AI is unusual because it can operate as a tool of control, an object of fear, and a quasi-religion simultaneously and at global scale. But it is a poor substitute for religion...

Analytical briefing · August 2026, by Claude

The hypothesis that a shared threat or belief system can unify and help control large populations is well supported by social science — but the evidence overwhelmingly describes emergent, decentralized processes (in-group/out-group psychology, ritual solidarity, media-driven panics, and “rally-round-the-flag” dynamics), not a single coordinated cabal executing a master plan. Deliberate top-down manipulation is real, but historically it is clumsy, limited, and frequently backfires.

Dozens of threats have served, or been theorized to serve, this cohering and mobilizing function — pandemics, war, terrorism, economic collapse, immigration “invasion” narratives, nuclear and asteroid annihilation, crime waves, ethnic scapegoating, climate change, and even a hypothetical alien invasion (invoked by Ronald Reagan in 1987). A mature body of scholarship — moral panic theory, scapegoat theory, securitization theory, and the “politics of fear” — explains the mechanism without any recourse to conspiracy.

Artificial intelligence is unusual because it can operate as a tool of control, an object of fear, and a quasi-religion simultaneously and at global scale. But it is a poor substitute for religion as a source of meaning, community, ritual, and answers to mortality, and the fragmentation of the information environment it produces may reduce shared cohesion rather than increase it. Crucially, real global governance — of AI and of everything else — is fragmented, contested, and weak: the opposite of the omnipotent coordinating body that conspiracy narratives imagine.

1. Cohesion is mostly emergent, not engineered. From Durkheim’s “collective effervescence” to Tajfel’s minimal-group experiments, solidarity and in-group bias arise spontaneously from ritual and mere categorization — no planner required. Rulers have genuinely exploited these dynamics, but exploiting a pre-existing mechanism is very different from manufacturing one from scratch.

2. The “common enemy” effect is real but bounded. Sherif’s Robbers Cave experiment, realistic conflict theory, social identity theory, and the “rally-round-the-flag” effect all confirm that external threat boosts internal cohesion — yet the effect is typically short-lived, conditional, and sometimes weak or absent. Terror Management Theory adds a mortality dimension, but its canonical paradigm suffered a major replication failure (Many Labs 4, 2022) that should temper strong claims.

3. Many fears can play the role. There is a well-developed vocabulary for how threats are constructed and by whom: Cohen’s moral panics, Girard’s scapegoat mechanism, the Copenhagen School’s securitization theory, and Furedi’s and Robin’s work on the politics of fear.

4. AI is a triple-role phenomenon (control infrastructure, existential fear, belief system) but is emotionally and communally thin compared with religion, and its informational effects are centrifugal — pulling a shared reality apart rather than together.

5. Global governance is polycentric and weak. International-relations theory is explicit that “global governance” is governance without government — overlapping, fragmented spheres of authority, not a world state. AI governance efforts illustrate both the reality of coordination and its severe limits.

Émile Durkheim, in The Elementary Forms of Religious Life (1912), argued that religion is fundamentally social: collective rituals generate “collective effervescence,” an intense shared emotional energy that reaffirms group identity and produces social solidarity. The sacred totem is, in his reading, a symbol of the group itself; worshipping it is society worshipping itself. Crucially, Durkheim saw this as a spontaneous, emergent process — he identified the same dynamics in secular gatherings (the French Revolution, national ceremonies, sporting events, political rallies), not as the product of a controlling designer.

Karl Marx offered the control-oriented counterpoint. In the Introduction to A Contribution to the Critique of Hegel’s Philosophy of Right (written 1843–44, published 1844), he wrote that religion “is the opium of the people” — pacifying the oppressed and reflecting the economic conditions that produce it, but again as a systemic product of class society, not a conspiracy. Max Weber, in The Protestant Ethic and the Spirit of Capitalism (1904–05), reversed the causal arrow, arguing that religious ideas (Calvinist predestination, the worldly “calling”) can themselves drive economic and social change. Together these founders frame religion as cohesion (Durkheim), as legitimation and pacification (Marx), and as an autonomous causal force (Weber) — none requiring a single coordinating elite.

