Office of Strategic Risk Assessment

Civilizational Risk Assessment

The Department assesses how the systems that sustain modern civilization depend on one another, how disruptions propagate between them, and how likely those systems are to remain stable over the coming decade.

The Civilizational Risk Assessment is the Department’s continuing assessment of systemic stress across the critical systems on which modern societies depend: energy, food, water, finance and trade, supply chains, communications, computation, public health, governance and population. It is conducted by the Office of Strategic Risk Assessment under the Integrated Systems Modeling Program, with contributions from the Office of Advanced Artificial Intelligence, the Office of Long-Term Systems Research and the Continuity Infrastructure Administration.

The current assessment is set out in Civilizational Systems Outlook: 2026–2035 (DSC-TR-2026-014), released on 15 October 2026. It is the first edition of an annual series. Its central finding is that, in 71 percent of weighted model runs, the Department’s Composite Systemic Stress Index enters a sustained breakdown regime before 1 January 2035. The Office assesses this outcome as likely, with moderate confidence.

This page summarizes the finding, explains how it was produced and describes how it should and should not be read. The full report, including its methods, assumptions, review record and limitations, is the authoritative statement of the assessment.

Flagship report · Technical Report

Civilizational Systems Outlook: 2026–2035

Report number: DSC-TR-2026-014
Issuing division: Office of Strategic Risk Assessment
Publication date: 15 October 2026

Read the full report

Key findings

The Composite Systemic Stress Index (CSSI), a 0–100 measure of concurrent stress across eleven modeled systems, is estimated at 54 for 2026, compared with reconstructed values of approximately 31 in 2000, 38 in 2010 and 46 in 2020.
In 71 percent of weighted ensemble runs (90 percent interval 58–82 percent), the CSSI enters the systemic breakdown regime, a CSSI of 70 or above sustained for 24 months, before 1 January 2035. The median entry year is 2033.
The principal drivers are tight coupling between automated infrastructure control and energy systems, synchronized food-system shocks, supply-chain concentration in semiconductors and critical minerals, financial contagion and erosion of institutional capacity. Climate change acts as a stress multiplier rather than a sole driver.
About a third of the variation in outcomes arises only from interactions among drivers. The risk is a property of the coupled system, not of any single sector.
Model runs that avoid the breakdown regime share identifiable conditions: retained non-automated fallback capacity in critical control systems, deeper inventories of critical inputs, and institutional capacity that holds under concurrent stress.

How the assessment was produced

The assessment was produced with version 4.2 of the Integrated Civilizational Systems Model (ICSM), a coupled simulation maintained by OSRA. The model represents eleven systems (Energy; Food and Agriculture; Water; Climate and Environment; Finance and Trade; Supply Chain and Manufacturing; Information and Communications; Computational Systems and AI; Public Health; Governance and Social Cohesion; and Demography) across 24 world regions. The systems exchange information at monthly intervals, so that a disturbance in one can propagate to others through explicitly represented dependencies. The model’s architecture, calibration and validation are documented in DSC-TR-2026-003.

Because the future conditions that matter most cannot be known in advance, the model was not run once but 40,000 times. The runs span 1,280 scenario configurations, each a combination of assumptions about five external drivers: the pace of automation in infrastructure control, the climate hazard pathway, the trade and geopolitical regime, the trajectory of the energy system and the background trend in institutional capacity. Within each configuration, uncertain parameters and the timing of shocks were varied at random. The ensemble was executed on the ALPHA-1 system at Continuity Research Facility Alpha between 24 August and 18 September 2026.

Summary of the 2026 assessment
Element	Value
Model	ICSM v4.2, 11 coupled modules, 24 regions
Ensemble	40,000 runs across 1,280 scenario configurations
Headline metric	Composite Systemic Stress Index (CSSI), 0–100
2026 nowcast	CSSI ≈ 54
Breakdown regime	CSSI ≥ 70 for 24 consecutive months
Entry before 1 January 2035	71% of weighted runs (90% interval 58–82%)
Median entry year	2033
Independent review	Model Review Panel; two dissenting opinions recorded

Before the projection was run, the model was tested against historical periods, including the financial crisis of 2008–2009 and the disruptions of 2020–2022, to establish whether it reproduces how stress spreads between systems and how systems recover. It does so with measurable skill, while tending to underestimate peak stress. The methods used to handle assumptions that cannot be assigned reliable probabilities are described in DSC-WP-2026-012. A full account of the methodology is given in the methodology section of the Outlook.

Interpreting the findings

What the breakdown regime means

The systemic breakdown regime is a defined state of the model: a Composite Systemic Stress Index of 70 or above, sustained for at least 24 consecutive months. The threshold was set so that sustained exceedance corresponds, in the model, to severe civilizational breakdown, which the Department defines as a sustained, multi-region loss of capacity to maintain critical infrastructure, food distribution and governance at 2020 levels.

The definition concerns capacity. It describes a condition in which, across several regions at once and for an extended period, systems can no longer deliver what they delivered in 2020, and in which recovery from further disturbances is slow and incomplete. The 24-month requirement distinguishes this condition from the short, sharp crises of recent history, from which systems recovered.

