The subway doors open, and a dense stream of people begins to pour out of the train. For a brief moment, every face appears strangely familiar, not because they are identical, but because the movement of the crowd, the architecture of the station, and the limitations imposed by the narrow platform dissolve individual differences into the appearance of a single indistinguishable body. Standing beside me, a man instinctively touches my shoulder and, with visible astonishment, says, “There were so many of them… Can you believe it? I didn’t think there would be this many. There are just too many.” Before I have the opportunity to respond, he mutters a curse under his breath as we struggle to enter the carriage before the doors close once again.

Inside, the impression persists. The same faces remain before us. They seem less like individuals than fragments of a single moving organism that could effortlessly divide itself at the next station and continue reproducing the same image of itself indefinitely. One suddenly acquires the peculiar sensation that the crowd possesses no clear limits, that it can always become larger than it already appears. Yet almost immediately another question interrupts this first impression. Were there really that many people? Or was this perception itself the product of spatial confinement? Perhaps the station was simply too small to accommodate an otherwise ordinary number of passengers. Perhaps the architecture of circulation magnified the experience of density without increasing the number of bodies present. Perhaps, had these very same individuals been dispersed across the streets of the city, nothing about them would have appeared remarkable. Or perhaps urban space itself distributes visibility unequally, concentrating some bodies while rendering others invisible. I cannot answer these questions. What I do know is that later that evening, when I encountered the same event on television, something no longer corresponded. The crowd that appeared on the screen referred to the same event I had experienced only hours before, yet it was no longer the same crowd. It was undoubtedly large, but not overwhelmingly so. Neither image appeared false. Neither completely exhausted what had happened. Somewhere between immediate experience, mediated representation, and political narration, reality had begun to fracture. What I had witnessed and what I had later been shown no longer occupied precisely the same world.

This discrepancy may appear trivial. It concerns neither constitutional law nor political philosophy. It could easily be dismissed as a simple psychological illusion produced by confined space, selective attention, or the inevitable imperfections of memory. Yet it is precisely through such apparently insignificant experiences that deeper political transformations begin to reveal themselves. Constitutional crises rarely announce their arrival through dramatic proclamations. They seldom begin with tanks in the streets or governments publicly suspending constitutions. More often, they begin quietly, concealed within ordinary experiences whose political significance remains almost invisible precisely because they appear so ordinary. The subway platform, therefore, is not important because of the number of passengers who occupied it. It is important because it forces us to confront a far more unsettling question: what becomes of constitutional order once reality itself can no longer be assumed to precede its representation?

For centuries, constitutional thought has rested upon an assumption so fundamental that it has rarely required explicit articulation. Political events occur. Institutions observe them. Citizens experience them. Courts establish facts. Legislatures enact laws in response to those facts. Governments derive their legitimacy from populations whose existence is assumed rather than questioned. Even disagreement presupposes a shared world in which competing interpretations refer back to the same underlying reality. Political conflict concerns how reality ought to be understood, not whether reality itself has become unstable. The distinction appears almost self-evident. There exists, first, an event; only afterwards does interpretation begin. Law, in this sense, has always assumed that facts precede norms, that reality precedes legal judgment, and that political representation follows political existence.

Yet this assumption has become increasingly fragile. Not because reality has disappeared, but because the technologies through which political reality is represented have acquired an unprecedented capacity to participate directly in its production. Representation no longer merely follows events. Increasingly, it precedes them, shapes them, amplifies them, and occasionally replaces them altogether. The political question therefore changes in a fundamental way. We are no longer concerned merely with distinguishing truth from falsehood, or propaganda from objective reporting. Such distinctions, although important, belong to an earlier vocabulary. The contemporary difficulty concerns the possibility that political reality itself may become technologically reproducible before legal institutions have acquired the conceptual tools necessary to distinguish authentic events from artificially generated ones.

