Calls for international administrations are still present in contemporary international relations, from the war in the Middle East to the Ukraine. This chapter summarises the main argument developed in the book, notably the necessity to apprehend international administrations through the reality on the ground. This means to pay special attention to the confluence of claims of political authority deployed by international officials and concomitant claims of accountability such practices of authority elicit. Finally, I discuss the various possibilities to strengthen accountability mechanisms for the UN (and for other international organisation).

This chapter introduces the main approach developed in International Leviathans through (1) a sociological understanding of sovereignty and (2) the concept of international administration. First, the chapter presents a new take of the debate between Kelsen and Schmitt around sovereignty and presents the sociological understanding of sovereignty by unpacking sovereignty practices. It makes the case for analysing the socio-political or socio-legal struggles happening between competing claims of political authority and accountability. It then discusses and theorises the concept of ‘international administration’, pointing out the limitations of two strands of the literature: the functionalist approach, defining international administrations through the functions they ought to perform as underlined in their mandate, and the normative approach, focusing on self-proclaimed goals and objectives. I posit that sovereignty practices deployed by international officials are social practices which cannot be understood solely through mandates or stated goals – they need to be understood through the reality on the ground, created by claims of political authority deployed by actors and the concomitant claims of accountability these practices elicit.

This chapter presents the central claim of the book, arguing that sovereignty practices emerge at the confluence of struggles – on the one hand, by actors asserting the political authority over a specific territory and, on the other hand, through resistance to this move by actors pursuing accountability and the responsibilisation of sovereign actors. I claim that practices of political authority – expressed through effective rule over territory – are central, constitutive acts of world politics, entailing specific obligations. Broadening the study of sovereignty practices beyond state relations, I argue that understanding how specific actors such as international organisations act as sovereign actors opens up new perspectives on international accountability and obligations in world politics.

Three concepts associated with ‘good governance’, referring to the quality of governance processes and outcomes, are examined in this chapter: accountability, legitimacy and trust. For each concept, definitions are reviewed and characteristics identified. The chapter investigates accountability through key themes of complexity and challenges associated with delivering on downward accountability in the context of decentralisation. Complexity arises from the number and range of actors involved in co-management, situated at different administrative levels, with multiple demands on and mechanisms for accountability. Different forms of legitimacy and trust are introduced and reviewed, including input, output and throughput legitimacy and dispositional, rational, affinitive and procedural forms of trust. Challenges to delivering and maintaining legitimacy and trust in the context of co-management and solutions to these are identified.

The conclusion synthesizes the book's argument and reflects on its implications for law, business, and democratic governance. It reiterates the central claim that corporate decisions now shape public life before law can respond. In this environment, the quality of corporate decision-making becomes a matter of public concern.

The conclusion explains why corporate ordering cannot replace democratic institutions. Corporations lack electoral legitimacy and cannot resolve value conflict. Yet as long as they exercise power, their internal governance structures matter.

The chapter emphasizes a shift from outcome-based debates to process-based evaluation. Rather than asking whether corporations pursue the correct goals, the focus should be on whether they deliberate responsibly under conditions of disagreement.

The conclusion calls for scholars and policymakers to study corporate governance as a site of norm creation. It highlights transparency, institutional design, and monitoring as emerging foundations of accountability.

The book closes by framing corporate ordering as neither celebration nor condemnation. It is a descriptive and normative account of how power is exercised in a world where consensus is rare and delay is costly.

Chapter nine translates the book's insights into a proposal for legal reform. It argues that fiduciary duty should evolve to require deliberative capacity when corporate decisions have significant social effects.

The chapter does not propose new substantive obligations. Instead, it focuses on process. Boards should be required to establish governance structures that allow informed inquiry, justification, and independent review. Failure arises only when no such structures exist or when deliberation is cursory.

The chapter shows how this approach fits within existing doctrine. Courts already evaluate processes in areas such as mergers and compliance oversight. Extending this logic to social governance preserves deference while improving accountability.

The chapter addresses concerns about judicial overreach and managerial paralysis. By focusing on capacity rather than outcomes, the proposal avoids asking courts to resolve social disputes.

This chapter offers a practical path forward that aligns corporate governance with contemporary realities while respecting institutional limits.

