The global academia is racing to craft regulations to govern the use of generative AI (GenAI). Yet, most policies and regulatory principles in GenAI governance are top-down and risk-averse, leaving the questions of ethics of users underexplored. By situating the “ethical use” at the center of the AI governance discourse, this article asks (i) how learners perceive the ethicality of GenAI use, and (ii) what considerations shape their decision to use GenAI ethically. Drawing on the Aristotelian virtue ethics theory as sensitizing tool and using data from nine focus groups of learners (n = 33) from a publicly funded university, we derived a Five-Rights Framework of Virtue Based GenAI Governance: right actions (ethical use cases), right outcomes (outcomes gained from using), right beliefs (new moral virtues of open-minded caution and intellectual autonomy), right approaches (technical skills and techno-practical wisdom), and right community (peer influences, usage policy). The analyses also revealed that (1) negative outcomes of using GenAI came from deficits in the right ways of using and technical skills; (2) the willingness to declare the use of GenAI is influenced by social stigma and the lack of fair and transparent detection regimes; and (3) peers’ attitudes influence the use or refusal of GenAI. The study advances GenAI governance scholarship by contributing a blueprint of a “virtue-based” governance regime, offering evidence-based and theoretically informed governance suggestions in global academia concerning the ethical use of GenAI.

As data-driven service delivery expands, data reuse holds significant potential to improve access to and quality of essential services for young people. However, limited youth involvement in decisions about how their data is reused risks perpetuating mistrust and deepening the inequalities that these services seek to address, particularly if young people choose to avoid seeking services or withhold critical information out of fear of misuse. Grounded in a social license approach, responsible data reuse aimed at enhancing service delivery therefore requires methodologies that meaningfully engage youth and reflect their preferences and expectations. This article presents findings from the NextGenData project, which developed and piloted a scalable methodology for engaging young people aged 19–24 in co-designing responsible data reuse strategies. Conducted as a year-long participatory action research initiative across India, Tanzania, Moldova, and Kyrgyzstan, the approach implemented youth assemblies, deliberative methods, and localised facilitation by national partners to engage young people. Through a cross-contextual analysis, this study emphasises the importance of context-sensitive, multi-phase engagement in supporting the development of a social license for data reuse and presents a publicly available toolkit designed to support the replication and adaptation of this engagement strategy in diverse contexts. Drawing on the findings, recommendations are presented for policymakers and practitioners to guide future initiatives.

The term data sovereignty—the principle that data is subject to the laws and governance of the country where it is collected or stored—faces implementation challenges. The legal deadlock over how to regulate transatlantic (E.U.–U.S.) data transfers reveal the difficulties of asserting “sovereignty” over data, a resource that flows transnationally through digital technologies. Among global regulatory efforts, the European Union’s struggle to ensure the supremacy of its data protection and privacy laws represents an interesting case from which to derive new theoretical and conceptual insights. This paper advances data agency—as agency mediated by data, not over it. The intention is to advance a term that moves beyond the limits of data sovereignty to explain how governance over data-driven systems is distributed and contested. The manuscript suggests pragmatic pathways for more effective governance as mere “control” over data is not enough to achieve meaningful insights derived from data technologies.

Academic peer review is in crisis—it is a structural reality that every editor, reviewer, and conscientious author now navigates daily. The symptoms are familiar: reviewer fatigue, inconsistent decisions, declining response rates, and a growing sense that the machinery of scholarly quality assurance is straining under a load it was never designed to bear. For a journal like Data & Policy—operating at the intersection of multiple disciplinary traditions and sectorial experiences, serving a field still in the process of constituting itself—these pressures are not abstract. They shape every editorial decision we make.

The practice of public administration is confronted by an overwhelming volume of unstructured textual data, creating a critical bottleneck that challenges the capacity for timely, evidence-based policymaking. While large language models (LLMs) offer a powerful solution, the academic and policy discourse on AI governance has primarily focused on the risks associated with quantitative, predictive models, leaving the governance of LLMs, which involve qualitative, interpretive analysis, critically underexplored. This article provides a systematic and empirical answer to this challenge. We evaluate four distinct archetypes for AI-assisted policy analysis: (1) traditional expert-driven analysis, (2) automated topic modelling, (3) human-in-the-loop LLM co-production (ChatGPT with prompting), and (4) institutional-scale API automation (ChatGPT via API), by applying them to a real-world corpus of 63 United Nations policy documents. Our results reveal the inherent trade-offs of each archetype in terms of analytical speed, cost, contextual depth, and the requisite level of human oversight, finding that the human-in-the-loop model offers the most effective balance. We conclude by presenting a durable framework to guide public institutions in making a strategic, evidence-based choice of analytical workflow, ensuring the adoption of AI strengthens, rather than erodes, the integrity of democratic governance.

