There’s a gesture all of us perform, hundreds of times a day, without thinking. It might be the most repeated gesture of our era—more than walking, more than drinking water “mindfully”. The thumb scrolling up on a screen. One scroll. Another. Another.
On 6 February 2026, the European Commission did something that, the way I read it, marks a before and after. It didn’t say “this content is forbidden”. It didn’t ask anyone to take down a hashtag or remove a video. It put an interaction pattern in question. A design choice.
According to preliminary findings under the Digital Services Act, TikTok’s infinite scroll pushes users into “autopilot mode”, feeding compulsive behaviour. So, Brussels says, the platform has to change the basic design of the service: disable infinite scroll, introduce mandatory pauses, rethink the recommendation system. If it doesn’t, the fine can reach 6% of global annual revenue. For ByteDance, that’s tens of billions—the kind of number that makes wrists tremble.
Two weeks later, on the other side of the ocean, Mark Zuckerberg is sitting in a Los Angeles courtroom. It’s the first jury trial on social media addiction in U.S. history. A young woman, identified as K.G.M., describes being pulled into Instagram as a child and developing depression and suicidal ideation. In the background sit more than 1,600 similar lawsuits.
The lawyers display Meta’s internal emails. One sentence sticks: “IG is a drug. We are pushing users.” Another email suggests the company knew that more than 30% of children aged 10 to 12 used Instagram, breaking its own policies. Zuckerberg replies that the goal is to make the service “useful”, not to maximise time on platform. Meanwhile the judge warns the audience not to record proceedings with AI glasses—some members of Zuckerberg’s entourage, apparently, had shown up wearing Ray-Ban Metas.
Two continents. Two approaches. The same problem.
Design is politics
It’s worth pausing on what happened on 6 February, because it’s more radical than it looks.
Europe didn’t regulate content. It said that the form itself of the interaction between a person and an app can constitute a systemic risk. That a thumb movement, if designed never to end, isn’t “your weakness”. It’s the platform’s problem.
This is, in fact, the first time a regulator has tried to set a legal standard on the addictive potential of design. It isn’t a law “against infinite scroll” in the abstract. It’s a larger principle: addictive design is a risk, and infinite scroll is one of its most visible manifestations.
The interesting part—and the slightly unsettling one—is that the principle doesn’t stop at TikTok. Meta has been under investigation since May 2024 for similar reasons. The famous “rabbit hole”, the tunnel you enter watching one piece of content while the algorithm pushes you toward more and more similar—and sometimes more extreme—content, is in the crosshairs. X has already taken a €120 million fine for “deceptive design practices”.
The EU’s approach looks more and more like this: not just “what you can’t show”, but “how you can’t make people feel”.
For people who build software, the ground really does shift here. Code has never been neutral, we know that, but now it’s officially also a question of legal liability. The Product Liability Directive coming into force on 9 December 2026 includes software among products. The Cyber Resilience Act demands security by design. The AI Act imposes risk assessments. And now the DSA suggests that even the way you organise a feed can become a violation.
A simple image comes to mind: the door handle that won’t let you out. It isn’t a problem with the person trying to leave. It’s a problem with the handle.
The quarterly vs. the neuron
The question, though, remains. Why did we get this far? Why do we need courts and commissions to say something anyone with a child—or just a thumb—already senses?
One answer, maybe, is in the internal documents emerging from the U.S. trials. It doesn’t look like negligence. It looks like a deliberate choice, repeated, driven by a metric: engagement. And engagement, in the end, is the antechamber of the quarterly.
Those documents show that increasing time spent on platforms and growing among teens wasn’t a side effect. It was a goal. Emails from 2015 and 2017 discuss how to prioritise growth among adolescents. Internal studies record some teenagers describing Instagram with words very close to behavioural addiction.
And still no one stops. The measuring continues.
Adam Mosseri, head of Instagram, testified that in his view it’s not “clinical addiction” but “problematic use”. Maybe the distinction matters in a courtroom. In real life, I wonder how much it changes. The point, in any case, is that they knew.
They knew that certain face filters simulating plastic surgery could affect adolescent girls, and after internal debate they chose a “more targeted ban” rather than a complete one. They knew the parental controls were easy to bypass, and they presented them as a solution. They knew children under 13 were on the platform by the millions, and they kept counting them as users.
Here an economic concept comes in that I find useful, even if it’s a little frightening how well it fits. It’s called externality: a cost the producer offloads onto society without paying for it. Pollution is the classic example: the factory produces, the river pays.
With social media we’ve had a new form of externality for twenty years. We could call it, without exaggeration, the attention externality: the cost society pays when a company extracts attention the way a mine extracts coal, without compensating for the damage. Lungs become neurons. Air becomes time. The river becomes someone’s childhood.
And the difference from industrial pollution, perhaps, is this: here the harm isn’t only a side effect. In many cases it was designed. It isn’t the production’s waste. It’s the product.
Autopilot and dignity
In the European findings there’s an expression that stuck with me: “autopilot mode”. Certain features, the Commission writes, shift users’ brains into autopilot, feeding the impulse to keep scrolling. Scientific research, they add, shows that these mechanisms can reduce self-control and contribute to compulsive behaviour.
Autopilot is a strange word, because in other contexts it’s almost positive. In aviation it makes sense: it frees cognitive resources for more important decisions. In a social media app, on the other hand, it removes the most important cognitive resource we have at that moment: the ability to choose when to stop.
Byung-Chul Han described contemporary malaise well by talking about the achievement society and self-exploitation. The idea, simplified, is that today no external master is needed. We squeeze ourselves: we optimise, perform, consume, until burnout. And the smartphone is the perfect instrument for it.
