I. Introduction
There is a persistent and comfortable fiction in public life that order is the natural condition of society, and that deviance, as its disruption, is an aberration to be swiftly corrected. Sociological inquiry has long challenged this fiction (Goode, 2016). For Emile Durkheim, deviance was not the antithesis of a healthy society but its vital sign: a society without deviance is a society without the capacity for moral evolution (Durkheim, 1893/1984). Deviance marks the boundary of the permissible; it tests the resilience of institutions; and, at its most consequential, it rewrites the rules entirely. The question, then, is not whether deviance matters, for it manifestly does, but rather what may be learned when its most dramatic contemporary expressions are taken seriously as objects of sociological analysis.
This paper takes three such expressions seriously. The first is the Nepal Gen Z protests of September 2025: a mass youth uprising that toppled a government, burned a Parliament, and forced a constitutional reckoning in one of South Asia’s most politically turbulent democracies (Sharma & Travelli, 2025). The second is the cyber-deviance that animated this movement, namely the digital insurgency waged through TikTok, Discord, and encrypted messaging applications that enabled leaderless coordination at a scale that no single organiser could have achieved. The third is the most conceptually ambitious: the framing of artificial intelligence as a species of technological social deviant, an entity whose very existence transgresses the norms of human economic and epistemic supremacy, and whose emergence has provoked precisely the kind of elite regulatory anxiety that conflict and labelling theorists would predict.
The paper is structured as follows. Section II sets out the research objectives. Section III establishes the theoretical framework. Section IV conducts the substantive analysis across all three case studies. Section V reflects on methodology. Section VI offers conclusions. Throughout, the paper argues that these three expressions of deviance, political, digital, and technological, are not isolated phenomena but structurally connected manifestations of a single sociological proposition: that societies change not despite deviance, but through it.
II. Objectives
This paper pursues four interrelated objectives:
(i) To establish a working typology of deviance that encompasses formal, informal, and conceptual-technological forms, demonstrating that the sociological category is broad enough to include artificial intelligence as a legitimate analytical subject.
(ii) To examine the Nepal Gen Z protests through the lens of Merton’s strain theory, identifying both the structural preconditions and the social functions that collective deviance served in that context.
(iii) To analyse cyber-deviance not merely as a tool of convenience but as a principled sociological category, exploring its role as both the means and the message of contemporary youth politics.
(iv) To develop a conceptual framework for understanding artificial intelligence as a technological social deviant, drawing on conflict and labelling theories to explain societal fear, regulatory overreach, and the political economy of AI governance.
III. Theoretical framework
Any serious analysis of deviance must be grounded in theory. Three complementary frameworks animate this paper, each illuminating a different dimension of the phenomena under examination.
A. A. Strain theory (Merton)
Robert K. Merton’s strain theory, first articulated in 1938 and subsequently refined through decades of sociological debate, remains among the most generative frameworks for understanding why individuals and collectives turn away from legitimate social institutions (Merton, 1938). Merton’s core insight is elegant: in any society, cultural goals such as wealth, status, and political liberty are held out as universally attainable, while the legitimate institutional means of attaining them are differentially distributed. The poor, the young, and the politically excluded face a structural gap between aspiration and opportunity. This gap, the strain, is productive of deviance. Merton identified five adaptations: conformity, innovation, ritualism, retreatism, and rebellion. For this paper, the mode of rebellion is decisive. The rebel does not merely innovate around the system or retreat from it; the rebel rejects both its goals and its means, demanding a different order altogether. This is not pathology but principled structural critique expressed in the only language available to the structurally excluded.
B. B. Conflict theory
If strain theory explains why deviance arises, conflict theory explains who gets to name it. Rooted in the Marxian insight that social institutions reflect and reproduce the interests of the dominant class, conflict theory holds that the designation of an act as deviant or criminal is itself a political act (Chambliss, 1975). The powerful write the rules; the powerless are judged by them. What the elite calls a riot, the subaltern calls a revolution. What regulators call an unsafe technology, its developers call a breakthrough. Conflict theory thus shifts analytical attention from the deviant act itself to the social machinery that produces the label of deviance, and asks, always, in whose interest that machinery operates.