Henri Tajfel and John Turner’s social identity theory (1979), built on the “minimal group paradigm” (Tajfel et al., 1971), showed that merely categorizing people into arbitrary groups — even by a coin flip, or by a stated preference for the painters Klee versus Kandinsky — is sufficient to produce in-group favoritism and out-group derogation, with participants even sacrificing absolute gains to maximize the relative advantage of their group. Muzafer Sherif’s Robbers Cave experiment (summer 1954, with 22 eleven-year-old boys), the basis of realistic conflict theory, showed that competition over scarce resources rapidly manufactures intergroup hostility, and that “superordinate goals” requiring cooperation dissolve it. Notably, Sherif’s earlier attempts failed — in one, the two groups united against a common enemy; in another, they turned on the experimenters — a caution about how contingent these effects are.

Building on Ernest Becker’s Pulitzer-winning The Denial of Death (1973), Jeff Greenberg, Sheldon Solomon, and Tom Pyszczynski (1986) proposed that awareness of death generates anxiety managed by clinging to cultural worldviews and self-esteem (the “anxiety buffer”). Hundreds of “mortality salience” studies found that subtle death reminders increase worldview defense, in-group favoritism, out-group derogation, and support for charismatic leaders. Important caveat: TMT’s canonical worldview-defense paradigm failed to replicate in a large pre-registered study — Klein et al., “Many Labs 4” (2022), which pooled 21 labs and roughly 2,220 participants and was unable to reproduce the effect under any conditions. The core insight remains influential, but the strength of specific effects is genuinely contested.

The Roman imperial cult, instituted under Augustus, identified emperors with the divinely sanctioned authority of the Roman state; participation was a marker of loyalty and refusal could be treated as treason. Constantine’s embrace of Christianity (after his 312 victory at the Milvian Bridge) is widely read as at least partly a calculated strategy to use religious unity as a foundation for political cohesion. The medieval divine right of kings extended sacred legitimation to monarchs. In the modern era, wartime propaganda, the Cold War “threat,” and the political-science “rally-round-the-flag” effect (John Mueller, 1970) all show leaders benefiting from external threat. The clearest case: per Gallup, George W. Bush’s job approval stood at 51% just before 9/11, then jumped to a record 90% in a September 21–22, 2001 poll — the highest approval in Gallup’s history. But such effects are short-lived and conditional, and the related “diversionary theory of war” has only mixed empirical support.

The evidence strongly supports emergent, decentralized processes over deliberate top-down manufactured control by a coordinated elite. Cohesion mechanisms are properties of human psychology and social structure that operate with or without planners. Deliberate manipulation genuinely occurs — propaganda ministries, securitized “wars on X,” staged provocations — but it is opportunistic exploitation of pre-existing dynamics, is usually clumsy, and frequently fails or backfires. The conspiratorial premise of a single omniscient “powers that be” orchestrating global belief does not survive contact with the historical record, which is one of competing elites, unintended consequences, and bottom-up contagion.

The categories below have each historically unified, mobilized, or been used to discipline populations. The scholarly point is that any sufficiently vivid, externalized threat can be “securitized” into a cohering force.

• Pandemics and disease. The Black Death triggered scapegoating (Jewish communities falsely accused of poisoning wells); COVID-19 produced both extraordinary emergency powers and surges of national solidarity, alongside anti-Asian scapegoating.

• War and military enemies. The paradigmatic external-threat unifier (see the rally-round-the-flag effect).

• Terrorism / “war on terror.” Post-9/11 securitization justified sweeping domestic measures and a durable in-group/out-group frame.

• Economic collapse / scarcity. Financial crises and depressions have repeatedly been mobilized into political projects and out-group blame.

• Immigration / demographic “invasion” narratives. A recurrent out-group scapegoating frame; refugees and migrants function as modern “folk devils.”

• Nuclear annihilation. The Cold War’s organizing existential fear, cohering entire blocs.

