This article examines how generative AI (GenAI) is reshaping strategic crisis decision-making through the emergence of ‘synthetic foresight’ – the algorithmic simulation of adversary intentions, escalation pathways, and imagined futures under conditions of uncertainty. Unlike traditional practices such as early warning, scenario planning, wargaming, or red teaming, which discipline strategic imagination through structured engagement with uncertainty, GenAI functions as a synthetic cognitive and strategic actor, shaping how leaders anticipate, interpret, and respond to crises in real time. While the implications of GenAI span multiple domains, this study focuses on nuclear crises as the most acute and consequential test of these dynamics. The article identifies three interrelated risks: the normalisation of low-probability escalation pathways, the misattribution of adversarial intent cloaked in algorithmic certainty, and the emergence of synthetic feedback loops that can transform foresight into a driver of escalation. Together, these dynamics may generate self-fulfilling escalatory prophecies, undermining crisis stability as simulated futures begin to shape the behaviours they were intended to anticipate or prevent. The article theorises synthetic foresight as a distinct epistemic force – not merely a predictive aid, but a transformative influence on how strategic futures are imagined, interpreted, and acted upon.

The impacts of poverty and material scarcity on human decision making appear paradoxical. One set of findings associates poverty with risk aversion, whilst another set associates it with risk taking. We present an idealised rational-choice model, the Desperation Threshold Model (DTM), that explains how both these accounts can be correct. The DTM assumes that there are basic needs whose satisfaction is not fully divisible. This generates an S-shaped utility function for material resources. The value of gaining a dollar is at first small (because even with the extra dollar, basic needs still cannot be met); then large (because the extra dollar enables basic needs to be met); and then small again. Just above the basic needs threshold, people’s main concern is not falling below, and they are predicted to avoid risk especially strongly. Below the threshold, their most important concern is jumping above, and they are predicted to take risks that would otherwise be avoided. Versions of the DTM have been proposed under various names across biology, anthropology, economics and psychology. We review a broad range of relevant empirical evidence from a variety of societal contexts. Though the model primarily concerns individual decision making, it connects to a range of population-scale and societal issues such as: the consequences of economic inequality; the deterrence of crime; and the optimal design and behavioural consequences of the welfare state. We discuss interpretative issues, and suggest areas for future DTM research that bridges disciplines.

This article examines how China’s central political inspections indirectly enhance municipal provision of invisible public goods. Such goods (e.g., underground pipelines, drainage systems) eludes reliable public assessment through daily observation. Drawing on Mani and Mukand, we emphasize their two defining attributes: (1) conditional evaluation (public judgment requires specific triggers like extreme weather), and (2) temporal accountability lag (delayed quality assessment). Unlike technical business inspections, political inspections prioritize provincial leaders’ political loyalty, generating cascading deterrent effects on municipal officials. Confronting heightened career risks, rational local officials strategically reallocate resources to rectify undersupplied invisible goods. Empirical analysis leveraging the first wave of nationwide inspection data confirms this causal mechanism.

Captivity is a complex phenomenon in international politics with a broad range of purposes, functions, and consequences. Existing scholarship suggests that states use captivity, for example, to facilitate hostage or prisoner exchanges, to extract material rewards, or, in the case of human shields, for deterrence purposes. This article argues that states may use captivity to deter not only traditional military threats emanating from other states, but also perceived threats to regime security posed by non-state actors, including individuals, and that emotions are central to this process. The argument is illustrated through three empirical vignettes that show how the Chinese government has detained foreign academics, publishers, and NGO workers engaged in activities seen as threatening regime security. Detention is interpreted as attempts to deter such actors. While fear is often seen as key to successful deterrence, the article indicates that paying attention to other emotions can help better understand deterrence failure. Specifically, because captivity, and deterrence, involve the denial of the captive’s agency and may trigger feelings of humiliation and shame, it can backfire as the target of deterrence efforts might seek to act to regain agency.

Scholars and policymakers have argued that territorial revisionism is dangerous because it risks setting off a cascade of claims by states dissatisfied with their borders. This Pandora’s box logic suggests that states that are vulnerable to an unraveling of the status quo have incentives to restrain their territorial ambitions to preserve stability. This paper explores this claim theoretically and empirically. It provides descriptive evidence to determine whether vulnerability to territorial threats has historically been associated with a lower likelihood of initiating territorial disputes. We find some evidence of such an effect in postindependence Africa, where this logic is most frequently invoked, and to some extent in Asia, but not in other regions. To help explain these empirical observations, we develop a multistate model of territorial conflict that identifies the conditions under which cooperation to preserve the territorial status quo can be sustained. The model shows that while an equilibrium of mutual restraint can exist, the necessary conditions are quite restrictive, and this cooperative equilibrium is never unique. Thus while a Pandora’s box of potential claims can provide the basis for a norm of restraint, the emergence of such a norm is neither straightforward nor guaranteed.

