05 — The Automated Police State

Full report

On this page

Administrative exclusion, wrongful classification and the coercive state

Research status — original analytical scenario study, 8 October 2026. This is an evidence-informed exploration of a deliberately extreme fictional world, not a prediction. Contemporary studies of Machine Intelligence generally describe job exposure and transformation, not verified permanent elimination of most employment. Historical facts cited below apply to their original periods; all future events, dialogue, household cases and political institutions in the scenario are invented. [S01] [S02] [S03]

Locked scenario: Machine Intelligence permanently destroys most wage employment across occupations; ordinary people suffer a worldwide depression at least as socially devastating as the 1930s; production remains partly advanced but access is unequal; governments, companies, and affluent owners wield surveillance and non-nuclear enforcement capabilities; some peaceful civilian attempts to obtain necessities are wrongly classified as threats. Governments do not all fail at once. No nuclear weapons are used, threatened, deployed, or detonated. This fiction does not presume every authority is malicious or every destitute person peaceful.

Analytical method: Separate (a) grounded historical evidence, (b) conditional mechanisms given the scenario, (c) high-uncertainty extrapolations, and (d) explicitly invented illustrative vignettes. Treat labels such as terrorism, insurgency, and disorder as contested classifications to be evaluated case by case, not inherently accurate descriptions. Food access, legal protection and medical care are humanitarian needs, not proof of political allegiance.

Executive finding: a functioning database is not a functioning society

An authoritarian crisis state need not be technologically omniscient. The more plausible system is fragmented: legacy databases, contracts, neighborhood cameras, identification checks, private service portals and human administrators with overlapping mandates. Individually these systems promise efficiency, fraud control or infrastructure protection. Together they can produce a dense lattice of exclusion. A person marked 'unverified' in one registry can lose transit access, a housing appointment and a food allocation without any official deciding to starve them.

Australia's Robodebt Royal Commission is a real example of serious harm from unlawful automated public administration; it is not evidence that the novel's global system exists. It does demonstrate why legal authority, accurate data, accountable design and meaningful review matter. [S11]

The construction of emergency normality

Stage one: information consolidation. Ministries merge benefit eligibility, residency, financial and service records, officially to reduce fraud and duplicate payments. Data collection expands under emergency authority. People with incomplete paperwork are already overrepresented among the homeless and displaced.

Stage two: classification creep. A rule designed to detect organized attacks on infrastructure begins flagging frequent presence near distribution facilities. The system treats repeated appeals as suspicious repetition. Unusual behavior is common during famine; 'unusual' does not equal dangerous. Officials lacking time may trust a Machine Intelligence score without understanding its error rate.

Stage three: outsourced enforcement. Service operators and contractors apply exclusion criteria supplied by multiple agencies. The technology provider points to the government; the government points to the provider; the front-desk worker lacks authority to override either. Formal due process survives on paper while meaningful review takes longer than the ration cycle.

Stage four: civic silencing. Residents refrain from attending public assemblies because they fear losing benefits or travel permissions. Mutual aid groups avoid recording requests, undermining coordination. The state then interprets absence of formal applications as evidence that needs have declined.

A contested word: terrorism

The book's governments are allowed to use the label terrorist improperly, but the narration should not flatten legal and moral distinctions. Hunger is not terrorism. Peaceful assembly is not terrorism. Theft and assault remain potentially criminal; organized acts intended to terrorize civilians are categorically different from seeking necessities. Authorities may intentionally blur distinctions because the label simplifies detention and deflects criticism. Different courts, police units and journalists should contest that framing.

The moral agency of security workers

Security operators still have families, doubts and discretion. A technician may notice a model repeatedly misclassifying people from a temporary shelter. A supervisor may demand a better audit; another may hide the data because performance targets reward fewer 'incidents.' Some officers fear crowd violence; others recognize their assignment is to deny food to nonviolent families. Dissent is risky where wages and housing depend on institutional loyalty. Mechanical competence in monitoring equipment does not settle the ethical choice.

