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.
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.