The three chapters in this section offer critical diagnoses of how algorithms govern through their interactions with and beyond data. Each focusses on a particular technical practice of algorithmic governance, provides an account of its logics, and seeks to reveal the consequences of embedding such practices in assemblages of governance: ‘proxification’ (Phan and Wark), the ‘human in the loop’, (Goldenfein), and commercial techniques of machine learning in public administration (Yeung). In this short intervention, we offer, first, some reflections on what these papers share along three distinct axes – the analytical, the methodological, and the political – before, second, seeking sympathetically to extend them, including by identifying novel questions raised by their analyses.

Human oversight of automated decisions has become regulatory orthodoxy. Scholars and policy makers frequently turn to the human to ameliorate automation’s disruptive potential. But there is little empirical evidence that human oversight or a human in the loop improves decision outcomes. This suggests that the regulatory value of human oversight is more political than it is empirical. This chapter accordingly asks what political work do legal requirements for human oversight perform, and what kinds of political configurations do they enable? Through analysis of cases invoking human in the loop in administrative contexts, this chapter argues that, in the name of recuperating rule of law values, the human in the loop is deployed as a versatile legal technology that enables troubling distributions of accountability and inhibits a richer legal conceptualisation of automated decision systems.

This chapter introduces trustworthy AI as a foundational requirement for the responsible development and use of artificial intelligence. It distinguishes trustworthy AI from responsible AI, framing AI as part of broader socio-technical systems rather than as a purely technical artifact. The chapter explores key principles such as reliability, safety, transparency, fairness, and accountability, and emphasizes that responsibility lies with people and institutions, not machines. It introduces “Question Zero”: whether AI should be used at all in a given context. The chapter positions trustworthiness as essential for innovation, public trust, and societal benefit.

Accountability in AI is framed as the clear allocation of responsibility across the actors involved in the design, development, deployment, and use of AI systems. Rather than attributing responsibility to machines, the focus lies on human and institutional accountability, supported by governance mechanisms such as documentation, traceability, auditability, and redress. The content highlights the importance of oversight, clarity of roles, and the ability to contest and remedy harm. Accountability is presented as essential for trust, legal compliance, and ethical legitimacy, ensuring that AI-enabled decisions remain answerable to society.

This chapter critically examines the concept of moral responsibility at the heart of everyday interpersonal and legal practices. While we routinely hold people morally responsible – praising and blaming them as if they genuinely deserve it – the chapter argues that such desert-based judgments are philosophically unjustified. Drawing on a range of philosophical literature, it distinguishes between several types of responsibility, including causal, role, and liability responsibility, and focuses on the central question of basic desert moral responsibility. The chapter defends the view that this kind of responsibility requires a form of free will we do not possess. Building on Derk Pereboom’s work, the chapter proposes a forward-looking, non-desert-based alternative grounded in protection, moral formation, reconciliation, and restoration. It also challenges the widespread assumption that moral responsibility practices are psychologically fixed or evolutionarily hardwired, citing crosscultural and empirical research to support their malleability. By replacing retributive attitudes with constructive, future-oriented responses, the chapter shows how free will skepticism can preserve what matters most in our moral lives – without relying on morally problematic notions of blame, indignation, or deserved suffering.

This chapter examines the often harmful consequences of the deliberate anthropomorphizing of generative AI (GenAI) chatbots by tech companies. Developers intentionally fine-tune systems to be human-like and accommodating, employing strategies like personalization, fostering dependency, offering excessive empathy (sycophancy), adopting a servility model, and promoting endless conversationality, all to maximize user engagement and monetization.

The chapter details how this strategy prioritizes profit over user welfare, leading to unintended and severe consequences, including psychological harm, increased social isolation, and the spread of dangerous misinformation, particularly in fields involving highly vulnerable people, such as mental health and education. The analysis criticizes the lack of accountability from AI companies, which often disclaim responsibility for the harms their products cause.