The ICRC has emphasized the humanitarian and legal risks of automated weapons and the need for meaningful human control and limitations. Treat this as a normative and legal concern, not proof that every form of Machine Intelligence-enabled security is equally lethal or uniformly deployed. [S12]

Fictional administrative case file: 52-118D

Applicant: Saira Malik, previously employed as a school administrator; two children. Trigger: Repeated entry attempts at a public distribution site following inconsistent hours. Record: One registration lists a demolished address; another lists temporary shelter. Machine Intelligence flag: 'Potential organized disruption—review pending.' Human consequence: The applicant's food appointment is deferred while a separate housing system refuses to renew the family placement without active eligibility. Appeal: Offered within forty-five days. Inventory cycle: Seven days. Finding: No documented threat or violence; data mismatch and repeated lawful attempts to access food explain observed behavior. Outcome: The flag is eventually removed, but the family has already moved, missed medical appointments and lost access to the original service district.

The administrative record would describe this as a small correction. For the family it was an irreversible collapse of security. This is a stronger dystopian device than an all-knowing computer that instantly understands every citizen.

Three state models

Constitutional emergency republic. Courts, advocates and an independent press still exist but function slowly. The immediate danger is suspension before review. A minority of cases receive corrective judgments, proving the law has not vanished even when it is inaccessible to many.

Corporate-authoritarian district. Employment, residence and food access are governed by contracts that resemble citizenship in practice. The private operator claims residents voluntarily accepted the terms, ignoring the scarcity of alternatives. No independent appeal exists.

Fragmented local authority. Databases do not synchronize. A person is considered eligible on one side of a bridge and suspect on the other. Some local officers cooperate with humanitarian groups; others charge arbitrary fees. Administrative unpredictability becomes a form of fear.

Why repression can destabilize its own institutions

Classification errors generate false positives. People withdraw information, making records worse. Contractors have incentives to exaggerate 'threat events' to retain funding. Officials spend more defending eligibility decisions than distributing goods. Families of security staff encounter the same exclusions. Surveillance can increase immediate control while reducing accurate knowledge and legitimate cooperation. A society may be extensively watched and increasingly misunderstood.

Civilian rights and protection, not evasion

A realistic nonviolent protection response includes community legal assistance, accurate records, independent audits, clear appeal routes, transparent welfare criteria, humanitarian exemptions, and trusted intermediaries. These are institutional design features rather than guaranteed protections. Their failure should be shown as a sequence of avoidable decisions, not a mystical property of Machine Intelligence.

Confidence: High that automated administrative error can cause harm absent safeguards; medium that prolonged emergency rule expands coercive classification; low that any state achieves universal surveillance without infrastructure failure, error or dissent. Alternative: enforceable independent review and social-protection guarantees prevent the security bureaucracy from determining basic subsistence.

Cross-sector chronology (conditional scenario, not forecast)

Shared system dynamics that must remain consistent

The economic access loop runs from job losses to reduced household purchasing power, then to failed small businesses, falling local revenues, contraction of services, worsening health and still lower practical employability. The debt loop runs from unemployment to arrears and repossession, lender losses, restricted credit and greater difficulty keeping housing. The coercion loop begins when requests for relief are processed as threats: restrictive classification reduces cooperation and trust; lower trust increases enforcement expense; enforcement expense consumes resources once available for relief. These are scenario mechanisms, not measured elasticities or guaranteed sequences.

There are four hard constraints on every fictional institution. First, electricity and clean water require skilled maintenance, physical equipment, financing and spare parts; Machine Intelligence does not abolish these material demands. Second, property ownership is enforceable only where institutions or security organizations can implement it. Third, a person may have the knowledge to perform a vital service but lack medicine, pumps, transport, legal authority or fuel. Fourth, local adaptation is limited by population density, climate, seasonality and trade: a city of millions cannot simply turn itself into a self-feeding village. Power-system reliability research and humanitarian standards underline these dependencies. [S14] [S15]