In recent years, the rapid development of artificial intelligence has raised a series of ethical issues. How to balance AI innovation and AI governance has attracted wide attention from society. Some principle-based and practical approaches have already emerged. However, these strategies mainly face problems of time, speed, and cost. They fail to fundamentally balance dynamic factors and the need for explainability. This study proposes an explainable and adaptive AI governance approach from the perspective of time to balance AI governance and AI innovation. Causal models have causal explanation capabilities. They meet the requirements of responsible innovation. They can improve governance speed and reduce governance and innovation costs. Taking Bayesian networks as an example, they have explainability and dynamic adaptation ability. Bayesian networks can be used as a governance step before a regulatory sandbox. Bayesian networks can implement a human-assisted regulatory approach. This approach can combine the EU Artificial Intelligence Act with scholars’ recommendations on quantifiable risk levels. The explainable and adaptive AI governance approach has three important meanings for AI governance. First, it can provide an inclusive governance path. Second, it can supplement static governance strategies. Third, it can automatically generate explainable value alignment schemes.

Gross Domestic Product (GDP) remains the principal benchmark of economic performance; however, its interpretative value depends fundamentally on the reliability, traceability, and calibration of underlying measurement systems. Despite extensive debates on alternative indicators, limited attention has been given to the reliability of GDP as a measured construct. This study aims to operationalize measurement uncertainty as a governance-relevant variable within macroeconomic interpretation. To address this gap, this study introduces the National Product Uncertainty Loss (NPUL) model, a framework designed to quantify the economic implications of insufficiently calibrated measurement processes embedded within sectoral GDP components, thereby revealing hidden fiscal risks that remain unaccounted for in conventional economic indicators. Using Albania as a pilot case, complemented by comparative insights from Western Balkan economies, the analysis demonstrates how traceability gaps in sectors such as energy, healthcare, hydrocarbons, and construction translate into measurable fiscal exposure. Empirical simulations suggest that systemic measurement inefficiencies may generate baseline deviations of ~0.3–0.4% of GDP, with upper-bound scenarios reaching 1–2.2% under adverse conditions. The findings indicate that strengthening calibration infrastructures and improving uncertainty transparency are not merely technical refinements but core elements of institutional governance reform. By integrating measurement reliability into macroeconomic interpretation, NPUL enhances the policy relevance of GDP and contributes to ongoing debates on data governance, digital sovereignty, and evidence-based policymaking.

Universities, governments, and health research infrastructures increasingly invoke the language of “trusted,” “secure,” and “safe” research environments to legitimise access to sensitive data. While technical infrastructures such as Trusted Research Environments, Secure Data Environments, and cloud-based analytical platforms have expanded rapidly, equivalent growth in operational governance capability has not always followed. This article argues that contemporary research governance risks drifting towards “governance theatre”: the performance of trustworthiness through visible technical controls, secure branding, and procedural symbolism while comparatively less attention is devoted to disclosure expertise, behavioural governance, independent oversight, and operational maturity. Using recent controversies surrounding the UK Biobank as an illustrative governance stress test rather than an isolated institutional failure, this article examines tensions emerging across contemporary research practices—including platformisation, global collaboration, artificial intelligence, code sharing, and increasingly complex data systems. Drawing on the five safes framework, this article argues that trustworthy governance cannot be reduced to technical infrastructure or formal compliance alone. Governance maturity depends upon layered socio-technical systems capable of managing disclosure risk, researcher behaviour, and organisational accountability across the full research life cycle. Moving beyond governance theatre will require greater investment in governance literacy, disclosure expertise, independent output review, and reflexive oversight capable of adapting to rapidly evolving systems.

This article examines how governance instruments for algorithmic systems are developed and framed, to answer the question of how notions of risk are constructed and institutionalized in governance instruments for the governance of algorithmic systems. Focusing on the Dutch context, we conducted an abductive discourse analysis on media articles from five different newspapers, advisory reports, and parliamentary documents to understand how these governance instruments not only respond to risk but also actively construct and institutionalize specific notions of risk in algorithmic governance. From this analysis, we conclude that governance instruments that emphasize procedural compliance are often chosen over instruments that create more substantial solutions, partly driven by the mediatization of politics and the prioritization of visibility over effectiveness. However, more substantive governance instruments, such as the Fundamental Rights and Algorithms Impact Assessment or educational programs for civil servants, are increasingly being picked up and implemented, suggesting a shift in focus to underlying issues such as discrimination or a lack of accountability.