At first glance, this observation appears to concern media rather than constitutional law. One might argue that manipulation, propaganda, and political deception are hardly new phenomena. Governments have always fabricated evidence. Political regimes have always manipulated public opinion. Photographs have been altered, speeches rewritten, testimonies coerced, statistics manipulated, and historical narratives reconstructed long before the emergence of artificial intelligence. Nothing about forgery itself is historically unprecedented. Indeed, one could argue that politics has never existed independently of representation. Every political order constructs narratives about itself. Every government attempts to shape public perception. Every constitution depends, to some extent, upon symbolic performances through which authority becomes visible and legitimacy becomes imaginable.

Yet artificial intelligence introduces a qualitative transformation rather than merely an incremental improvement in existing techniques of manipulation. The novelty does not lie in the existence of political fabrication but in its scale, accessibility, speed, and increasingly convincing character. Never before has the production of apparently authentic political realities become so inexpensive, so rapid, and so widely available. Images can now be generated rather than captured. Voices can be synthesized rather than recorded. Speeches can be produced without speakers. Demonstrations may appear to have occurred despite never taking place. Citizens may express opinions they never held. Public figures may deliver statements they never uttered. The technological distinction between documentation and fabrication begins gradually to dissolve. Consequently, the constitutional problem can no longer be reduced to misinformation. We are instead confronted with the emergence of technologies capable of manufacturing political reality itself.

It is precisely here that the classical theory of the state of exception begins to reveal its limitations. Since Carl Schmitt, discussions surrounding exceptional authority have largely revolved around a familiar constitutional question: under what circumstances may sovereign power temporarily suspend the normal operation of law? The historical examples are well known. Emergency provisions, martial law, constitutional crises, war, rebellion, insurrection, national catastrophe — all describe moments in which governments claim extraordinary authority because ordinary legal procedures appear incapable of preserving the constitutional order. Whether one ultimately accepts Schmitt’s conclusions or rejects them, the structure of the problem remains remarkably consistent. The state of exception appears as a rupture, an interruption of ordinary constitutional life, a legally ambiguous moment in which political decision temporarily exceeds legal regulation.

The 21^st century, however, increasingly confronts us with a different phenomenon. Exceptional power no longer necessarily requires dramatic constitutional declarations. Governments need not suspend constitutions publicly in order to exercise extraordinary forms of control. Instead, exceptional techniques gradually disperse themselves throughout ordinary administrative life, digital infrastructures, communication networks, algorithmic systems, surveillance technologies, and increasingly sophisticated mechanisms of information production. The exception no longer arrives as an event clearly distinguishable from normality. It settles within normality itself. It becomes continuous rather than episodic, ordinary rather than spectacular, infrastructural rather than explicitly constitutional.

This transformation carries consequences far beyond institutional design. If the classical state of exception concerned the temporary suspension of law, contemporary political power increasingly concerns the continuous reconstruction of the conditions under which law recognizes reality. Courts remain capable of applying legal norms only insofar as they retain confidence in evidence. Elections remain meaningful only insofar as citizens remain distinguishable from fabricated political actors. Public opinion continues to function democratically only insofar as “the public” itself exists independently of the technologies that claim merely to represent it. The moment these assumptions become unstable, constitutional order encounters a difficulty for which neither classical constitutional theory nor contemporary public law possesses an adequate vocabulary.

The crisis, therefore, is not technological before it is legal. Nor is it merely political before it is constitutional. It is, above all, a crisis of recognition. Constitutional law has always developed concepts through which new forms of power become visible. Before rights can be protected, violations must first be recognized. Before institutions can respond, they must first identify the phenomenon demanding a response. Every constitutional crisis begins, therefore, not with legislation but with language. It begins when existing concepts cease to describe political reality adequately, while new realities continue to emerge without names.

It is from within this conceptual gap that the present argument begins. I argue that contemporary constitutional thought is confronted with an emerging form of exceptional power that cannot be adequately understood through the traditional language of emergency powers alone. Artificial intelligence has become more than a technological innovation; under authoritarian conditions, it increasingly functions as an instrument through which political reality itself may be manufactured, reproduced, and strategically reorganized. Consequently, we are no longer confronted simply with a permanent state of exception. We are confronted with something different, something that has not yet received sufficient conceptual attention. I propose to describe this emerging phenomenon as the Artificial State of Exception — a form of exceptional political power that operates not primarily by suspending law, but by reconstructing the very reality upon which law depends. This, I shall argue, constitutes one of the most significant constitutional crises of our time, not because legal institutions have already responded inadequately, but because they have not yet fully recognized that the crisis has already begun.