• Asteroid / cosmic impact. A genuine (if low-probability) natural existential risk that also serves as a rhetorical unifier — “a threat to the whole species.”

• Ideological enemies. Communism, fascism, “the West” — each has served as the cohering “other.”

• Crime waves and moral panics. Stanley Cohen’s Folk Devils and Moral Panics (1972) showed how media and authorities amplify minor deviance (the Mods and Rockers) into perceived threats to social order.

• Ethnic, racial, and religious scapegoating. The most destructive form, theorized by Girard below.

• Climate change / environmental catastrophe. A real threat that also functions as a mobilizing, unifying civilizational frame.

• Existential / civilizational risk. Toby Ord’s The Precipice (2020) estimates roughly a 1-in-6 chance of existential catastrophe this century, overwhelmingly anthropogenic — his single largest specific risk being unaligned AI (about 1 in 10), ahead of engineered pandemics (about 1 in 30).

• Technology panics. The printing press, railways, telegraph, television, comic books, video games, and social media have each triggered waves of fear about social corruption.

• The hypothetical unifying external threat. Ronald Reagan, addressing the UN General Assembly on September 21, 1987, mused that humanity’s differences “would vanish” in the face of “an alien threat from outside this world.” The same device drives the fiction of Watchmen (a manufactured attack staged to unify humanity).

Four bodies of theory explain this without conspiracy:

• Moral panic theory (Cohen, 1972; later Goode & Ben-Yehuda): media and “moral entrepreneurs” construct “folk devils,” and panic follows a predictable arc of identification, amplification, escalating anxiety, policy response, and eventual fading.

• Scapegoat theory (René Girard, Violence and the Sacred, 1972): mimetic rivalry escalates toward a “war of all against all,” resolved by collective violence against an arbitrary victim — the “all against one” that restores unity and, Girard argues, founds ritual, myth, and religion itself.

• Securitization theory (Copenhagen School — Buzan, Wæver, de Wilde, 1998): security is a “speech act.” An issue becomes an existential threat not because it objectively is one, but because a securitizing actor successfully frames it as one to an audience that accepts the framing — licensing extraordinary measures beyond normal politics. This is the single most useful lens on how threats are constructed and by whom.

• The politics of fear (Frank Furedi; Corey Robin, Fear: The History of a Political Idea, 2004): fear as an organizing principle of public life and an instrument of governance. Robin argues fear becomes “the justifying language of public life” as faith in positive political principles recedes.

AI is unusual in being able to occupy three roles at once:

1. Tool of control. AI powers surveillance, predictive policing, and algorithmic governance. Shoshana Zuboff’s The Age of Surveillance Capitalism (2019) describes how firms claim private experience as free raw material, render it into behavioral-prediction products, and increasingly intervene to shape behavior — an “instrumentarian power” she calls “Big Other.” Important corrective: China’s “social credit system,” the usual exhibit for AI-driven population control, is (per MERICS and Cambridge research) highly fragmented, lowly digitalized, focused mainly on businesses, and has no single omniscient nationwide personal score. The dystopian “master algorithm” is largely a Western projection.

2. Object of fear. Existential AI risk is now a mainstream elite concern. The Center for AI Safety’s one-sentence “Statement on AI Risk” (May 30, 2023) reads: “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.” It was signed by hundreds of researchers, executives, and public figures, including Turing laureates Geoffrey Hinton and Yoshua Bengio and the CEOs of OpenAI, Google DeepMind, and Anthropic.

3. Belief system. Anthony Levandowski, the former Google and Uber engineer, founded “Way of the Future,” a religious nonprofit dedicated to “the realization and worship of a Godhead based on Artificial Intelligence.” Techno-optimism, transhumanism, “dataism,” and internet-native eschatologies such as Roko’s Basilisk function as a digital-age theology.

Mass personalization of persuasion, recommendation algorithms that shape the information environment, deepfakes and synthetic media, and behavioral prediction give AI reach and granularity no prior technology possessed. It scales globally and personalizes simultaneously — the strongest argument for its uniqueness.