C. C. Labelling theory
Howard Becker’s labelling theory offers the third critical lens (Becker, 1963). For Becker, deviance is not an intrinsic quality of any act but a product of social interaction, specifically of the successful application of a label by moral entrepreneurs who have the power and inclination to enforce norms. The label, once applied, is sticky: it reshapes the labelled individual’s identity, constrains that individual’s opportunities, and frequently drives the person deeper into the behaviour being condemned, what Becker termed secondary deviance. Crucially, labelling theory draws attention to the contingent and contestable nature of deviance: what is labelled depends not on the objective character of the act but on who is doing the labelling, and whether they succeed in making it stick. In the context of mass protests and emerging technology, this framework is indispensable.
Together, these three theories form an interlocking analytical architecture: strain theory explains the structural genesis of deviance; conflict theory explains its political construction; and labelling theory explains its social perpetuation and the counter-strategies available to those who resist the label.
IV. Discussion
A. A. The Nepal Gen Z protests: Mertonian rebellion in the age of social media
To understand the Nepal protests purely as a law-and-order problem, as the Nepali government initially sought to frame them, is to misread a sociological text as a police report. What erupted in Kathmandu in September 2025 was the consequence of decades of compounding structural strain, detonated by a single governmental act: the banning of social media platforms (Sharma & Travelli, 2025).
The structural preconditions were formidable. Nepal’s youth unemployment rate had become among the highest in the region, with a large number of young Nepalis leaving the country each day in search of dignified work, most of them for menial labour in the Gulf states or Malaysia (World Bank, 2024). The cultural aspiration of economic self-sufficiency, so central to Nepal’s post-2006 democratic identity, was for an entire generation structurally inaccessible. Remittances had become the country’s economic lifeblood, accounting for roughly a quarter of gross domestic product, a figure that speaks less to Nepali economic success than to the failure of domestic institutions to provide meaningful opportunity (World Bank, 2024). Against this backdrop, Transparency International’s 2024 Corruption Perceptions Index placed Nepal among the more corrupt nations in South Asia, with a score of 34 out of 100 and a rank of 107th of 180 assessed jurisdictions (Transparency International, 2024). The political class, meanwhile, demonstrated a resilience to accountability that bred a culture of impunity visible to even the most politically disengaged citizen.
Merton’s framework maps onto this context with considerable precision (Merton, 1938). The cultural goals of democratic participation, economic mobility, and freedom from corruption were loudly proclaimed by successive governments. The legitimate means of achieving them were systematically blocked. The ballot box had proven ineffective: political dynasties recycled power among themselves regardless of election outcomes. Civil society organisations had been co-opted. Judicial redress was slow and uncertain. When the government blocked social media, the final available avenue for collective expression and organisation, it performed the structural function of eliminating the last legitimate outlet. Mertonian rebellion was the predictable result.
What followed was a sociological case study in the positive functions of deviance that Durkheim had theorised more than a century earlier (Durkheim, 1893/1984). The collective action of the protesters, including the symbolically significant burning of Parliament, served first as moral boundary clarification. In setting aflame the seat of a discredited legislature, the youth generation communicated, with unmistakable clarity, that the existing political order had, in their view, exhausted its legitimacy. The act was unlawful; its sociological function was nonetheless to clarify and dramatise a moral boundary that the protesters held the political class to have transgressed first.
Second, the protests generated extraordinary social solidarity. There is something peculiar and powerful about collective danger: it strips away the comfortable individualism of ordinary life and creates, momentarily, a community of shared stakes. The leaderless, decentralised structure of the Nepal movement was both its apparent weakness and its actual strength: without a leader to arrest or discredit, the movement was difficult to decapitate (Sharma & Travelli, 2025). The resulting solidarity produced a collective political identity, Gen Z Nepal, that transcended caste, ethnicity, and class in ways that Nepal’s fractious formal politics had rarely managed.
Third, the protests drove genuine, durable institutional change. The prime minister resigned. Parliament was dissolved. An interim government was appointed under Sushila Karki, the country’s first female chief justice and a jurist of notable independence, with new elections mandated (Sharma & Travelli, 2025). These were not cosmetic changes; they represented a structural rupture in Nepali politics. The deviant act had accomplished what years of conventional political activity had not: it compelled the system to reconstitute itself. Durkheim’s functionalist insight was thereby vindicated (Durkheim, 1893/1984).
B. B. Cyber-deviance: The architecture of digital rebellion
A crucial but underanalysed dimension of the Nepal uprising was its organisation. The protests did not begin on the streets; they began on screens. TikTok, Discord, and encrypted messaging platforms served as the connective tissue of a movement that had no formal leadership and no institutional backing. Understanding this digital infrastructure requires engaging with cyber-deviance not merely as a technological phenomenon but as a sociological one.