From the State of Exception to the Artificial State of Exception

The concept of the state of exception has never merely described an extraordinary legal situation. It has always sought to explain something more fundamental: the relationship between law and political power at the precise moment when the former becomes incapable of containing the latter. From the constitutional provisions of the 19^th century to the constitutional experiments of the 20^th, exceptional powers were understood as temporary deviations from ordinary legality, justified by circumstances that threatened the very existence of the political community. Whether one follows Carl Schmitt in identifying sovereignty with the power to decide upon the exception, or Giorgio Agamben in demonstrating how the exception gradually colonizes ordinary political life, the conceptual structure remains largely unchanged. Exceptional power continues to presuppose the existence of an identifiable event — a war, an insurrection, a national catastrophe, or some other crisis sufficiently visible to justify extraordinary governmental authority.

Yet the political landscape of the 21^st century increasingly refuses to conform to this structure. Contemporary authoritarianism rarely depends upon dramatic constitutional ruptures. It seldom requires the public suspension of constitutions or the explicit declaration of emergency. Instead, it governs through continuity rather than interruption. The mechanisms through which exceptional power operates are no longer confined to constitutional provisions or emergency decrees. They have become embedded within ordinary administrative practices, digital infrastructures, communication systems, and technological environments that accompany citizens throughout their daily lives. The exception has ceased to appear exceptional. It has become an ordinary technique of government.

This transformation, however, cannot be explained solely through the language of permanent emergency. To say that the state of exception has become permanent is to describe only its temporal extension. It tells us that exceptional authority no longer disappears once the emergency has ended. It does not explain the profound transformation that has occurred in the object of exceptional power itself. Historically, the state sought to suspend rights, concentrate authority, control territory, discipline bodies, or regulate populations. Today, while these objectives certainly remain important, another ambition has quietly emerged alongside them: the ambition to govern reality itself.

This is precisely where artificial intelligence enters the constitutional discussion — not as a technological curiosity, nor as another instrument of administrative efficiency, but as a political technology capable of reorganizing the relationship between power and reality. Public debates frequently describe artificial intelligence in terms of automation, economic productivity, labor displacement, or ethical regulation. These discussions are necessary, yet they leave untouched what may ultimately become the most significant constitutional implication of the technology. Artificial intelligence has dramatically expanded the state’s capacity not simply to manage information, but to manufacture politically meaningful realities whose authenticity becomes increasingly difficult to verify.

The argument advanced here is not that artificial intelligence inevitably produces authoritarianism. Such a claim would be both historically naive and analytically misleading. Technologies possess no independent political destiny. Printing presses served both censorship and enlightenment; photography documented atrocities while simultaneously enabling propaganda; the internet expanded democratic participation while also facilitating surveillance. Artificial intelligence belongs to the same historical category. Its political significance emerges not from the technology itself but from the structures of power within which it becomes embedded. The constitutional problem therefore begins not with artificial intelligence as such, but with its strategic appropriation by political regimes seeking to expand the techniques through which power is exercised and legitimacy is reproduced.

Under these conditions, artificial intelligence becomes something more than an instrument of communication. It becomes an instrument for the production of political existence. Images no longer merely represent political events; they may constitute them. Speeches no longer merely communicate governmental intentions; they may be generated independently of those who supposedly deliver them. Public enthusiasm no longer necessarily emerges from actual citizens; it may be simulated through networks of automated accounts, artificially generated videos, fabricated interviews, synthetic voices, and algorithmically produced visual evidence. The distinction between documentation and fabrication, once relatively stable despite occasional manipulation, becomes increasingly unstable. Political reality begins to acquire the peculiar characteristic of technical reproducibility. Here we encounter a phenomenon that classical constitutional theory never anticipated. Traditionally, governments sought to persuade existing populations. Even the most authoritarian regimes ultimately depended upon real citizens whose obedience, participation, or silence carried political significance. Propaganda attempted to shape the consciousness of actual people. Elections, whether free or manipulated, still referred to electorates that physically existed. Demonstrations, whether spontaneous or organized, still required bodies occupying public space. Reality could certainly be manipulated, but it remained anchored to populations whose existence preceded governmental representation.

