The Facebook and Instagram agreement with the American states: an epochal change

When time becomes a design problem

In the United States, the relationship between teenagers and two of the most used social platforms in the world is about to change significantly. At the end of August 2026, Meta reached an agreement with 52 attorneys general of American states and territories and the District of Columbia that introduces a series of automatic protections for users under 18 on Facebook and Instagram. The agreement, approved by the judge on August 27, provides for a potential payment of approximately $18 billion spread over ten years and requires that most of the new measures remain in force for the same period.

The innovation that attracted the most attention is the introduction of a default limit of 2 hours in total per day between Facebook and Instagram. It is therefore not two hours for each platform: the time is counted cumulatively and Meta plans to aggregate it even when it manages to identify multiple accounts belonging to the same teenager. Once that threshold is exceeded, the ability to continue using the applications normally will require parental consent.

However, the agreement intervenes on many other aspects of the digital experience, introducing interruptions, nighttime restrictions, less stress during school hours and new parental supervision tools.

The meaning of the agreement therefore does not only consist in having identified a maximum amount of time, because it intervenes directly on the architecture of the experience: it modifies the default settings, introduces moments of interruption, reduces the ability of the platforms to continuously recall the user and gives parents a greater role in the management of teenagers' digital time.

It is also important to immediately clarify another element. The new rules only affect those under 18 in the United States and territories participating in the agreement; they do not automatically apply to Italian and European adolescents. Social platforms generally require a minimum age of 13 for the ordinary opening of an account, so the group directly affected is above all that between 13 and 17 years old, even if part of the new measures is intended to identify younger children or adolescents who have falsely declared that they are adults.

Then there is a lesser-known aspect of the agreement that could prove even more important in the long term. Meta explicitly asked TikTok and YouTube to take similar measures, arguing that intervening only on Facebook and Instagram could simply shift teenagers' time from one platform to another. The request is not purely symbolic: approximately 5.3 billion dollars, corresponding to approximately 30% of the total amount envisaged by the agreement, are linked to specific conditions which include the adoption by TikTok and YouTube of an even more restrictive daily limit, equal to 1 hour per day, together with night modes and age verification systems. Meta therefore requests that these protections progressively become a common standard for the entire sector.

The agreement does not represent an admission of liability. Meta continues to reject accusations that he deliberately designed Facebook and Instagram to create addiction in young people and disputes the possibility of establishing a simple causal relationship between the use of social networks and mental health problems. It would therefore be incorrect to interpret the 2-hour threshold as if new scientific evidence had suddenly demonstrated that one hour and fifty-nine minutes are harmless while two hours and one minute become harmful. The meaning of the decision is different and, in my opinion, much more interesting: a defined amount of time enters such a relevant way into the default architecture of the product for the first time.

A few months ago, in Social Dilemma. Tristan Harris, notifications and the Center for Humane Technology, I had tried starting from an apparently banal gesture: deactivating non-essential notifications. The messages, news and updates would have remained waiting for us, while something much more important than it might seem would have changed, that is, who decides when to interrupt our time. With active notifications it is often the applications that summon us; without notifications we choose when to enter.

In Stopping as an act of freedom. Because we read less and less, we scroll more and more and we risk running out of time to understand. I subsequently tried to broaden the same reflection to the way in which digital environments are modifying our relationship with attention, reading and the ability to interrupt a flow. The problem, I argued, is not simply the brevity of the content nor the number of minutes spent in front of a screen. Five minutes intentionally dedicated to a video is not equivalent to two hours of automatic scrolling, just as a chosen content is not equivalent to a feed that continues to produce another before we have even formulated a new intention. The decisive difference concerns control over our time.

The Meta case now brings that same question onto much more concrete ground, in which what for years has been discussed mainly in cultural, psychological and economic terms progressively also enters the field of legal responsibility. The question no longer just concerns what children, parents and educators should do to make better use of a smartphone, because it is also starting to concern how the digital environments in which we spend an increasing part of our days should be designed.