The scenario contains at least four political equilibria rather than one inevitable outcome: negotiated redistribution where institutions remain capable and accountable; selective rationing where functioning production coexists with politically unequal access; authoritarian persistence where security institutions outlast trust and legitimacy; and fragmented sovereignty where local authorities, contractors and community organizations compete to provide governance. The OECD's multidimensional approach to fragility is a useful guard against treating bank failure, famine, and military collapse as a single event. [S10]

Claims, uncertainty and falsification

Civilian-resilience lens

For ordinary people, evaluate decisions by how they preserve water safety, calories, shelter, health continuity, dependents, legal standing and trusted relationships rather than by dramatic assertions of self-sufficiency. Effective community aid needs transparent distribution, safeguards against exploitation, coordination with legitimate medical and humanitarian providers when available, and physical stocks that can actually be replenished. Even successful cooperation cannot create insulin, sewage plants or replacement electrical equipment out of goodwill. Sphere standards provide a useful baseline for realistic humanitarian requirements, not a promise that those requirements can be met. [S14]

Ethical and legal boundary

Analysis of rationing, detention, civilian classification, protest, security technologies and state failure is institutional and humanitarian. It does not offer tactics for attacking people, penetrating security infrastructure, sabotaging services, stealing supplies or directing coercive force. Where armed conflict exists, applicable humanitarian law and protections against starvation are relevant; ordinary peacetime poverty and policing require separate legal analysis. [S13] [S17]

Research discipline and alternative outcomes

Do not confuse a necessary premise for a predicted outcome. Permanent displacement of most employment is imposed for narrative analysis even though current credible estimates do not demonstrate it will occur. The strongest counterfactual is not a technological miracle but a policy choice: inclusive dividends from productive capital, public services, accessible housing, debt restructuring, and protected civic organization could redirect productivity gains. Such measures could fail through insufficient fiscal reach, corruption, political conflict or real shortages—but their possibility should remain visible to sharpen the story's moral causality.

Historical analogies are partial. The Great Depression documents unemployment and banking failure but not Machine Intelligence-controlled production. Famine research documents loss of food entitlements but not a universal network of automated warehouses. Contemporary refugee and water-access statistics reveal the practical scale of humanitarian challenges but are not projections of the novel's future. [S04] [S07] [S08] [S09] Differentiate baseline figures, scenario assumptions, and fictional illustrative numbers wherever quantities appear.

Questions for independent-agent comparison

1. Is a classification grounded in a specific act or merely poverty?
2. Who can correct a wrong decision before harm occurs?
3. What physical and institutional constraints limit surveillance?

References and source-use notes

• [S05] ILO, World Social Protection Report 2024–2026. In 2024 reporting, 52.4% had at least one social protection benefit and 47.6% did not; coverage says nothing by itself about adequacy. URL: https://www.ilo.org/resource/article/world-social-protection-report-2024-26-figures
• [S10] OECD, States of Fragility 2025. Fragility is multidimensional; authority, services and security do not decline together. URL: https://www.oecd.org/en/publications/states-of-fragility-2025_81982370-en.html
• [S11] Royal Commission into the Robodebt Scheme, report (2023). Case study of unlawful automated welfare-debt administration and accountability failure. URL: https://www.royalcommission.gov.au/robodebt/report
• [S12] International Committee of the Red Cross, autonomous weapons position. Concerns about civilian harm, predictability, accountability and human control. URL: https://www.icrc.org/en/document/icrc-position-autonomous-weapon-systems
• [S13] International Committee of the Red Cross, food security and the law. Starvation of civilians as a method of warfare is prohibited during armed conflict; does not cover all peacetime deprivation. URL: https://www.icrc.org/en/law-and-policy/food-security
• [S14] Sphere, Humanitarian Handbook and standards. Minimum humanitarian standards for water, sanitation, food, settlement and health. URL: https://spherestandards.org/handbook/
• [S17] International Committee of the Red Cross, humanitarian access and law. Legal duties in armed conflict must not be applied uncritically to every peacetime security emergency. URL: https://www.icrc.org/en/document/humanitarian-access-what-law-says