Artificial intelligence gradually weakens this dependence. It introduces the possibility that governments may no longer need to rely exclusively upon existing political actors in order to produce the appearance of legitimacy. They acquire, instead, the capacity to generate what might be described as Artificial People.

By Artificial People, I do not refer merely to automated social media accounts or internet bots, although these undoubtedly form part of the phenomenon. The concept is considerably broader. Artificial People are politically meaningful representations of citizens generated through artificial intelligence for the purpose of producing the appearance of public opinion, political participation, collective emotion, or democratic legitimacy. They may appear as digitally generated demonstrators celebrating governmental achievements; as fabricated eyewitnesses confirming official narratives; as synthetic journalists, commentators, or experts reinforcing political discourse; as artificially generated victims legitimizing exceptional measures; or even as digitally reconstructed political leaders delivering speeches they never actually delivered. Their common characteristic lies not in their technical construction but in their political function. They exist in order to occupy the constitutional place traditionally reserved for actual citizens.

The implications of this transformation are profound. Constitutional democracy has historically rested upon a simple yet indispensable assumption: the people exist prior to political representation. Governments derive legitimacy because citizens precede institutions. Elections matter because electorates already exist before ballots are counted. Public opinion matters because it reflects attitudes genuinely held by members of a political community. Even authoritarian governments, despite manipulating these processes, implicitly acknowledge this sequence. First there are people; afterwards comes representation.

Artificial intelligence quietly reverses this order. Representation increasingly precedes the people it claims merely to depict. Political legitimacy no longer depends exclusively upon convincing actual citizens. Increasingly, it may depend upon generating sufficiently convincing representations of citizens.

The constitutional danger therefore lies not simply in misinformation but in a gradual inversion of political ontology itself. Governments begin not merely to represent the people but to manufacture politically useful versions of “the people” whenever existing populations fail to produce the desired image of legitimacy.

It is at precisely this point that I propose the concept of the Artificial State of Exception. The Artificial State of Exception does not suspend constitutional law in the classical sense. Nor does it merely extend emergency powers indefinitely. Rather, it reconstructs the informational environment within which constitutional institutions attempt to distinguish genuine political reality from artificial political production. Instead of temporarily interrupting legality, it continuously reshapes the conditions under which legality becomes meaningful. Exceptional power no longer depends primarily upon the suspension of norms. It depends upon the strategic production of realities that render the distinction between authentic political events and artificially generated political appearances increasingly impossible to maintain.

This distinction may appear subtle, yet it fundamentally alters our understanding of contemporary authoritarianism. Under the classical state of exception, governments exercised extraordinary authority because reality itself was assumed to have become extraordinary. Under the Artificial State of Exception, governments increasingly acquire the capacity to produce the appearance of extraordinary realities whenever such realities become politically advantageous. The emergency no longer necessarily precedes exceptional authority. Exceptional authority may instead participate in constructing the emergency it subsequently claims merely to manage. The constitutional consequences of such a transformation remain largely unexplored. Public law has developed elaborate doctrines concerning emergency powers, judicial review, proportionality, constitutional necessity, and the temporary limitation of rights. Yet it possesses remarkably few conceptual resources for addressing situations in which the central object of political manipulation is neither law nor rights, but reality itself. How should constitutional institutions respond once evidence itself becomes technically reproducible? That question cannot be answered by existing constitutional doctrine alone because it arises from a transformation that constitutional doctrine has not yet fully recognized. We must first acknowledge that the constitutional crisis emerging before us is not simply another variation of authoritarian politics. It represents the emergence of a new architecture of power — one in which artificial intelligence expands the state’s capacity to construct the political realities upon which constitutional government has always depended.

Only after recognizing this transformation can we begin to understand its final political consequence: the emergence of what I shall call Artificial Dictatorship.