One of the most interesting testimonies that emerged during the proceedings was that of Adam Mosseri, head of Instagram, who has led the platform since 2018 and who previously held important product roles within Facebook. Mosseri was questioned, among other things, about the "Take a Break" function, introduced by Instagram to invite users to stop using it after a certain period.

When activating the function was left entirely to the user's choice, only a very small percentage of adolescents actually decided to use it. Mosseri argued that many kids simply didn't want to turn it on, and noted that Instagram later turned several protections into automatic settings for teen accounts. Other testimonies reported the existence, within the company, of discussions on the possible impact of some security measures on engagement metrics.

The circumstance is interesting above all because it allows us to observe a problem that goes far beyond Meta's single decision: how much can we rely exclusively on the user's self-control when the product he is using has been simultaneously designed to make continuing simple and pleasant?

Almost five hours a day

To understand the dimensions of the phenomenon we must also look at the time spent on the platforms. Gallup research conducted in 2023 on over 1,500 American teenagers estimated an average of around 4.8 hours a day spent on seven major social platforms. 51% of the adolescents interviewed declared at least 4 hours a day, while the average increased from 4.1 hours for thirteen-year-olds to 5.8 hours for seventeen-year-olds.

In the same study, YouTube was the platform on which teenagers spent the most time on average, with around 1.9 hours a day, followed by TikTok with 1.5 hours and Instagram with around 0.9 hours.

These are self-reported data and must therefore be interpreted with caution, because remembering and precisely quantifying one's time online is not easy. The order of magnitude remains relevant: 4.8 hours a day means more than 33 hours a week, an amount of time now comparable to the duration of many working weeks.

However, it would be wrong to automatically transform those hours into useless or harmful time. On social media we converse, listen to music, search for information, participate in communities, produce content, discover books, ideas, places and people. The problem therefore does not concern technology itself nor the possibility that it occupies a significant part of our day.

The more interesting question is how much of that time is deliberately chosen and how much comes from an environment designed to prolong the interaction.

The distinction becomes particularly important during adolescence, when skills such as self-regulation, impulse control and the ability to delay gratification are still developing. Here too we must avoid simplifications.

The Advisory published in 2023 by the Surgeon General of the United States underlined that up to 95% of children between the ages of 13 and 17 use at least one social platform and that more than a third declare they use them "almost constantly". The same document recalled studies in which more than 3 hours a day of use were associated with approximately double the risk of experiencing some mental health problems, including symptoms of anxiety and depression.

The decisive word, however, is association. Spending many hours on social media does not automatically demonstrate the existence of an addiction and does not, on its own, allow us to establish that any anxiety, depression or relationship difficulties have been caused by the platforms. The scientific literature provides a more complex picture, in which the effects may depend on the type of use, the content encountered, individual conditions, family and social relationships and above all on the activities that online time ends up replacing.

A teenager may use social media more because he experiences a condition of loneliness or discomfort and, at the same time, some methods of use can contribute to amplifying that same condition. Transforming this complexity into the formula “social media causes depression” would be scientifically inappropriate.

The Meta trial poses a different question, which does not require demonstrating that every additional hour automatically produces damage: what responsibility does anyone who designs an environment have when they know more and more about the mechanisms through which that environment manages to prolong the user's stay?

The conflict incorporated into the product

This is where the Meta case connects directly to the critique of the attention economy developed by Tristan Harris many years before the current litigation. When the success of a product is also measured through dwell time, return frequency, views and interactions, there is inevitably an economic incentive to reduce the occasions in which the user could interrupt the experience.

It is not necessary to imagine that someone explicitly decides to create addiction. It is sufficient to observe the functioning of an optimization system in which a specific graphic choice, a notification, a suggested content or a different video sequence method are continuously compared on the basis of their ability to produce measurable results. If a solution on average increases the time spent in the application or the probability that the user will return, the platform can learn this through experiments conducted on enormous quantities of interactions and progressively use that information to improve its retention capacity.

The disproportion that I described when talking about Harris acquires a particular meaning here. On the one hand there is a person with inevitably limited attention span and self-control, influenced by tiredness, boredom, curiosity, the need to belong and the search for approval; on the other hand there is a system equipped with enormous quantities of data, capable of continuously experimenting and adapting what it proposes based on the user's previous responses. In the case of an adolescent this asymmetry takes on further relevance because it concerns a person whose self-regulation skills are still in the development phase.

The question, therefore, does not consist of establishing whether the user formally has the ability to close Instagram, because naturally he can do so. The problem is understanding how substantial that freedom remains when it is exercised within an environment whose economic success also depends on the ability to postpone the moment in which we choose to leave.

Because two hours matter

It is precisely from this point of view that the 2-hour limit introduced by the American agreement takes on a meaning that goes beyond the number itself. There is no biological frontier capable of making one hour and fifty-nine minutes harmless and two hours and one minute dangerous, nor does the agreement pretend to establish it. What changes is the starting point of the choice.

Until now, the dominant architecture of the platforms has essentially entrusted the user with the task of establishing when to stop using it, while the product could continue to offer new content indefinitely. With the new limit, at least for the American teenagers affected by the agreement, a discontinuity is inserted into the architecture of the platform itself. After 2 hours, continuing no longer simply represents the automatic continuation of the experience, because it becomes necessary to modify a protective configuration with the involvement of a parent.

The difference may appear subtle, but it concerns one of the most important concepts in behavioral economics: the weight of default options. What is presented as the initial configuration frequently tends to be maintained, both because changing it requires further action and because that choice is implicitly perceived as normal or recommended. The “Take a Break” story shows something similar: asking a teenager, while using a platform, to voluntarily choose a tool that will subsequently invite him to stop produces a very different effect than incorporating that break directly into the ordinary functioning of the product.

The fact that Meta is now asking TikTok and YouTube to adopt similar rules adds another element to the discussion. The company's argument is understandable: If a teen reaches the limit on Instagram and can immediately transfer the same behavior to TikTok or YouTube, the overall effect on their digital time is likely to be modest. Meta therefore goes so far as to argue that the problem must be addressed at an ecosystem level and not at a single platform level.

There is something historically significant in seeing a company that has built an important part of its development on the growth of engagement today asking its competitors to jointly introduce systems capable of limiting it. It is perhaps precisely this apparent contradiction that makes the agreement more important than the 18 billion dollars that accompany it, because it signals a possible change in the way in which we begin to consider attention: no longer just as a resource that the user is free to offer and the company is free to conquer, but as something for which technological architecture can also assume a responsibility.

And after eighteen?

Finally, there remains a question that the agreement, by its nature, does not address. The new protections concern adolescents and this choice is understandable: the greater vulnerability of minors, the responsibility of parents and the particular stage of development make specific forms of protection more easily justifiable. At 18, the architecture of the platforms does not change and the mechanisms through which our attention is captured and maintained do not suddenly disappear.

An adult certainly possesses capacities for self-determination and responsibility that cannot be equated to those of a teenager, and it would be difficult to imagine the State automatically determining how much time an adult can spend on Instagram, Facebook, TikTok or YouTube. However, this does not mean that the question of design suddenly loses meaning.

Even an adult can reach the end of an evening and realize that they have spent an hour scrolling through content without ever having consciously decided to dedicate that hour to it; he can open his phone to reply to a message and find himself twenty minutes later in a sequence of notifications, videos and suggestions that he hadn't looked for; he can even know perfectly how these mechanisms work and still continue to suffer their effectiveness.

In What will we do with the freed time (and Keynes' prophecy) I addressed an apparently different question: if new technologies, and in particular artificial intelligence, really succeed in giving us back part of the time currently absorbed by work, who will take that freed time? Technology can reduce the time needed to carry out an activity without guaranteeing that that time actually returns to the person, because it can be immediately reabsorbed by new demands for productivity, communication, availability or consumption.