Religion provides meaning, community, ritual, moral regulation, and answers to mortality that AI does not. AI lacks the emotional and communal dimensions; “AI as god” churches have attracted at most a few thousand adherents and one folded entirely within a few years. Most importantly, AI’s dominant effect on the information environment is centrifugal: it fragments a shared reality into personalized streams, which tends to reduce the common focus that Durkheimian cohesion requires. And AI-as-threat is abstract, contested, and temporally diffuse — unlike a bombing or an invasion, it offers no single dramatic “rally point.” Yuval Noah Harari, in Homo Deus (2016), frames the contest as one of narratives: he argues a new techno-religion, “Dataism,” could displace humanism — but presents it as a worrying possibility rather than a functioning faith, and critics note it offers little of the meaning, morality, or community that sustains actual religions.

AI is best understood as an unusually versatile infrastructure — simultaneously control apparatus, fear object, and faith object — rather than a superior replacement for religion’s cohering function. Its comparative advantage is influence and control at scale; its comparative weakness is meaning-making and communal solidarity. It is far more likely to be wielded within existing cohesion and fear dynamics than to become a unifying worldview in its own right.

International-relations scholars are emphatic that global governance is not world government. Rosenau and Czempiel’s Governance Without Government (1992) established the field’s core claim: order among states emerges from overlapping “spheres of authority,” networks, norms, and regimes — not centralized hierarchy. Robert Keohane’s institutionalism (After Hegemony, 1984) similarly explains cooperation through regimes and mutual interest, not a central planner. Recent network analyses describe the system as simultaneously fragmented, polycentric, and complex — organized, but without a controlling center.

These are the real, stated objectives of actual global governance structures:

• Ending poverty and advancing development — the UN’s 17 Sustainable Development Goals (the 2030 Agenda, adopted September 2015 with 169 targets), spanning poverty, hunger, health, education, gender equality, clean water and energy, decent work, reduced inequality, sustainable cities, climate action, life below water and on land, peace and justice, and global partnership.

• Global public goods provision — pandemic prevention and response (WHO), climate coordination (UNFCCC, Paris Agreement), and nuclear non-proliferation (NPT/IAEA).

• Peace and conflict prevention; economic stability; human rights; managing the global commons — the oceans, the atmosphere, and outer space.

• Preventing catastrophic and existential risks and coordinating on transnational problems — migration, finance, health, and AI governance itself.

The Bletchley Declaration was signed on November 1, 2023, by 28 countries plus the EU at the UK AI Safety Summit — the first intergovernmental statement to formally acknowledge potentially catastrophic frontier-AI risks and commit signatories (including both the US and China) to cooperation. It was followed by summits in Seoul (2024), Paris (2025, notably rebranded from “AI Safety Summit” to “AI Action Summit,” shifting emphasis toward innovation), and India (2026). The EU AI Act (Regulation (EU) 2024/1689) is the first comprehensive, risk-based AI law; the US pursued executive action (Executive Order 14110, October 2023, later rescinded in January 2025). These efforts are real but partial, largely non-binding in key respects, and already fracturing between “safety” and “innovation” framings.

The striking feature of actual global governance is its weakness: it is fragmented, under-resourced, contested by great-power rivalry and sovereignty concerns, and frequently unable to enforce its own commitments (the SDGs are officially off-track at their midpoint; the WHO’s pandemic authority was sharply limited; AI governance is a patchwork of voluntary declarations). The empirical record is the opposite of the coherent, all-powerful coordinating body that conspiracy narratives posit. Put bluntly: if a secret cabal were running the world, it would be doing a conspicuously poor job of it. The honest description is many hands, competing interests, and chronic coordination failure — governance without government, not government in the shadows.

• For readers evaluating “single coordinated plan” claims: apply a simple test — does the claim require a level of centralized coherence and enforcement capacity that real institutions demonstrably lack? If yes, prefer the emergent, decentralized explanation. What would change this assessment is direct primary evidence of a single body successfully directing global belief or policy across rival states — which the record does not provide.