Artificial Dictatorship and the Constitutional Crisis of Manufactured Reality

If the preceding discussion is correct, then artificial intelligence should not be understood merely as another technological innovation requiring regulatory attention. The emergence of artificial intelligence undoubtedly raises familiar legal questions concerning privacy, intellectual property, data protection, discrimination, and algorithmic accountability. These issues are important, yet they belong largely to the domain of legal regulation. The argument advanced in this article concerns a different order of difficulty. It concerns the possibility that artificial intelligence is contributing to the emergence of a new constitutional environment in which political authority acquires an unprecedented capacity to manufacture the very reality upon which constitutional government has historically depended.

The history of constitutional thought repeatedly demonstrates that new forms of political power almost always emerge before constitutional language develops the concepts necessary to describe them. Absolutism preceded constitutional monarchy. Industrial capitalism preceded modern labor law. Totalitarianism preceded the international legal vocabulary of crimes against humanity. Constitutional theory has therefore always been reactive. It learns to recognize transformations only after those transformations have already begun reshaping political life. There is little reason to believe that the contemporary moment is fundamentally different. Artificial intelligence has already entered the institutional practices of governments, political parties, intelligence agencies, communication systems, and digital media. Yet constitutional theory continues, for the most part, to approach it as a question of technological governance rather than as a transformation in the structure of political power itself.

It is precisely this transformation that I propose to describe through the concept of Artificial Dictatorship. The concept should not be misunderstood. Artificial Dictatorship does not describe every authoritarian regime that happens to employ artificial intelligence. Instead, it refers to a particular configuration of political power in which artificial intelligence becomes an indispensable instrument for producing, reproducing, and managing political reality itself. It designates a form of authoritarian rule that no longer depends exclusively upon coercion, censorship, surveillance, or even mass mobilization. Rather, it increasingly depends upon the systematic production of artificial political existence.

Historically, dictatorships have sought to monopolize violence, communication, and political organization. They imprisoned opponents, censored newspapers, controlled broadcasting institutions, prohibited political parties, and restricted public assembly. Their objective was relatively straightforward: to suppress competing centers of political authority while preserving their own monopoly over legitimate political action. But even the most sophisticated propaganda machine still had to work on a population it did not create. Artificial intelligence fundamentally alters this historical condition.

The defining characteristic of Artificial Dictatorship is not simply that governments become more efficient at producing propaganda. Propaganda has accompanied political authority for centuries. Rather, Artificial Dictatorship emerges when political authority acquires the technological capacity to replace the uncertain behavior of actual citizens with increasingly convincing artificial substitutes. Instead of persuading existing publics, governments may begin producing artificial publics. Instead of waiting for spontaneous political support, they may generate synthetic demonstrations of support. Instead of documenting political reality, they may increasingly construct it.

The inversion described earlier now acquires its full political form. Where the Artificial State of Exception makes representation capable of preceding the people, Artificial Dictatorship makes that precedence a technique of rule: political representation no longer merely speaks in the name of the people, it begins producing the people in whose name it claims to speak.

This reversal should not be understood literally. Governments cannot create biological populations. They cannot manufacture citizens in the physical sense. What they increasingly possess, however, is the ability to manufacture politically effective representations of citizens whose practical consequences may become indistinguishable from those produced by actual political participation. Artificial crowds, artificial patriotic movements, and even artificial political opponents may gradually populate the informational environment within which constitutional legitimacy is publicly negotiated. Their importance lies not in whether they physically exist but in whether they successfully perform the constitutional function traditionally reserved for actual citizens.

This development introduces a constitutional difficulty that extends beyond ordinary concerns regarding misinformation. False information has always existed. Constitutional democracies have long relied upon public debate, investigative journalism, judicial review, and institutional transparency to expose deception. Artificial Dictatorship presents a qualitatively different problem because it systematically blurs the distinction between deception and political existence itself. The question is no longer whether governments occasionally lie. The question becomes whether constitutional institutions retain the practical capacity to distinguish authentic political actors from technologically generated political simulations.