But there’s another step the TikTok-DSA case makes more visible. Self-exploitation isn’t always self-generated. Sometimes it’s designed by someone else and then sold as freedom.
Infinite scroll isn’t the user’s decision. It’s a decision by a product team, validated by an A/B test, approved by a VP of Growth, celebrated in a quarterly business review. The user isn’t self-exploiting. They’re being exploited inside an architecture that simulates choice.
And here, if you want, you could even bring in Kant without going full salon-philosophy. Treat people as ends, never only as means. The user-as-KPI is the user-as-means promoted to a business model. You aren’t a person using a service. You’re a unit of attention to extract, an eye to monetise, a thumb to keep moving.
Maybe that’s why what Europe is doing feels, underneath, like a defence of dignity. Not just consumer protection. The idea that we have a right not to be manipulated, even when the manipulation is elegant, well-designed, and packaged in a colourful interface with rounded corners.
Two models of the world
The contrast between Brussels and Los Angeles tells two different philosophies.
In the United States the system is adversarial. A family takes a company to court, presents documents, witnesses, expert reports. A jury decides whether Instagram was a “substantial factor” in the harm of one specific person. It’s a case-by-case model, where the burden of proof falls mostly on whoever was harmed.
And in that context everything becomes ambiguous. YouTube, in the same lawsuit, tries to defend itself by saying it’s more like Netflix than a social platform. Meta says the problem isn’t the app, it’s the content—or maybe it’s the girl’s personal life. The lines between “platform harm” and “content harm” are hard to draw. People argue over the definition of “addiction”. They postpone. They negotiate.
In Europe the approach is more structural. You don’t wait for the harm, you regulate the design. You don’t ask the individual to prove that infinite scroll ruined their life. You ask the platform to prove its design isn’t intrinsically risky. The DSA treats addictive design as a systemic risk on a par with disinformation or election interference.
It’s an asymmetry that reflects two ideas of freedom.
In the United States, often, freedom is absence of constraint: you’re free to use TikTok or not, and if it hurts you that’s your problem—or your lawyer’s.
In Europe, at least in this framing, freedom includes protection from manipulation: you’re free only if the conditions you operate in aren’t designed to strip you of the ability to choose.
Neither model is perfect. But one of them, today, is trying to change the game before millions of people have to walk into a courtroom and recount their childhood in front of a jury.
The question for builders
I build software. Every day decisions get made that look small and aren’t. Where to put a button. How to structure a notification. How easy or hard to make leaving a flow. Whether to show a counter, a badge, a red dot.
I don’t build social media. I work on platforms for public administration, e-learning, B2B services, legal tools. My world is a thousand kilometres from infinite scroll. And yet the principle Europe is trying to fix sounds universal: design isn’t neutral. Every interface choice is also an ethical choice. And now, like it or not, also a choice with legal consequences.
For years the social platforms enjoyed a kind of de facto impunity, because regulation couldn’t keep pace. That time looks over. The DSA, the Product Liability Directive, the Cyber Resilience Act, the AI Act are pieces of a mosaic redrawing the relationship between those who produce technology and those who use it.
Those who endure it, you might say, after reading certain internal emails.
In the end, a conviction stays with me, and it probably gets stronger every time I see a product “grow” thanks to a cognitive trick: software that respects the people using it is better software. Not because Europe imposes it on us, but because it’s the right thing to do.
And maybe the most radical thing today is to do something radically obvious. Build technology that doesn’t need a court to order it to stop doing harm.
Infinite scroll will end, in one form or another. The real question is what we put in its place. If it’s just another trick to keep you glued, with a different name and a slightly subtler mechanism, we won’t have solved anything.
If instead it’s a design that starts from the question “what does this person need?” rather than “how do I keep them here one more minute?”, then maybe 6 February 2026 will be a date worth remembering.
Not for the scroll it called into question, but for the question it made unavoidable.
Key takeaways
Design is not neutral: every interface choice is a political choice—and now also a legal one.
The attention externality is the new pollution, except it isn’t a side effect, it’s the product.
Two models of freedom: absence of constraint vs. protection from designed manipulation.
Questions & answers
What is the European Commission charging TikTok with under the Digital Services Act?
Infinite scroll and design mechanics that produce cognitive autopilot—compulsive use the user did not consciously choose. Under the DSA, very large online platforms have to assess and mitigate the systemic risks of their design, including effects on mental wellbeing. TikTok sits at the centre of enforcement because its algorithm is the most effective at inducing involuntary attention states.
Why is design becoming a matter for regulation?
Because design isn’t neutral: an interface is a choice about how to allocate attention, time, and desire. For decades it was treated as the user’s individual choice (“if you don’t like it, close the app”). Research on compulsive behaviour and measurable consequences for adolescents and public health has shown that design has systemic effects—and what is systemic falls under public regulation.
What changes for tech companies operating in Europe?
Mandatory impact assessments, fines up to 6% of global revenue, researcher access to data to verify the claims. It isn’t just TikTok: the DSA applies to Facebook, Instagram, X, YouTube, Snapchat, LinkedIn, Amazon, Booking. Anyone designing social media or e-commerce for the EU market has to document design choices with cognitive impact and be ready to defend them on technical grounds.
Why is this a cultural fight as well as a regulatory one?
Because it redraws the relationship between business model and public health. For twenty years the attention economy treated human attention as a resource to extract, maximising “engagement” as a KPI. Europe is saying: users’ brains are not a KPI for your company. It’s a moral statement dressed as technical regulation—and that’s why it makes more political noise than a debate about algorithms ought to.
The author
