The idea that we can separate "wellbeing" from "digital wellbeing" gets more distant by the day. The latter is defined as "the balanced, measured use of technology to improve people's quality of life" (Tayiz et al., 2025) [1]. Technological progress has undeniably improved quality of life in many ways — access to education, healthcare, and convenient services. But excessive screen time has increasingly been linked to serious health issues, from sedentary behavior and its impact on metabolic health, to mental health concerns including depression, anxiety, social isolation, and addiction.
In 2012, when the term started gaining academic traction, the separation was physical and obvious: if you left your phone at a relative's house for a weekend, life went on just fine. You'd miss a few notifications, maybe a call or two, but nothing too costly. In 2026, it's different: you lose access to the authentication keys for logging into work platforms, health services, payments, and transportation. The cost isn't just FOMO (Fear of Missing Out), but the countless disadvantages of living at an analog pace and losing access to services altogether.
This shift from optional tool to a kind of mandatory infrastructure has produced a certain environmental compulsoriness. When we say the air is "compulsory," we mean you can't choose not to breathe it — and when someone puts pollutants into it, you breathe worse air with no alternative. The digital world now works the same way. Since opting out is functionally expensive, exposure to everything that happens within it stops being well described as voluntary, through the metaphor of free choice. Following the analogy, the digital world works like air: whatever gets injected into it, we absorb by default — advertising, engagement algorithms, recommendation systems, and so on. Digital environments, as they exist today, are hostile to autonomy.
The digital world is mandatory, and what it charges us isn't an optional fee — it's closer to a toll, paid in attention. According to DataReportal's 2026 report, the average Brazilian spends 9 hours and 13 minutes a day in front of screens [2]. In a quick, conservative estimate we ran, that translates to more than 1 hour and 40 minutes a day spent on advertising content alone.
Digital advertising also has its own model. The classic model was the TV commercial break: the show stops, a message clearly marked as advertising plays, and the show resumes. The line between content and ad is sharp, and the viewer knows exactly when they're being persuaded. Today's model is different. Content itself functions as continuous commercial surface: influencers don't interrupt the feed to advertise — the feed is the ad. On social media feeds, the average ratio is 1 ad for every 4 or 5 organic posts. On video platforms, 15- to 30-second insertions appear before, during, and after content. And there's a third, less obvious but equally significant layer: influencer marketing. In Brazil, 87% of users follow at least one influencer, and the purchase conversion rate from influencer recommendations reaches 73% — an unprecedented figure globally (DataReportal, 2026). Add these three layers together, and roughly 20% of all connected time is spent financing the digital ecosystem with our attention. The line between what we "choose" to see and what someone paid for us to see gets blurrier by design — because it works.
This model wasn't born in the digital world. Exposure to advertising already existed in places you couldn't easily leave — buses, airports, elevators. What the digital world did was scale that logic up, removing the limits physical space imposes. All of it with far less regulation and oversight.
The digital wellbeing tools available today promote more "mindful" use of devices and digital literacy — night mode that reduces blue light, a reminder that "you've been on this app for 2 hours," a weekly screen time report. But they rarely give people more real autonomy, and tend to individualize the problem — supporting self-control — rather than redesigning the platforms themselves. In fact, such tools are used remarkably little, as I recently suggested in an empirical study. When digital environments, like social media, are designed to hold attention beyond conscious choice, exercising self-control is an incomplete solution. Mechanisms like the infinite scroll (reminiscent of a slot machine), the lack of natural pauses, the variable reward system of notifications and likes, and a means of quantified social validation exploit decision-making vulnerabilities, keeping us on platforms regardless of our "deliberate will." These mechanisms were described by former Google ethics specialist Tristan Harris in testimony before the U.S. Senate [3]. Even behavioral-science-based models, like Nir Eyal's Hooked [4], are cited in the design of many digital products. So on one side, you have a user with finite attention and self-control, partial knowledge, and 24 hours in a day. On the other, teams of engineers, data scientists, and interaction designers, backed by AI models running around the clock, optimizing systems to maximize attention. It's an uneven race.
