You Spin Me Round: Three Features Built to Keep a Child in the Loop
For some background music when reading this post, we encourage you to (re)discover Dead or Alive performing “You Spin Me Round (Like a Record)”.
The opening post in this series, Non, a Ban Is Not a Plan, described how the French Conseil constitutionnel struck down the under-15 ban because the legislator had set out to address the risks posed by certain features of social media services and then wrote a rule about the user’s age instead. That post asked the question our Social Media Ban Repository focuses on the second half of this year: so, what is a plan? This post begins to answer it, three features at a time, and it starts where the harm is most familiar, with the design built to keep a child scrolling.
Three main features make up that design: the feed that never ends, the video that plays itself, and the notification built for return. They sit at the core of the sticky design (we will not use addictive, as we find the word too strong to cover all the possible reactions these features can potentially generate), each with a single job: to remove a natural stopping point and keep the child engaged. They are also where this series has to begin, because a ban leaves them untouched.
The good news for a drafter is that none of this needs inventing. EU law already names these three as harms, the European Commission is already enforcing against them, and at least two jurisdictions outside the European Union have already written the rule the EU has not. This post looks at what each feature does to a child, what European law already contains, what is still missing, and what the drafting should say. The through-line is the one the French decision supplied: regulate the feature, not the user.
The endless feed makes it difficult to stop
A feed with an endpoint is a feed a child finishes. Reaching the end is itself the cue to look up, the built-in moment where attention is handed back. The feed that never ends removes that moment by design. There is no last post, so there is no natural place to stop, and stopping becomes a deliberate decision the child has to make, faced with a system engineered to defer it indefinitely. For an adult, that is friction. And for some adults, that can be a lot of friction. For a child still building the capacity to regulate their own attention, it is a game arranged so the house wins by default.
EU law already treats the endless feed as a harm, not an innocuous feature. The Digital Services Act (DSA) requires platforms accessible to minors to secure a high level of safety by design, and the Commission’s July 2025 guidelines on the protection of minors name addictive patterns, including endless feeds, as risks a provider is expected to address. The European Commission has since moved from principle to enforcement. Its preliminary findings against TikTok of 6 February 2026 and against Meta of 10 July 2026 both identify infinite scroll as part of the addictive design under examination. These are proceedings in progress rather than final decisions, and should be read as such, but they indicate that the rules to handle these features seem to be in place and actionnable.
What EU law does not yet contain is a more general rule. The DSA reaches the largest platforms, works through risk assessment and guidance rather than a bright line, and asks a regulator to catch each feed after it has already shaped a generation of sessions. Brazil wrote that rule instead. Its ECA Digital statute, in force since March 2026, bans infinite scroll for minors outright, with penalties reaching up to ten per cent of a group’s revenue. New York did the same in a different register, namely its SAFE for Kids Act, whose final rules took effect on 28 July 2026 and restrict addictive feeds for minors by default. Meanwhile, the EU holds the diagnosis and the enforcement muscle but has not yet written the sentence that says the feed must end.
The fix is not to take the feed away. It is to give it a natural stopping point by default for every minor. Older teens, or parents, can switch the endless feed back on. The default protects the child who cannot yet win the game alone. The switch respects the older teenager who can, and the parent entitled to decide. This is the shape the French court decision seemed more willing to consider, a rule that distinguishes services by what they do and leaves the individual something to decide, and it is the shape the repository’s contributors have proposed, from Brazil’s legislators to 5Rights to the UN High Commissioner for Human Rights.
Autoplay chooses the next video for the child
Autoplay reverses what takes effort. Without it, watching another video is a choice the child makes each time. With it, the next video plays unless the child stops it, so stopping becomes the only decision left. The platform has quietly swapped which way the child has to act to keep watching, and doing nothing now means watching on. The decision is lifted out of the child’s hands and handed to the queue, again and again, for as long as the child fails to intervene.