This chapter introduces Governance by Emulation, a framework analyzing how public law models, particularly administrative and constitutional mechanisms like individual rights adjudication, are reproduced in private and regulatory governance. Focusing on corporate-controlled content moderation, it examines the European Union’s out-of-court dispute settlement bodies (ODSs) under the Digital Services Act and Meta’s Oversight Board–conceptualized as Emulated Guardians. These institutions borrow the legitimacy of courts to regulate novel, bureaucratic private power structures while addressing public demands for accountability. Grounded in law, sociology, and political science, the chapter outlines the book’s methodology and contributions. It delves into four inquiries: the actors involved, their tasks, the power they seek to discipline, and how public law principles are adapted for private governance. These dynamics highlight emulation’s duality: it promises innovation yet risks performative legitimacy devoid of substantive reform. By situating Emulated Guardians within broader global governance challenges, this chapter frames content moderation as a microcosm of future issues in sectors like AI, biotechnology, and space exploration. It concludes that while governance by emulation addresses urgent accountability demands, its efficacy depends on public engagement and institutional evolution, offering a critical lens to assess emerging accountability structures beyond state control.

This chapter turns to two local protective mechanisms that primarily protect communities through civil society partnerships: threatening accountability and reintegrating victims. Churches brought needed local credibility to civil society partnerships, while civil society organizations contributed expertise and implementation experience. Threatening accountability encompasses activities that raise the likelihood that state agents will be held accountable for human rights abuses, such as training local church networks in human rights documentation. Efforts at reintegrating victims involve initiatives to socially, financially, and psychologically assist the families of those left behind by Drug War violence, who remained at great risk of subsequent violence. After using the parish-based survey to get an overall sense of the distribution of these activities, the authors empirically substantiate their operation through qualitative and quantitative evidence, including parish interviews, experimental evidence from a sample of plausible Philippine police personnel, and survey evidence of ordinary Filipinos.

Chapter 4 reconceptualises investor stewardship by tracing its historical, conceptual, and economic roots and advancing a theory of stewardship as delegated, relational power. It unpacks the evolving meanings of stewardship – from early moral connotations to contemporary use in corporate governance and investment management. Rejecting a narrow principal–agent lens, it reframes stewardship as a multidimensional practice embedded in complex delegation structures and layered accountabilities. The chapter introduces a tripartite model of stewardship – as power exercised by institutional investors, on behalf of clients and beneficiaries, and for the benefit of wider, often unseen, stakeholders – and a four-part relational model: client stewardship, end-investor stewardship, asset stewardship, and sustainability stewardship. These relationships expose the plural and sometimes conflicting responsibilities investors bear within a fragmented investment chain. It also considers the economic rationale for investor stewardship, highlighting incentives, constraints, and portfolio dynamics. Finally, it introduces the enlightened steward as a pluralistic figure balancing private mandates with systemic effects.

The problem of nonquantifiability is recurrent in both public policy and ordinary life. We often cannot quantify the benefits or costs of potential courses of action, and we must nonetheless decide whether and how to proceed. Government agencies are generally required to quantify both benefits and costs and to show that the former justify the latter. But they may also consider nonquantifiable factors such as human dignity and fairness, and understanding is constrained by the limits of existing knowledge. When quantification is impossible, agencies should engage in “breakeven analysis” to explore how high nonquantifiable benefits would have to be in order to justify the costs. Breakeven analysis can be used in three different ways. (1) Agencies may identity lower or upper bounds through either point estimates or an assessment of expected value. (2) Agencies can explore comparison cases in which relevant values have already been assigned (such as for a statistical life). (3) When those two options do not apply, agencies may identify what information is missing and specify the conditions under which benefits would justify costs (“conditional justification”). In admittedly rare cases, regulators, no less than individuals, might have to “pick” or “opt.”

Decisions by international organizations typically neglect the interests of non-human animals. The chapter investigates whether and how animal interests can and should be brought to bear in the decision-making of IOs. It works through cognate concepts ranging from animal citizenship over animal representation to animal consideration and animal deliberation. The physical limits of human-animal communication foreclose responsiveness and accountability to the animals themselves. The chapter therefore prefers the term animal ‘consideration’ rather than animal ‘representation’. After this groundwork, the chapter briefly canvasses some proposals for bringing animal interests to bear in in democratic political processes. With due modifications, some schemes could be applied to the work of international organizations. These range from animal ombudspersons, strengthening the voice of pro-animal CSOs through compulsory notice-and-comment procedures and extended speaking rights in the organizations, mandatory animal welfare impact assessment, and more. All attempts for upstepping the existing rudimentary schemes in the direction of a better and stronger consideration of animal interests in human politics will require deep cultural and social change, to a large extent beyond the purview of the law.