Beyond the imbalance of power around attention capture, there's also the steering of decisions in the online environment in ways that favor the company over the user — a kind of evil "nudge." These design patterns are called dark patterns. Examples include a "cancel subscription" button that requires five confirmation screens, a privacy setting that's off by default and has to be manually re-enabled after every update, or a pop-up asking "are you sure you want to leave?" with the "stay" button highlighted, among others. The site deceptive.design documents and categorizes types of dark patterns using real-world examples.
A useful analogy is food safety. We don't ask consumers to bring a lab kit to the supermarket to test whether each food item has pesticide levels above the legal limit, whether meat was stored at the right temperature, or whether preservatives are safe for long-term consumption. That's why we have health agencies, food legislation, and inspection systems. We recognize that individual consumers can't compete with the scale and sophistication of industry. The same reasoning should apply to the digital environment.
Regulatory precedents point to possible paths forward. The European Union's Digital Services Act (DSA), in force since 2024, requires large platforms to explain how their recommendation systems work and to offer a non-personalized feed option. The DSA also explicitly bans the use of dark patterns in consent interfaces.
The UK's Age-Appropriate Design Code takes a complementary approach, requiring platforms accessed by minors to default to higher-privacy, lower-exploitation settings. The default can't be "collect everything and show personalized ads" — it has to be "collect the minimum and don't track."
The GDPR (General Data Protection Regulation), since 2018, has required explicit user consent for tracking and data collection in Europe. That sounds like a win for autonomy. In practice, what happened (until 2024, when new legislation took effect) was that pop-up windows started appearing everywhere on every site — and the design of those windows was telling: the "accept all" button was large, colorful, and one click away; the "reject" button was buried in submenus, sometimes requiring three or four clicks [5]. While the GDPR marked major progress in European legislation, without rules governing not just what is required but how it's presented to the user, the mere existence of a formal right likely doesn't translate into real autonomy.
In Brazil, legislative discussions on the digital environment and platform regulation are moving forward, though in a fragmented way and often reactively, in response to specific crises. The Marco Civil da Internet (2014) and the LGPD (2018, Brazil's data protection law) laid important groundwork, but were written for a less sophisticated digital ecosystem than the one we have in 2026. The most ambitious regulatory attempt was Bill 2630/2020, known as the "Fake News Bill."
Bill 2630 proposed algorithmic transparency obligations (platforms would have to explain how their recommendation systems work and make data available to researchers), a duty of care that would hold platforms accountable for proactively moderating content that incites violence, hatred, or disinformation, and advertising transparency rules — including a requirement to maintain public repositories of ads run, with information on who paid, how much, and what targeting criteria were used [5]. The bill was shelved in 2024, and a much softer proposal was introduced in its place [6].
None of these initiatives is perfect. All face challenges around implementation, enforcement, and keeping pace with change. But they all share a common principle: user autonomy needs regulatory infrastructure, not corporate goodwill. We'd add one thing: autonomy needs to be designed and operationalized with real human behavior in mind, grounded in what behavioral science, cognitive science, and neuroscience actually know about digital behavior. If our lives are also digital, social rules can't stop at the edge of that space.
Looking at our relationships with technology — and the systems built around them — through a behavioral lens is one of our core interests here at the Brazilian Institute of Behavioral Sciences.
[1] Tayiz, V., Vangölü, M., Özok, H., & Tanhan, F. (2025). Concept of Digital Well-being. Psikiyatride Güncel Yaklaşımlar, 17, 673–686.
[2] Screen time and digital behavior figures cited throughout this piece come from the Digital 2026: Brazil report, published by DataReportal in partnership with We Are Social and Meltwater. Available at datareportal.com/reports/digital-2026-brazil.
[3] Tristan Harris Congressional Testimony: Understanding the Use of Persuasive Technology.
[4] Eyal, N., & Hoover, R. (2014). Hooked: How to Build Habit-Forming Products. Portfolio/Penguin.
[5] Brazilian Bill 2630/2020 ("Brazilian Law on Freedom, Responsibility and Transparency on the Internet") was proposed by Senator Alessandro Vieira (Cidadania-SE) and approved by the Senate in June 2020.
[6] The shelving of Bill 2630 and its replacement with a softer proposal (Bill 4691/2024) was documented by the Coalizão Direitos na Rede.
Enjoyed this? Subscribe to Café Comportamental for weekly essays in Portuguese, or explore more in Resources.