reflections are not directly germane to the amendments, they are important as we try to understand what we are seeking to achieve here, which is a sort of hermetically sealed absence of risk for children. I do not think it will work. I said at Second Reading that I thought the flavour of the debate was somewhat similar to a late medieval conclave of clerics trying to work out how to mitigate the harmful effects of the invention of movable type. That did not work either, and I think we are in a very similar position today as we discuss this.
There is also the question of harm and what it means. While the examples being given by noble Lords are very specific and no doubt genuinely harmful, and are the sorts of things that we should like to stop, the drafting of the amendments, using very vague words such as “harm”, is dangerous overreach in the Bill. To give just one example, for the sake of speed, when I was young, administering the cane periodically was thought good for a child in certain circumstances. The mantra was, “Spare the rod and spoil the child”, though I never heard it said. Nowadays, we would not think it morally or psychologically good to do physical harm to a child. We would regard it as an unmitigated harm and, although not necessarily banned or illegal, it is something that—