Hello!
Yet again, it’s been months and months since I last wrote a newsletter. My genius-level content strategy is to just keep surprising everyone by sending it at random intervals with no discernible connecting theme. 

Anyway, I’ve decided to break this particular dry spell with some thoughts on why the UK government keeps making poor choices about online safety policy and outline some ways it could be more effective. I mean, I realise that’s a bit of a swing, but if you’re only going to send a newsletter 4 times a year and pay $29 a month for the privilege,* it may as well be ambitious. 

(*I may need to move to a different newsletter platform.)

Is the social media ban good policy or bad policy? 

In case you missed it, the scope of the under-16 social media ban has been expanded in the pre-Christmas legislative rush and it now includes any popular digital service that has a “content recommender system” or includes private messaging. (See the Draft Statutory Instrument for some updated definitions.) The stated aim of this expansion is to stop children engaging in dastardly and nefarious exploits such as commenting on unpopular NPR podcasts, but the actual impacts are more likely to include: 

• kids turning up on the most godforsaken bits of the Internet, which will almost certainly be less safe than scrolling through 37 different Minecraft or GRWM influencers on YouTube Shorts or TikTok;

• kids using their parents’ or carers’ log-ins, which is also likely to be less safe, as they’ll be exposed to adults’ content recommendations;

• VPN sales going up, which means children will be pretending to be adults;

• increased use of free AI bots, because that seems fine; and

• someone (or perhaps, many people) vibe coding alternative message platforms and running them in secret. 

Also, depending on how broadly the notion of a “content recommender system” is interpreted, it doesn’t seem impossible this could include access to any algorithmically driven search engine.

So, overall, this feels like a disaster: it removes access to most of the busiest and most useful parts of the Internet while creating new risks and giving children and young people nothing in return. The ban isn’t going to be sweetened by increased investment in local libraries or new youth clubs or funding for great days out; it’s just taking stuff away from people who haven’t done anything wrong while allowing tech companies to carry on regardless. So it seems like a flagship example of extremely poor policymaking – and the puzzle is, why have successive UK governments been so bad at online safety policy? 

The problem is two-fold. The first aspect is that online safety has become such an emotive and high-stakes issue that policymakers cannot convene useful conversations about it. The lack of useful conversations then makes it harder to identify the real causes of the problem and turns online safety into a dynamic and exciting political issue that demands big, dramatic solutions. These dramatic solutions then feed the hyperbole and make it more difficult to propose specific and more precise interventions that would make a long-term meaningful difference.  

The second is that, politically, we are now in a place where it’s easier for government to intervene directly in the lives of millions of children than it is to make effective product safety legislation that would affect a handful of companies. That seems problematic in a functioning democracy as it means public resources are used in the service of political theatre rather than policy impact. 

Why can’t we have good conversations about children’s safety online? 

I’ve worked in online safety on and off since 2000, initially on the product side and latterly in regulation and research. Of all the tech policy areas I’ve worked in, children’s online safety is notable for being particularly difficult to talk about constructively.

For a start, there are many experts. We’ve all been children, many people are parents, nearly everyone has a problematic relationship with their phone that they wish they’d never started, and we’ve all heard terrible, terrible stories. 

Within tech policy, it is also a topic that gives rise to some deeply held, binary opinions on privacy, surveillance, and freedom of expression. This, unhelpfully, turns many conversations about child safety into ruminations on adults’ right to privacy and free speech. 

There is also little agreement on how safe children need to be and what being “safe” means across the different ages and stages children go through. We all know what it is to be unsafe – to be bullied, groomed, abused, or perniciously influenced – but most people can’t describe what it means to be “safe” online. 

Keeping children safe is also incredibly emotive, which can make it difficult to discuss in practical, logical terms – especially when terrible stories are circulating or bereaved parents have been thrust into the spotlight.

All of this obtuse and imprecise discussion also makes children’s online safety particularly susceptible to terrible, magical snake oil tech solutions that guarantee their advocates a spot on the media round. 

So, politically, it is a nightmare. Everyone cares about it, no one knows what to do about it, and the Online Safety Act has been a disappointing juggernaut. So the problems persist and the conversations become more heightened.

What stops good policy?

Hyperbole and imprecision. 

Unsurprisingly, bad conversations about online safety are not improved by media coverage that lingers over tragic outcomes (which are, thankfully, the exception not the rule) and leads with scare stories that make it sound as if every picture on the Internet some kind of torture porn. This adds to framing online safety as an ineluctable, snowballing crisis that demands theatrical measures and big announcements. 