What the finding does not mean

It is not a projection of human extinction. The definition of breakdown concerns the capacity of systems, not the survival of populations.
It is not a prediction of certainty. The finding is that the model enters the breakdown regime in 71 percent of its runs under documented assumptions. In 29 percent of runs it does not.
It does not identify a date. The median entry year of 2033 summarizes a distribution; individual runs enter in different years, and many do not enter at all within the horizon.
It does not describe a single event. In the model, entry into the breakdown regime most often follows a sequence of incompletely resolved disturbances rather than one catastrophe.
It does not apply uniformly. Stress differs substantially among types of regional system, and the assessment does not rank or forecast outcomes for any country.
It is not a statement that the outcome is unavoidable. Identifying the conditions under which model runs avoid the breakdown regime is one of the assessment’s main purposes.

How to read the probability

The 71 percent figure is the weighted fraction of model runs in which the outcome occurs. It is a property of the model under its assumptions, not a measured frequency of real-world events. The Office’s confidence that the model’s behavior is informative about the world is stated separately, as moderate: the model has been validated against historical periods of stress, but no historical period contains conditions of breakdown-regime severity at global scale against which it could be tested. The 90 percent interval of 58–82 percent reflects uncertainty in the weighting of scenarios; uncertainty in the structure of the model itself is examined through alternative model variants and independent review, and is discussed in the uncertainty and limitations section of the report.

Model-based assessment

The Civilizational Risk Assessment is a research product. It states what the Department’s models indicate under documented assumptions, with explicit uncertainty. It is not a forecast of specific events and contains no operational guidance.

Frequently asked questions

Is the Department predicting the end of humanity?

No. The assessment concerns the capacity of critical systems, not the survival of the human species. Severe civilizational breakdown is defined as a sustained, multi-region loss of capacity to maintain critical infrastructure, food distribution and governance at 2020 levels. That is a serious condition, and its consequences for people would be serious, but it is a different and far narrower claim than extinction. The model does not represent extinction outcomes and the report makes no finding about them.

Why 2035?

The Outlook series uses a ten-year horizon. Ten years is long enough for slowly changing factors, such as institutional capacity, infrastructure stock and population structure, to change materially, and short enough that the model can still be tested against historical periods of comparable length. The date 1 January 2035 is the end of that horizon for the 2026 edition; it does not mark an expected event. Results are reported for each year in the full report, and the cumulative probability of entry rises gradually across the period rather than at a single point.

Has the model been independently reviewed?

Yes. ICSM v4.2 and the draft results were reviewed by the Model Review Panel, the Department’s independent body for review of the model, whose members are external to OSRA. The Panel accepted the model for use in the Outlook. Two members recorded dissenting opinions on the calibration of the Governance and Social Cohesion module, one indicating that the central result may be too high and the other that it may be too low. Under their proposed alternatives, the probability of entry before 2035 is 61 percent and 78 percent respectively. Both dissents and the Panel’s findings are summarized in Appendix D of the report.

How confident is the Department in the result?

The Office assesses the result with moderate confidence. The model reproduces the propagation of stress in historical crises with measurable skill, its main sensitivities are documented, and the result holds, with varying magnitude, across alternative model versions, definitions and weighting schemes. Confidence is not high because the model has not been, and cannot yet be, tested against conditions as severe as those it projects, and because the module representing governance and institutional capacity is the least well constrained by data.

How does this relate to the Neural Continuity Research Program?

The Department’s mission includes research into the preservation of human knowledge and, as a long-term and explicitly aspirational objective, of human cognition. The Neural Continuity Research Program investigates whether the latter could eventually be achieved; whether it can is not known, and no such capability exists. The Outlook bears on that program only indirectly: its findings lengthen the period over which the Department assumes that preservation infrastructure may need to operate without dependable external support. The report recommends that the program continue on its chartered schedule, subject to ethics and scientific review, and does not recommend accelerating it.

What is the Department doing in response to the findings?

The report makes ten recommendations, addressed principally to the Department’s own research program. They include strengthening the evidence base for the model’s governance module, expanding the evaluation of automated systems used in critical-infrastructure control, research on control systems that degrade gracefully and retain manual fallback, quarterly publication of the stress index with early-warning indicators, and continued investment in long-duration information preservation. The Department is a research organization; the recommendations concern research and institutional priorities rather than operational measures.

Will the assessment be updated?

Yes. The Civilizational Systems Outlook is an annual series. The next edition is scheduled for October 2027 and will use an updated version of the model, incorporate new data and the results of the calibration work recommended in this edition, and record any change to the central estimate together with its causes. Earlier editions will remain available unchanged in the publications library.

Where can I find the data?

The full report, DSC-TR-2026-014, contains the ensemble design, the key assumptions and their ranges, annual cumulative probabilities, driver attributions, results under alternative model variants and the scenario configuration definitions. The model is documented in DSC-TR-2026-003, and the dependency mapping on which it draws in DSC-TR-2026-010. Enquiries about the assessment may be made through the Department’s contact page.

Supporting research

DSC-TR-2026-003Technical Report28 April 2026