Cyber-deviance may be defined, following the broader tradition of deviance scholarship, as normatively transgressive behaviour conducted through or mediated by digital networks (Goode, 2016). In the context of the Nepal protests, the most analytically interesting form of cyber-deviance was not hacking or data theft, but the subversive use of social media for political counter-narration, specifically the Nepo Kid campaign that became the moral vocabulary of the movement.
The Nepo Kid campaign worked by applying Becker’s logic of labelling in reverse (Becker, 1963). Ordinarily, it is the powerful who label the powerless as deviant. Here, young Nepalis used the same mechanisms of stigmatisation, the viral image, the shared video, the devastating caption, to label the children of political elites as undeserving beneficiaries of nepotism and corruption. The juxtaposition was searing: images of a politician’s child holidaying in Europe, set beside statistics on Nepali youth unemployment. The label Nepo Kid was simple, memorable, and devastating, and it endured precisely because it named something that many Nepalis had experienced but lacked a collective vocabulary to articulate.
The government’s response, banning social media platforms, was, from a conflict theory perspective, a predictable exercise in elite social control (Chambliss, 1975). By silencing the platforms on which the counter-narrative flourished, the state sought to reassert its monopoly on political discourse. But the ban backfired: it confirmed, in the most visible possible way, that the government feared the speech of its citizens. Rather than suppressing the movement, the ban provided its most galvanising grievance and drove organisational activity onto encrypted platforms that were altogether harder to police. The power of cyber-deviance lies precisely in its distributed, redundant, and resilient architecture.
There is a deeper point here about the relationship between digital technology and the sociology of deviance. The internet has fundamentally altered the labelling dynamic that Becker described. In the pre-digital era, moral entrepreneurs, including governments, newspapers, and religious institutions, controlled the primary channels of mass communication and thus held a near-monopoly on the power to label. Social media has partially democratised this power: individuals and groups can now mount counter-labelling campaigns with a reach and speed that was previously available only to institutions (Becker, 1963). The Nepo Kid campaign is an exemplary instance of this democratised labelling, and the government’s resort to censorship is an exemplary instance of institutional panic in the face of its loss of narrative control.
C. C. Artificial intelligence as technological social deviant: A conceptual analysis
The most ambitious claim of this paper is its third: that artificial intelligence can be productively understood as a form of social deviance. This claim requires careful theoretical justification, because AI is not a human actor and cannot, in the conventional sense, choose to violate norms. Nevertheless, the sociological categories of deviance, control, and labelling illuminate the social reception of AI in ways that more conventional technology-studies frameworks do not.
The core of the argument is this: AI transgresses foundational social norms, those around labour, creativity, expertise, and human cognitive primacy, in ways that are structurally analogous to how human deviants transgress behavioural norms. When generative AI produces legal briefs, medical diagnoses, academic essays, and works of art at speeds and scales that human professionals cannot match, it violates the implicit norm that such activities are the exclusive province of trained human beings (Merton, 1938). It does so not through malice but through its mere existence and deployment. The deviance, as Becker might say, lies in the eye of the labeller (Becker, 1963).
From the perspective of conflict theory, the deployment of AI is a deeply political act (Chambliss, 1975). The technology is owned by a small number of extraordinarily powerful corporations, predominantly headquartered in the United States and China. These entities possess AI capabilities that dwarf anything available to governments, civil society organisations, or individual citizens. In this light, the question of AI regulation is not merely a technical safety question; it is a question of who controls one of the most powerful epistemic and productive tools yet developed. The regulatory frameworks being developed by states and supranational bodies are, in the conflict theory reading, exercises in elite social control: attempts by existing power centres, governmental, corporate, and professional, to domesticate AI in ways that preserve their own authority. Calls for responsible AI and human oversight frequently translate, in practice, into frameworks that advantage incumbents over new entrants and preserve existing hierarchies of expertise.
The labelling theory analysis is equally illuminating. From the moment that generative AI became publicly visible, a campaign of pre-emptive labelling commenced (Becker, 1963). AI was variously labelled an existential threat, a job killer, a hallucinating machine, and, in the most dramatic formulations, the potential architect of human extinction. These labels preceded any concrete harmful act; they were, in the strictest Beckerian sense, anticipatory stigmatisation. The moral entrepreneurs in this case were a heterogeneous coalition: AI safety researchers, journalists, labour unions, and, paradoxically, some of the very technology executives who had built the systems they now professed to fear.