• For analysts of AI: track AI primarily as control infrastructure and information-environment shaper, not as an emerging unifying religion. The threshold that would revise this is evidence of AI-based movements attaining mass (millions-scale), ritualized, community-forming adherence — currently absent.

• For those worried about the politics of fear: use securitization theory as a diagnostic. When any actor frames an issue as an existential threat licensing extraordinary measures, ask who benefits, what normal checks are being suspended, and whether the threat’s magnitude is established by evidence or by assertion.

• For policymakers on AI governance: the divergence across the Bletchley → Seoul → Paris → India summit series is the central risk. Pursue binding, verifiable model-testing and incident-reporting regimes through existing bodies; the benchmark for real progress is enforceable cross-border commitments, not further declarations.

• Replication concerns. Terror Management Theory’s canonical mortality-salience paradigm failed a major pre-registered replication (Many Labs 4, 2022); Sherif’s Robbers Cave study has been criticized for artificiality and for undisclosed prior failed attempts; the minimal-group paradigm is criticized for ecological validity. Treat these as suggestive frameworks, not settled laws.

• Contested effect sizes. The rally-round-the-flag effect is real but short-lived and, in several studies, weak; the diversionary-war thesis has mixed support.

• Popularization risk. Harari’s “Dataism,” popular accounts of surveillance capitalism, and especially depictions of China’s social credit system are influential but contested; where relied upon here, scholarly pushback has been flagged.

• Subjective estimates. The AI existential-risk figure (about 1 in 10) is Toby Ord’s own subjective estimate from The Precipice, reported as such, not as an objective measurement.

• Normative neutrality. Describing a mechanism (securitization, scapegoating, moral panic, fear politics) is not endorsing its use; several mechanisms discussed here are frequently destructive, and naming them as objects of academic study is precisely the point.

All sources consulted, grouped by theme, with associated URLs.

[1] Émile Durkheim, The Elementary Forms of Religious Life (1912).https://en.wikipedia.org/wiki/The_Elementary_Forms_of_Religious_Life

[2] Karl Marx, A Contribution to the Critique of Hegel’s Philosophy of Right, Introduction (1844).https://www.marxists.org/archive/marx/works/1843/critique-hpr/intro.htm

[3] Max Weber, The Protestant Ethic and the Spirit of Capitalism (1904–05).https://en.wikipedia.org/wiki/The_Protestant_Ethic_and_the_Spirit_of_Capitalism

[4] Henri Tajfel & John Turner, “An Integrative Theory of Intergroup Conflict” (1979); Tajfel et al., minimal-group experiments (1971).https://en.wikipedia.org/wiki/Social_identity_theory · https://en.wikipedia.org/wiki/Minimal_group_paradigm

[5] Muzafer Sherif, the Robbers Cave experiment (1954) and realistic conflict theory.https://en.wikipedia.org/wiki/Robbers_Cave_experiment · https://en.wikipedia.org/wiki/Realistic_conflict_theory

[6] Ernest Becker, The Denial of Death (1973). https://en.wikipedia.org/wiki/The_Denial_of_Death

[7] Terror Management Theory (Greenberg, Solomon & Pyszczynski, 1986; Rosenblatt et al., 1989).https://en.wikipedia.org/wiki/Terror_management_theory

[8] Richard A. Klein et al., “Many Labs 4: Failure to Replicate Mortality Salience Effect With and Without Original Author Involvement,” Collabra: Psychology 8(1):35271 (2022). https://doi.org/10.1525/collabra.35271 · https://online.ucpress.edu/collabra/article/8/1/35271/168050/Many-Labs-4-Failure-to-Replicate-Mortality

[9] John E. Mueller, “Presidential Popularity from Truman to Johnson” (1970); the “rally-round-the-flag” effect. https://en.wikipedia.org/wiki/Rally_%27round_the_flag_effect

[10] Gallup, Presidential Approval Ratings — George W. Bush (90% approval, Sept 21–22, 2001 poll). https://news.gallup.com