However, while theatrical measures may feel like the right response to hyperbole, they are often unsuitable for tackling the substance of a complex problem. For instance, the social media ban will only reduce or displace the risks social media poses. The ban seeks to completely eradicate these risks, but control only extends so far: a child can always get around it by using another device or another person’s account, using a less popular service, or turning to an AI bot rather than a social-media platform to pass the time. Moreover, a child who is unfamiliar with social media or private messaging might be placed at more risk because they have reduced digital literacy, particularly if they are using an adult log-in and shown content suitable for people in a much older age group. The potential impact of these new risks could undermine the ban completely – particularly if the furthest edges of the ban also do things like stop children from listening to music, playing games online, or finding useful information. 

The very broadbrush nature of big-picture, theatrical policymaking is also unhelpful. For instance, the social media ban appears to treat “children” as a single, undifferentiated mass. However, a specific measure that tries to manage how a 15 year old consumes information, conducts relationships, and negotiates risk should have very little in common with a policy that puts guidelines around how a toddler engages with streaming video. 

It is also the case that different children grow and mature in different ways and at different rates, and that their development is influenced by many different factors. The social media ban will have a completely different impact on a Year 6 child whose parents don’t want to buy them a smartphone when they go to secondary school, a Year 10 teen who uses YouTube or TikTok to support their homework or learn a new skill, and a young carer who takes on adult responsibilities in the family home and needs to work out how to call the electricity company.  

The ban also manages to completely overlook the main source of risk reduction in many young people's lives: their parents and carers. In the offline world, while specific legal protections exist that reduce the possibility of physical harm to children (such as appropriate car seats in cars) and there are professionals and institutions that deliver services such as education and healthcare, many aspects of children’s lives are also shaped by the individualised and relational mixture of love and support we know as parenting and care-giving. As an example, although everyone knows children need lots of sleep, there is not a legally enforced National Bedtime Policy in which every 5 year old is mandated to get themselves to bed at 7; instead there is a mix of best practice advice for parents and carers that helps them do whatever works in their context. 

The ban also changes the location of legal safety interventions: it would be more usual and easier for government to enforce product safety regulations on a small number of providers to make the products safer. Instead, the ban allows the products to remain dangerous for everyone while expecting millions of users to change their behaviour without any supporting social infrastructure. 

What good policy looks like 

Because so many aspects of risk and safety are contextual, good safety policies are often specific, measurable, and targeted at reducing named and known risks.  For instance, even without age verification, platforms such as YouTube, Meta, and TikTok know enough about their users to work out whether an account is more or less likely to belong to a 14 year old or to a 24 year old - and government could mandate that users inferred to be children are shown age-appropriate content that is optimised to be less compulsively scrollable. A single intervention of that kind wouldn’t solve every problem, but several precise interventions that address known risks would create baseline safer services. 

As laws and regulations cannot be designed to meet the myriad individual needs outlined above, a focus on high-quality outcomes, rather than magic-wand solutions, would also  make it a lot easier to meet children’s individual safety needs. The Prime Minister has announced a “root and branch review of childhood: some part of that review needs to address the modern challenges of growing up online and becoming an adult in a  digital society.

Childhood is a also process of learning; being a young teenager, particularly, is all about negotiating risks and making good decisions. The expanded scope of the ban effectively puts a “closed” sign across most digital products and services and then expects young people to magically work it out when they’re 16. In addition to making digital products safer, government needs to build and invest in the institutions and norms that make good digital health and wellbeing possible for children and young people. If social media use is a public-health problem, we should use the established norms of engagement and support people to achieve better outcomes. 

And finally, rather than just taking things away, there needs to be investment in the kinds of social infrastructure that has been erased over time by the mix of austerity, Covid lockdowns, and increased digitisation. Young people have nowhere to go – stopping them from using Vinted won’t improve that. 

Conclusion

The social media ban may well deliver some short-term improvements, but it won’t work in the long-term or lead to systemic change because it’s leaning on the wrong people. A government that thinks it’s easier to make millions of children change their behaviour overnight than it is to mandate a handful of companies make their products safer is engaging in performative policymaking - chasing headlines rather than creating societal change. Good online safety policy is possible, but it requires the confidence and expertise to be specific, precise, and relentless. 

Thanks so much to Georgia Iavoucu and Ben Whitelaw for inviting me to speak at Marked As Urgent on the social media ban last week, which gave me the opportunity to start writing about this again. See also Georgia’s great post on the social impacts of the social media ban and this really important research from Ellie Harris at IPPR on young people and loneliness.