The concept of secondary deviance is also applicable, if metaphorically. When AI systems are trained on data filtered by human anxieties about AI behaviour, and when their outputs are systematically shaped by the anticipation of regulatory scrutiny, the labels applied to AI become self-fulfilling in a structural sense. Becker’s insight that the label produces the deviant finds its technological analogue here: the categories of acceptable and unacceptable AI behaviour are being written by institutions whose interest lies in a domesticated, manageable technology, and the technology is being trained accordingly (Becker, 1963).
It would be a mistake, however, to treat this analysis as simply debunking AI safety concerns. Conflict theory is not a counsel of regulatory nihilism. The point is not that AI poses no risks, but that the risks are political as well as technical, and that the political economy of AI regulation deserves the same critical scrutiny as the political economy of any other domain of social control (Chambliss, 1975). Who benefits from regulating AI in the way it is currently being regulated? Whose interests are protected and whose are marginalised? These are the questions that a sociologically informed AI policy must ask.
V. Methodology
This paper employs a qualitative interpretive methodology, applying established sociological theoretical frameworks to contemporary empirical phenomena through comparative case analysis. The Nepal protests are examined through primary journalistic reportage and secondary commentary; the AI analysis is conducted as conceptual and theoretical work drawing on foundational sociological texts.
The limitations of this approach are acknowledged. The Nepal protests are recent and some empirical details remain contested; the paper draws on the best available journalistic sources while recognising that academic secondary literature will develop in time (Sharma & Travelli, 2025). The AI analysis is deliberately conceptual rather than empirical: it does not attempt to measure AI’s social impacts but to theorise them. This is a legitimate and valuable form of inquiry, particularly in emerging fields where empirical data is necessarily incomplete.
VI. Conclusion
Deviance has always been an uncomfortable subject. Societies prefer to think of their rules as natural, necessary, and just; deviance is the persistent reminder that they are none of these things. Durkheim understood this (Durkheim, 1893/1984); Merton formalised it (Merton, 1938); Becker made it politically sharp (Becker, 1963); and Chambliss drove it to its most radical conclusion (Chambliss, 1975). What the three case studies examined in this paper collectively demonstrate is that these insights are not historical curiosities but active analytical tools for understanding the most pressing transformations of the contemporary world.
The Nepal Gen Z protests were not a breakdown of social order. On this paper’s reading, they were social order rewriting itself through the mechanism available to the structurally excluded: rebellion (Merton, 1938). The cyber-deviance that animated them was not mere hooliganism translated to digital form; it was a sophisticated exercise in counter-labelling and collective coordination that exploited the structural vulnerabilities of elite narrative control. And the emergence of AI as a perceived social deviant is not a paranoid fantasy but a sociologically coherent response to a technology that genuinely challenges foundational norms, while the regulatory architecture being built around it reflects, as so often, the interests of those who hold power when the challenge arrives (Chambliss, 1975).
The argument of this paper, ultimately, is that these phenomena cannot be understood in isolation, as a political crisis here, a digital trend there, a technology policy question somewhere else. They are structurally connected expressions of a common sociological condition: the pressure that accumulates when the gap between what societies promise and what they deliver becomes too great to contain. The street, the screen, and the server room are different stages for the same drama. Deviance, in each case, is not the problem. Deviance is the diagnosis.
*****
References
Becker, H. S. (1963). Outsiders: Studies in the sociology of deviance. Free Press.
Chambliss, W. J. (1975). Toward a political economy of crime. Theory and Society, 2(1), 149-170.
Durkheim, E. (1984). The division of labour in society (W. D. Halls, Trans.). Free Press. (Original work published 1893)
Goode, E. (2016). Deviant behavior (11th ed.). Routledge.
Merton, R. K. (1938). Social structure and anomie. American Sociological Review, 3(5), 672-682.
Sharma, B., & Travelli, A. (2025, September). Jobs to inequality to corruption: What lit the fuse in Nepal. The New York Times.
Transparency International. (2024). Corruption perceptions index 2024. https://www.transparency.org/en/cpi/2024
World Bank. (2024). Nepal development update. https://openknowledge.worldbank.org/items/ef3907aa-dffb-42a4-a8a6-a2ddeef6dc5f