[11] Imperial cult of ancient Rome. https://en.wikipedia.org/wiki/Imperial_cult_of_ancient_Rome

[12] Constantine the Great and Christianity. https://en.wikipedia.org/wiki/Constantine_the_Great_and_Christianity

[13] Divine right of kings. https://en.wikipedia.org/wiki/Divine_right_of_kings

[14] Stanley Cohen, Folk Devils and Moral Panics (1972); moral panic theory.https://en.wikipedia.org/wiki/Folk_Devils_and_Moral_Panics · https://en.wikipedia.org/wiki/Moral_panic

[15] René Girard, Violence and the Sacred (1972); the scapegoat mechanism. https://en.wikipedia.org/wiki/Violence_and_the_Sacred · https://en.wikipedia.org/wiki/Ren%C3%A9_Girard

[16] Barry Buzan, Ole Wæver & Jaap de Wilde, Security: A New Framework for Analysis (1998); securitization theory (Copenhagen School). https://en.wikipedia.org/wiki/Securitization_(international_relations)

[17] Frank Furedi, Culture of Fear / How Fear Works. https://en.wikipedia.org/wiki/Culture_of_fear

[18] Corey Robin, Fear: The History of a Political Idea (Oxford University Press, 2004). https://www.goodreads.com/book/show/364629.Fear

[19] Toby Ord, The Precipice: Existential Risk and the Future of Humanity (2020). https://theprecipice.com/

[20] Ronald Reagan, Address to the 42nd Session of the UN General Assembly (Sept 21, 1987).https://www.reaganlibrary.gov/archives/speech/address-42d-session-united-nations-general-assembly-new-york-new-york

[21] Shoshana Zuboff, The Age of Surveillance Capitalism (2019); “instrumentarian power” / “Big Other.”https://news.harvard.edu/gazette/story/2019/03/harvard-professor-says-surveillance-capitalism-is-undermining-democracy/

[22] On behavioral prediction and modification: “How to ‘Improve’ Prediction Using Behavior Modification” (arXiv).https://arxiv.org/abs/2008.12138

[23] China’s Social Credit System (the fragmented reality; MERICS and Cambridge research).https://en.wikipedia.org/wiki/Social_Credit_System

[24] Center for AI Safety, “Statement on AI Risk” (May 30, 2023). https://safe.ai/work/press-release-ai-risk ·
https://aistatement.com/

[25] Anthony Levandowski’s “Way of the Future”; AI-as-religion. https://www.thebrink.me/the-rise-of-ai-religion/ · https://en.wikipedia.org/wiki/Anthony_Levandowski

[26] Yuval Noah Harari, Homo Deus: A Brief History of Tomorrow (2016); “Dataism.”https://en.wikipedia.org/wiki/Homo_Deus:_A_Brief_History_of_Tomorrow · https://en.wikipedia.org/wiki/Dataism

[27] James N. Rosenau & Ernst-Otto Czempiel, Governance Without Government: Order and Change in World Politics (1992).https://en.wikipedia.org/wiki/Governance_Without_Government

[28] Robert O. Keohane, After Hegemony (1984); “Governance in a Partially Globalized World.”https://en.wikipedia.org/wiki/After_Hegemony

[29] United Nations, Transforming Our World: The 2030 Agenda for Sustainable Development (Res. A/RES/70/1, 2015); the 17 SDGs.https://sdgs.un.org/2030agenda · https://sdgs.un.org/goals

[30] The Bletchley Declaration by Countries Attending the AI Safety Summit (1–2 November 2023).https://www.gov.uk/government/publications/ai-safety-summit-2023-the-bletchley-declaration/the-bletchley-declaration-by-countries-attending-the-ai-safety-summit-1-2-november-2023

[31] Regulation (EU) 2024/1689 (the EU Artificial Intelligence Act). https://eur-lex.europa.eu/eli/reg/2024/1689/oj

[32] Executive Order 14110, “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence” (Oct 30, 2023).https://www.federalregister.gov/documents/2023/11/01/2023-24283/safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence