Back to School 2026: Under Pressure
The autumn of 2026 opens onto a political and technological landscape of rare density. The United States is entering an electoral sequence whose outcome will shape the architecture of global digital infrastructure for a decade, while France prepares for a presidential election that promises to be the first fully contested on the terrain of algorithmic information.
What we still call, for convenience, "social networks" no longer designates a set of leisure applications, but the very infrastructure of public speech, civic debate, and now, population movements.
Three seemingly distinct sequences converged over the summer. The French legislature ratified a ban on social media for minors under fifteen and sketched out a disinformation observatory capable of invisibilizing accounts deemed at fault. The European Central Bank secured the green light from the parliamentary committee for its digital euro, while the US Senate barred the Federal Reserve from doing the same. And in Ceuta, within a few hours, between fifty and sixty thousand people crossed a European Union border, resulting in over seventy deaths, driven by a viral rumor amplified by algorithms whose metrics no one has yet been able to examine.
These three events are not separate chapters of the same book. They are manifestations of a single, unified pressure bearing down on open societies — a pressure we can call "digital convergence." The State, which had lost control over communication, payment, and information infrastructures, is seeking to regain it. The movement is legitimate in principle. It becomes worrying in its method, when institutional consolidation merges with the capture of spaces of freedom, when the punitive precedes the educational, and when the line between protecting citizens and controlling their speech becomes so thin it disappears.
Digital Identity: Necessary Remedy and Possible Poison
It would be unreasonable to deny the potential benefits of robust digital identity. Protecting minors from harmful content is a genuine concern, supported by converging mental health studies on the effects of doomscrolling, online harassment, and early exposure to pornography. Fighting bots, large-scale opinion manipulation, fraud, and impersonation also requires the ability to verify the identity of interlocutors. Dual anonymity, recommended by France's CNIL, where a trusted third party certifies age without revealing identity to the consulted site, represents an elegant technological compromise. The European digital wallet would offer a minimal-disclosure proof infrastructure, allowing one to certify an attribute — age, for example — without exposing one's full civil identity.
But the problem lies not there. It lies in the mechanics of repeated use that settle in. Asking a user to prove their age to access a social network, then to consult an adult-only website, then to regain control of a hacked account, creates a habit. By the fifteenth use, the logical extension will no longer be proving an attribute, but permanently linking civil identity to the account — what cognitive psychology calls a nudge. And this extension has already been explicitly requested, on multiple occasions, by the ruling party in France, through amendments aiming to link pseudonymous accounts to a civil identity accessible by the administration. This shift is not a conspiracy theory. It is documented, publicly debated, and announced by its proponents.
The question is not whether digital identity is good or bad. It is who accesses it, within what framework, with what adversarial guarantees. A digital identity system that protected minors without exposing adult speech would be progress. A system that, under the pretext of protection, installed a generalized identification infrastructure for public speech would be a regression.
The boundary between the two does not lie in the technology, which is neutral, but in the law that frames it and in the independence of the institutions that apply it. This is not a constant, but a variable subject to the political flows and pressures of a given time and place.
Public Speech Under a Regime of Attenuation
The report of the French Senate Culture Committee, published on July 9, proposes the creation, before the 2027 presidential election, of an independent disinformation observatory — a domestic counterpart to Viginium. This observatory could request platforms to invisibilize a user deemed at fault upon simple administrative request and execution. Centrist senator Laurent Lafon, who sponsored the text, used a revealing expression during his press conference, speaking of "domestic interference." This slip, or deliberate formulation, designates an unprecedented and vertiginous concept: a group of citizens coordinating to intervene in public debate with partially foreign funding could be classified as domestic interference and, on the sole decision of a state-funded association, be invisibilized without recourse to a judge, without adversarial debate.
Invisibilization is shadow-banning — that artificial de-amplification of an account's reach. Meta, Twitter, and YouTube executives have admitted it under oath before US congressional committees. The Twitter Files, published in December 2022, documented the existence of visibility filtering — those lists of accounts whose reach was minimized because their discourse did not align with the official narrative.
This is not a conspiracy theory. This is sworn testimony, in a legal system where perjury leads to prison.
The danger is not only frontal censorship. It lies in the opacity of visibility. When one does not know which opinions are tolerated, slowed, or rendered invisible, self-censorship sets in — not out of fear of explicit sanction, but out of permanent insecurity. The citizen no longer knows which discourse will endanger their account, which analysis will be labeled disinformation, which coordination will be labeled interference. This insecurity is incompatible with the exercise of a pluralist democracy, particularly during an election period.
Emmanuel Macron formulated it without ambiguity: he spoke of the need to move toward a gradual lifting of all forms of anonymity and a hierarchy of speech where the expert, the scientist, the political leader would carry more weight than the ordinary citizen. The hierarchy of speech is a seductive concept at first glance. Who would refuse to give more weight to the doctor than the charlatan, to the climatologist than the denialist? But the problem is one of selection. Who decides who is an expert? The State? A state-funded association? For a year, doctors maintained that Covid's origin was the pangolin, before the lab-leak hypothesis, initially labeled conspiracy by those same experts, became plausible. Political or scientific authority cannot be decreed. It is earned, verified, and lost. A society that institutionalizes the hierarchy of speech at the very moment when trust in its institutions collapses does not strengthen expertise. It discredits it further.
Punitive Before Educational
France's ban on social media for under-fifteens, adopted on July 21 by the French Parliament, illustrates a structural flaw in contemporary digital governance: the temptation to respond to a real problem with a coercive mechanism rather than an educational investment. No one seriously disputes that children's exposure to violent, pornographic, or addictive content constitutes a public health issue. The studies on adolescent mental health, the Facebook Files revelations showing that Meta knew about Instagram's harmful effects on young people — everything converges toward the need to act.
But the chosen method rests on a fragile technological premise. Verifying a user's age requires verifying everyone's identity — through facial recognition, France Connect, or the European digital wallet. The Australian experience, which preceded France's, shows that nearly eighty percent of minors bypass the system. In China, where the State sovereignly controls its digital infrastructure behind the Great Firewall, the ban can be enforced. In Europe, where platforms are American or Chinese and cloud infrastructure is over sixty percent American, technical sovereignty does not exist. We are therefore about to educate a generation to circumvent the law from the earliest age, just as young Tunisians circumvented Ben Ali's censorship during the Arab Spring.
An alternative existed. Requiring smartphone and computer distributors to set up extremely restrictive parental filters by default, unlockable only by parents after identity verification, would have transferred responsibility to parental authority. Combined with a nationwide communication campaign warning about the effects of doomscrolling and harassment, this approach would have addressed the problem at its root without building a generalized identification infrastructure. But this approach requires considerable educational and cultural investment.
A ban, by contrast, requires a law and an algorithm. The punitive is always cheaper than the educational — until the bill for mistrust comes due.
The Digital Euro: Monetary Sovereignty or Command Infrastructure
Mark Zuckerberg's announcement of Libra in June 2019 triggered a salutary panic. A private company was about to mint money on a planetary scale, for its two and a half billion users, with a single click. The reaction of central banks was of rare violence and remarkable effectiveness. Libra was put to death within a few months by a global coalition. But on that day, central banks discovered that their monopoly over money — so ancient an obviousness that no one thought to name it — could be attacked.
The digital euro was born of that fear, not of citizen demand. The initial promise was beautiful: offering everyone a banknote in their phone, the security of central bank money without intermediaries — digital cash. But reality eroded every promise. The anonymity of the banknote was publicly abandoned. Christine Lagarde acknowledged it before the European Parliament: digital money leaves a trace; it will not be fully anonymous. Disintermediation, which was supposed to restore a direct link between the central bank and citizens, was abandoned in favor of commercial banks, which will distribute the digital euro and verify identities. The project designed to bypass intermediaries will be distributed by intermediaries. Finally, the holding ceiling, discussed at around three thousand euros, reflects a structural incapacity: if the digital euro were genuinely safer than bank deposits, it would empty banks at the speed of fiber optics.
Yet the digital euro also defends a legitimate sovereignty. Stablecoins — those dollar-pegged tokens — already weigh approximately three hundred billion dollars, and ninety-seven percent are denominated in dollars. The euro's share is below zero point three percent. The US Senate, on June 22, 2026, barred the Federal Reserve from creating a digital dollar, betting on the private sector. Europe, twenty-four hours later, validated the legal framework for the digital euro. Two models are clashing. The first leaves the market to steer digital money. The second entrusts it to the public institution that manufactures the euro. The second is not perfect. But the absence of European digital monetary sovereignty, in a world where dollar stablecoins dominate, would be a strategic abdication.
The decisive question is not the existence of the digital euro. It is programmability. The European regulation explicitly excludes it: no euros with expiry dates, no earmarked spending. But infrastructure, once built, cannot be uninvented. What one text prohibits, another text can authorize — and one day of crisis, for excellent reasons. The Shenzhen experiment of October 2020, where public money evaporated from phones at midnight, is today theoretically illegal in Europe. But it is no longer impossible. Augustin Carstens, General Manager of the Bank for International Settlements, described it bluntly: central bank digital currency offers absolute control over the rules and the technology to enforce them. This description does not come from an opponent. It comes from inside the system, at the highest level, and it is perfectly owned. A programmable currency is no longer quite money. It is an instrument of behavioral policy — money that no longer obeys its holder but its issuer.
Ceuta: Physical Border and Informational Border
On July 30 and 31, 2026, the border between Morocco and the Spanish exclave of Ceuta was crossed by tens of thousands of people within a few hours, driven by a carefully constructed rumor. A distorted interpretation of a Spanish Supreme Court ruling, a rumor of a nighttime opening, a Facebook group gathering over twenty thousand members sharing maps, routes, and fundraisers, relayed by TikTok, WhatsApp, and Instagram. Over seventy dead. Thierry Breton, former European Commissioner, described this event as the first large-scale hybrid algorithmic attack conducted against an external border of the Union.
The classification must be examined rigorously. What is known is that viral content preceded and accompanied a massive population movement. What is unknown is infinitely more important: no platform has communicated the circulation metrics of this content. How many views? What geolocated reach? What share of algorithmic amplification, bots, and authentic virality? The opacity is total. It is this opacity that makes analysis impossible and prevention illusory.
The DSA — the European Digital Services Act — precisely provides the instruments to respond to this type of situation. Article 34 requires very large platforms to assess the systemic risks their services pose to public security and democratic processes. Article 35 imposes proportionate mitigation measures. Article 36 enables the activation of a crisis response mechanism. Article 40 opens data access for vetted researchers. None of this would have prevented the residents of Fnideq from hoping to reach Europe. But all of this would have made the manipulation traceable, contestable, and costly for its initiators.
The 21st-century border is simultaneously physical and informational. Pretending to secure one without the other is an illusion. But the European response cannot consist of replicating the Chinese model of integrated surveillance, nor of accepting the American laissez-faire that has prevailed since Section 230. The European model — that of binding transparency and ex-ante platform responsibility — is the third path. Its value depends on its effective application. A celebrated but unenforced regulation at its first serious test is not a regulation; it is posturing.
Interference: Real, Alleged, and Politically Useful
The question of foreign interference runs through all these debates and must be treated with particular intellectual discipline. There is established interference. The instrumentalization of migration by Belarus at the borders of Poland, Lithuania, and Finland since 2021 is documented. Russia's use of troll farms, influence purchases by state actors on Western platforms, the deployment of disinformation campaigns by all means — these are established facts. The possibility of foreign instrumentalization in the Ceuta crisis is enough to require a thorough, independent, and public investigation. But this possibility never dispenses with the need for proof.
It is essential to distinguish three registers. Established interference, which calls for a proportionate diplomatic, economic, and technical response. The plausible hypothesis, which calls for an investigation and the mobilization of existing legal instruments. And the convenient accusation, which consists of labeling as foreign interference what is merely a disturbing domestic opinion. It is on this boundary that the concept of "domestic interference" becomes dangerous. Can a group of French citizens, partially funded by foreign actors but expressing a political opinion, be labeled as interference? If so, then any opinion that does not conform to the dominant narrative can be disqualified by stigmatization, not by argument.
The temptation is strong, in the electoral context of 2026 and 2027, to use the accusation of interference as an instrument of domestic politics. It would protect democracy if it exclusively targeted proven external manipulations. It would destroy it if it served to disqualify the opposition. The line of demarcation can only be drawn by an independent, adversarial, and public investigation — never by a closed-door administrative decision. This is precisely what the DSA provides with data access for vetted researchers and transparency of moderation measures. But it must be applied.
Three Models in Collision
The convergence of digital pressures in the fall of 2026 highlights three governance models clashing without any having yet demonstrated lasting superiority. The Chinese model rests on integrated technological sovereignty — from the firewall to national platforms, from programmable digital currency to social credit. It is effective within its framework but incompatible with an open society. The American model bets on the market, private stablecoins, and Section 230, abandoning public regulation in favor of platform self-regulation. It has produced the most dynamic innovation of the century, but also the most extreme concentration and the most opaque manipulation. The European model, finally, seeks the regulatory path — with the DSA, the digital euro, the digital wallet, and the Migration Pact. It is ambitious in its normative ambition but fragile in its technical sovereignty.
Europe depends on American cloud infrastructure for over sixty percent of its needs. Its social spaces, from X to TikTok, are American or Chinese. Its ability to enforce the DSA depends on the goodwill of platforms whose headquarters lie outside its direct jurisdiction. The European regulation is an instrument of power, but it is not itself a power. This dissonance between normative ambition and technical dependence constitutes the structural flaw of the European model. Until it is filled by massive investment in technological sovereignty (infrastructures, semiconductors, artificial intelligence, and platforms), European regulation will remain an elegant shell whose content escapes its control.
Governing Without Capturing
The fall of 2026 thus condenses a series of milestones that outline the architecture of digital politics for the coming decade. Digital identity is being built. The digital euro is advancing. The DSA exists. The Migration Pact has entered into force. Each of these instruments, taken in isolation, responds to a real need. Protecting minors, monetary sovereignty, border security, the quality of public debate — these are legitimate objectives. The question is no longer whether the State must regain control. It must.
The question is how.
The conditions for legitimate digital governance can be formulated simply, even if their application is difficult. The purpose of each instrument must be limited and declared: protecting minors, not surveilling adults; securing money, not programming behaviors; tracing manipulations, not ranking opinions. Independent oversight must be real, exercised by a judge or a body guaranteeing impartiality — never by an association funded by the executive. Algorithmic transparency must be binding, with data access for researchers and public auditing. Recourse must be effective, with adversarial debate and the possibility of contestation. Finally, the separation between identity, expression, and money must be maintained as a structural principle, to prevent a single infrastructure from simultaneously serving to identify, to measure reach, and to program behaviors.
We should never have to choose between control and freedom. Open societies have always needed to control certain flows — whether illicit financial flows, foreign manipulation content, or critical infrastructure. The question is to prevent tools designed to protect from becoming architectures of docility. The digital euro, digital identity, the DSA, and the Migration Pact can be the pillars of a renewed democratic sovereignty. They can also become elements of an integrated surveillance apparatus whose model China would supply and whose logic Europe, through lack of vigilance, would accept while claiming to refuse it.
The American and French elections to come go beyond simple political ballots. They are the first full-scale test of democracies' ability to organize free debate in an information space saturated with algorithms, rumors, manipulations, and attenuation mechanisms. What will be decided in the coming months — in terms of legal frameworks, enforcement of existing regulations, and technological sovereignty — will determine not only the outcome of the elections, but the very nature of the public space for the generation beginning to vote. Technologies are elusive, often subverted, sometimes subvertible.
Governance is battered, but not yet defeated. Everything depends on what we demand of those who claim to take it in hand.