On autoplay, EU law has gone furthest, as the Commission’s guidelines under the DSA, in their final July 2025 text, set autoplay off by default for minors. Beyond the text of the rules, the Meta proceedings name autoplay among the features under examination. So the standard already exists, albeit as guidance, attached to the obligations of the largest platforms, rather than as a binding default that reaches every service a child can open. Guidance can be read down, argued around, and confined to the biggest players. On its own, it misses the distinction the fix depends on, between a feed the platform queues to keep a child watching and a playlist the child chose.
Our proposed fix writes that distinction in. Autoplay off by default for minors, except where the child chose the queue, such as a playlist of their own. It is the cleanest of the three to draft and the hardest to defend against. A platform that cannot switch autoplay off for a fifteen-year-old is making a commercial argument, not a technical one. Brazil, 5Rights and the UN High Commissioner converge on it, and the European Commission has already written the default into its own guidance. The drafting now only has to make that default binding and universal, while leaving the child’s own choices in place.
The notification calls a child back after they leave
The feed and the autoplay work on a child who is already there. The notification is the only one of the three that reaches out of the device to fetch a child who has left. It manufactures the reason to come back, and it could pop up when the child is doing something else, in a lesson, at a meal, in bed. A buzz reporting that someone reacted, that someone is live, that the child has not visited in a while, is not information the child necessarily requested.
Again, fixes exist at various levels and in various geographies. Brazil prohibits excessive notifications to minors. New York’s SAFE for Kids Act restricts notifications to minors through the night by default. The Meta proceedings name push notifications as one of the addictive features under examination. And the UN High Commissioner for Human Rights names persistent notifications among the design choices that serve a commercial purpose rather than a child’s need.
EU law contains no binding quiet-hours default and no notification budget for minors, and it has not drawn the line the fix depends on, between a notification whose only function is to re-engage and a message a child genuinely needs, from an accepted contact or a real safety alert. Without that distinction, a crude rule fails in one of two directions, silencing the platform’s plea and the friend’s message together, or silencing neither.
The fix we suggest draws the line precisely. Quiet hours by default for all minors. Engagement notifications off by default; messages from accepted contacts and genuine safety alerts still get through. The line is the whole point. The child keeps the messages that matter and loses the noise that does not.
Off by default, with the switch left to the child or the parent, answers all three
Read together, the three fixes are the same drafting move. The harmful pattern is off by default for minors, and turning it back on is a choice left to the older child or the parent. This is what regulating the feature rather than the user looks like when it comes to legislative drafting. A ban asks who the user is and turns them away at the door; these fixes ask what the feature does and change it.
This more graduated approach is also what carries the fix past the objection the ban could not survive, since a default setting is not the same as a prohibition. It is a starting position the older teenager or the parent can move, a switch to operate rather than a birthday to wait for. And it asks nobody to prove their age, because a default that leans toward safety costs nothing when it guesses wrong.
Three features answered, and not one of them needed a ban
We spent the first half of this year asking whether a ban is a plan. The repository now sets out what a plan looks like, feature by feature, and these three are its engine room. The repository holds the other nine, and this series will continue to take them three at a time. The next post turns from the features that keep a child engaged to the features that turn that engagement into revenue: the streak and the reward loop, the profile built to sell, and the box you pay to open. The two posts after that take up the features that decide what a child is shown and who can reach them, and the newer arrivals a ban never contemplated at all.
The Dead or Alive song is a loop by design, the same hook coming round again with no resolution, the record spinning until someone lifts the needle. A product built to keep a child scrolling is the same machine. It does not push, it just never lets the groove end, and no amount of digital literacy will teach a child to out-sit a loop built to have no last turn. The answer is not to make the child better at stepping off. It is to build the stop back into the design and be willing to let the record end. A plan for the young generation should be made with them and not against them, and it should give the loop a bottom, so a child can step off when they choose.
Written by Caroline De Cock, LL.M., Head of Research