The problem of the attention economy seems to represent the other side of the same phenomenon. On the one hand we build increasingly powerful technologies to save time; on the other hand we have built environments capable of competing with extraordinary effectiveness for that same newly freed time. The question therefore does not only concern how much we will work in the future, but also who will be able to fill the hours that will no longer be necessary for work.

The difference between adolescents and adults concerns the level of protection we are willing to impose, while the question of the quality of choice remains surprisingly similar. If we deem it appropriate to remind a teenager of the time that has passed, reduce notifications in the hours in which he should be at school and introduce a discontinuity after 2 hours, we can at least ask ourselves why many of these tools should not be easily available, clearly visible and perhaps proposed as a default option also to adults, naturally leaving the latter the full freedom to modify them.

It would not be a question of paternalistically establishing how much time an adult can dedicate to social media. It would be a question of designing environments in which the user's freedom also includes truly effective tools for recovering awareness of the time that has passed, instead of entrusting this responsibility entirely to the individual ability to resist systems perfected precisely to keep attention alive.

In Stopping as an act of freedom I argued that interrupting a flow means recovering the possibility of deciding what to do with one's time. I continue to think so, but the Meta case suggests adding something to that reflection: freedom does not depend only on the theoretical existence of a button with which to close an application, because it also depends on the architecture of the environment within which that choice is exercised. A system can make a decision easier or more difficult, remind us that there is an exit point, or make that point continually disappear behind subsequent content.

For this reason, the debate on the hours spent by teenagers on social media risks becoming misleading if it focuses exclusively on finding a threshold beyond which to establish that use is excessive. The number of hours is certainly important, especially when it reaches dimensions capable of occupying a very significant part of the day, but the deeper question concerns the quality of the choice through which those hours are accumulated.

The age limit of 18 necessarily serves the laws to identify a threshold beyond which the level of protection changes. Our vulnerability to the attention economy, however, has no birthday after which it suddenly disappears. For this reason, the question that the agreement between Meta and the American States certainly concerns adolescents, but probably not only them: when we spend hours in built environments to gradually get to know what manages to hold us back, how much of that time really continues to belong to us?

---

Sources and references

Meta, Our Agreement With Bipartisan Attorneys General: Calling on TikTok and YouTube to Join Us in Supporting Teens, Meta Newsroom, August 26, 2026, updated following judicial approval August 27, 2026.

Meta, An Open Letter to TikTok and YouTube to Join Us in Supporting Teens, Meta Newsroom, August 26, 2026.

Reuters, Meta agrees to pay $18 billion to settle US lawsuits over children's social media addiction, 26 August 2026.

Associated Press, Instagram chief takes the stand in a trial pitting Meta against states over social media harms, August 2026.

Gallup, Jonathan Rothwell, Teens Spend Average of 4.8 Hours on Social Media Per Day, October 13, 2023.

Office of the U.S. Surgeon General, Social Media and Youth Mental Health: The U.S. Surgeon General's Advisory, U.S. Department of Health and Human Services, 2023.

Di Nardo, A., Social Dilemma. Tristan Harris, notifications and the Center for Humane Technology, Appunti Corsari – armandodinardo.it, 17 June 2026.

Di Nardo, A., #AI#4 What will we do with freed time (and Keynes' prophecy), Appunti Corsari – armandodinardo.it, 23 June 2026.

Di Nardo, A., Stopping as an act of freedom. Because we read less and less, we scroll more and more and we risk no longer having time to understand, Appunti Corsari – armandodinardo.it, 1 July 2026.

Image Adolescent, Pablo Picasso, August 2, 1969 oil on canvas Paris, private collection © Succession Picasso by SIAE 2024

Note on the use of artificial intelligence — This article was drafted with the support of artificial intelligence tools for research, analysis, source organization and writing. The selection of content, critical interpretation and final responsibility for the text remain with the author.