Backers of nuclear deterrence are thought to use strategic logic, while nuclear disarmament advocates are believed to embrace moral reasoning. Yet policy makers and diverse publics may hold both—ostensibly contradictory—preferences. Recent studies find that publics in Western democratic countries support the nuclear strikes underpinning long-standing conceptions of deterrence policy. But other scholarship indicates that these very same publics want to abolish nuclear arsenals. A lack of comparative analyses across the Global North and the Global South limits the generalizability of these claims. Does a categorical dichotomy between nuclear deterrence and disarmament really reflect global public views on the bomb? What explains a multitude of seemingly inconsistent scholarly results? In this reflection essay, we argue that deterrence and disarmament are not necessarily incompatible tools for reducing nuclear dangers. We point to several ways that individuals might simultaneously accommodate both pro- and antinuclear weapons policy positions. To investigate this proposition, we offer a new observational dataset on global nuclear attitudes from a survey we conducted in 24 countries on six continents (N = 27,250). Unlike isolated studies of these phenomena, our data strongly confirm that publics do not subscribe to categorical views of nuclear weapons. This headline finding and novel dataset open new possibilities for studying nuclear politics.

Military decision-making institutions face new challenges and opportunities from increasing artificial intelligence (AI) integration. Military AI adoption is incentivized by competitive pressures and expanding national security needs; thus, we can expect increased complexity due to AI proliferation. Governing this complexity is urgent but lacks clear precedents. This discussion critically re-examines key concerns that AI integration into resort-to-force decision-making organizations introduces. Beside concerns, this article draws attention to new, positive affordances that AI proliferation may introduce. I then propose a minimal AI governance standard framework, adapting private sector insights to the defence context. I argue that adopting AI governance standards (e.g., based on this framework) can foster an organizational culture of accountability, combining technical know-how with the cultivated judgment needed to navigate contested governance concepts. Finally, I hypothesize some strategic implications of the adoption of AI governance programmes by military institutions.

Thomas Schelling’s 1966 classic, Arms and Influence, became one of the major strategic works of the Cold War, and it remains the clearest argument for the implicit logic of American and Russian coercive forms of diplomacy. Schelling is incisive about the credibility of deterrence, but the credibility of leadership is reduced to the Cold War assumption that power is decisive. While the rise of China and Putin’s invasion of Ukraine have rekindled interest in Schelling’s approach, the diffusion of agency and the interrelationship of issues in the current multinodal era have undermined the efficacy of hegemonic coercion. Rather than restoring Cold War bipolarity, the rise of China has created an asymmetric parity with the United States in which overlapping interdependencies inhibit the formation of camps. In the new era, the pursuit of strategic advantage by any state, large or small, must aim at securing its multidimensional welfare in a complex and unpredictable environment. The global powers are not hegemonic contenders, but rather the largest powers in a multinodal matrix of autonomous states in which each confronts uncertainty. A strategy based on coercion is likely to be less effective against its targets and more costly in its collateral effects. In a post-hegemonic era, Schelling’s premise that arms are the primary path to influence must be reexamined.

Contemporary deterrence scholarship remains disproportionately focused on military instruments, often neglecting the strategic utility of diplomacy, information, and economic statecraft. Our study addresses this imbalance through a new methodology for analysing how authoritarian states respond to the range of foreign policy tools: diplomatic, information, military, and economic (the DIME framework). Using a state’s propaganda, official statements, and media (POSM) to capture target states’ reactions to adversarial DIME actions, we offer an innovative analytical framework that enhances understanding of deterrence dynamics beyond the military sphere. Within the framework, we use computational text analysis, statistical analysis, and data visualisation to create a replicable process for analysing POSM big data. Applying this methodology to a case study of China, we find that Beijing’s POSM-based responses to information tools – such as public criticism of censorship and information control by NGOs – are more negative than to diplomatic, military, or economic tools. Our methodology contributes to deterrence theory and policy through its insight into non-military effects and by offering a scalable process for empirical analysis ripe for AI implementation. For policymakers, our process and findings hold implications for crafting more effective and sustainable deterrence strategies in an increasingly complex international security environment.