At this point, the observations of Carl J. Friedrich acquire renewed significance. Friedrich distinguished between constitutional dictatorship, whose extraordinary powers remain directed toward preserving the constitutional order, and unconstitutional dictatorship, which ultimately destroys that order altogether. Yet he also recognized that no institutional mechanism could permanently guarantee that extraordinary powers would remain confined within constitutional purposes. As he observed, no superior constitutional device exists capable of ensuring that emergency authority will always be exercised in defense of constitutional government; ultimately, only the vigilance of the people themselves can perform that function.

Friedrich’s observation becomes even more unsettling under contemporary technological conditions. What becomes of constitutional vigilance once “the people” themselves become susceptible to technological reconstruction? How can citizens supervise governmental power if governments increasingly acquire the capacity to simulate citizenship? How can constitutional accountability function when public opinion itself becomes technically reproducible? The constitutional problem therefore shifts dramatically. Friedrich assumed the continuing existence of an identifiable political community capable of monitoring governmental conduct. Artificial Dictatorship calls precisely this assumption into question. It does not merely threaten constitutional institutions from above. It threatens the constitutional subject from below by destabilizing the very category of “the people” upon which constitutional government has always depended.

For this reason, Artificial Dictatorship should not be understood simply as a third historical category of dictatorship alongside constitutional and unconstitutional dictatorship. Rather, it represents a transformation in the architecture of authoritarian power itself. Its distinctive feature lies in its ability to operate without relying exclusively upon legal repression or physical coercion. Instead, it continuously reconstructs the informational environment within which political judgment becomes possible. It governs not only through force but through the production of credibility. It expands not merely the state’s monopoly over violence but its emerging monopoly over political appearance.

Seen from this perspective, the Artificial State of Exception appears not as an isolated constitutional anomaly but as the normal condition within which Artificial Dictatorship operates. The two concepts are inseparable. The Artificial State of Exception describes the continuous reconstruction of political reality through artificial means; Artificial Dictatorship describes the political regime that increasingly depends upon that reconstructed reality for the reproduction of its authority. One concerns the constitutional environment. The other concerns the form of political domination made possible within that environment.

The danger of this transformation lies precisely in its invisibility. Classical states of exception announced themselves through emergency declarations, constitutional suspensions, military mobilizations, or explicit legal ruptures. Citizens knew that extraordinary powers had been invoked because the legal order itself acknowledged the interruption. The Artificial State of Exception behaves differently. It rarely announces its arrival. It appears through ordinary technologies, familiar interfaces, persuasive images, convincing videos, synthetic voices, algorithmically generated conversations, and apparently spontaneous political communication. Its exceptional character is concealed beneath the appearance of normality. The constitutional crisis therefore becomes progressively more difficult to identify because nothing outwardly appears exceptional. For this reason, the first constitutional response cannot yet be legal regulation alone. Law cannot regulate a phenomenon it has not conceptually recognized. Before constitutions can respond to Artificial Dictatorship, constitutional theory must first acknowledge that the relationship between political power and reality has undergone a fundamental transformation. The language inherited from classical emergency powers, however valuable, no longer fully captures the emerging structure of contemporary authoritarianism. We require new constitutional concepts not because theoretical innovation is intellectually fashionable, but because existing concepts increasingly fail to describe existing realities. The concepts of an Artificial State of Exception, Artificial People, and Artificial Dictatorship are proposed here as an initial attempt to name this emerging constitutional condition. They do not claim to provide a complete theory of artificial intelligence or authoritarian politics. Rather, they seek to identify a transformation that constitutional law has not yet sufficiently recognized: the gradual movement from governments that merely govern reality to governments that increasingly participate in producing it.

Every constitutional crisis begins long before constitutional texts acknowledge its existence. By the time constitutions formally respond, the political transformation is usually well underway. The constitutional challenge posed by artificial intelligence belongs to precisely this category. It is not, at least in its first instance, a crisis of regulation. It is a crisis of recognition. The most urgent task before constitutional scholars is therefore neither to celebrate nor to condemn artificial intelligence in the abstract, but to understand the unprecedented political environment it is helping to create. Only by recognizing this new architecture of power can constitutional thought begin to formulate legal principles capable of resisting it. Before law can regulate Artificial Dictatorship, it must first learn to see it. That act of recognition is not the conclusion of constitutional inquiry; it is its indispensable beginning.