Academics and practitioners recognize the importance of data to achieve critical public value outcomes: more effective services, more efficient use of resources, more equitably distributed benefits, and increased transparency and accountability of government action. Scholars are increasingly trying to understand the conditions that influence public organizations’ ability to use data to meet these goals, but existing research largely follows two separate approaches to studying this phenomenon. Some researchers take organizations as their main unit of analysis to understand how organizational configurations and resource endowments explain differences across governments. Other scholars study how individuals’ competencies and behaviors affect effective interpretation of data and its use to make decisions. Between these two poles lies a missing link: the collective and coordinated capabilities of groups of individuals responsible for collecting, processing, sharing, and using data in the service of tackling public problems. Using a comparative case-study design, we study teams leveraging data to tackle property distress and abandonment in three US cities. This exploratory empirical research allows us to derive insights into the main constitutive components of data-related capabilities, their enabling factors, and their effects. This understanding can advance future scholarship by providing researchers with a more operationalizable construct to study data capabilities, bridging an important gap between the main existing approaches to studying data-informed governance.

The UK and Australian Modern Slavery Acts require large corporations to disclose annually how they address modern slavery risks in their operations and supply chains. Existing methods assess only a small proportion of these disclosures, or narrowly against explicit legal criteria, overlooking deeper indicators of corporate commitment. This paper presents AIMS-QA, an AI-driven framework for assessing modern slavery statements at scale and beyond compliance. It is built on a structured taxonomy synthesising 13 benchmarking methodologies from academic and civil-society analyses, consolidating 473 metrics into 209 streamlined questions across 115 thematic topics. From these, a 37-question priority framework captures the most widely used indicators while retaining key statutory requirements. The seven reporting criteria of the Australian Modern Slavery Act provide the analytic anchor, with each question tagged as either a statutory baseline requirement (reflecting jurisdiction-specific legal obligations) or a beyond compliance expectation (reflecting stakeholder, investor, and civil-society priorities). The paper then introduces a two-stage AI-driven question answering pipeline. The first stage uses a fine-tuned classification model to extract relevant evidence, which is passed to a large language model, guided by domain-specific prompts, to generate targeted answers. A second-stage retrieval-augmented generation (RAG) Watchdog intervenes where the classifier finds insufficient evidence or the language model cannot answer confidently. Designed for triage and screening rather than enforcement, with human-in-the-loop oversight, AIMS-QA gives policymakers, regulators, and stakeholders a research-informed means of strengthening the monitoring and impact of modern slavery legislation.

COVID-19, climate change, and Russia’s invasion of Ukraine have underscored the importance of environmental sustainability and digitalisation for maintaining the EU’s economic stability, social resilience, and global position. Green and digital transition strategies have emerged as a response to these immediate challenges, aiming to mitigate their negative impact on European citizens and businesses and ensure the EU’s smooth transition to an environmentally sustainable digital economy. While green and digital transitions are often viewed as separate issues of conflicting nature, their interaction presents an immense potential for the EU’s increased growth and global competitiveness, indicating the importance of a twin transition for the future of Europe. As an intersecting central element of both the green and digital transitions, data can play a critical role in the achievement of the European twin transition. Through regulation, data sharing can be used as an effective instrument to leverage data towards the EU’s mutually reinforcing green and digital transitions. Embracing this approach, this article first demonstrates how data as an economic good can be leveraged through regulation as a catalyst for the twin transition by using data sharing as a redistribution mechanism that reallocates the socio-economic value of data. It then adopts Pareto efficiency as a conceptual benchmark to analyse to what extent the current EU data strategy contributes to the twin transition based on this framework. Finally, the article concludes with a brief discussion on how the Digital Omnibus may affect the interplay between the EU’s data strategy and its twin transition.

Notwithstanding its size and importance in a globalized world, there is little systematic information about the volume and direction of human travel that considers both migration and other kinds of mobility. This article introduces new estimates of the volume of migratory and non-migratory human travel across country borders worldwide between 1995 and 2022. Building on pioneering work by Recchi et al. (2019, Estimating Transnational Human Mobility on a Global Scale), the estimates integrate information from three major sources: tourism data, air passenger data, and migration data. Our estimates reveal that total transnational mobility increased from 5.29 billion trips in 1995 to 10.66 billion in 2019, largely outpacing global population growth and telling a story of continued globalization in that period—or, more precisely, of regional integration, since four out of five cross-border travels occur within the same world region. Across the board, migration (i.e., travel followed by residence change) constitutes a tiny and decreasing fraction of international travel; less than 1% worldwide and as low as .12% in Europe. Amidst the sustained growth in volume, the global network of human cross-border mobility has experienced slow “tectonic shifts” in some of its major clusters, while maintaining a stable configuration in others.

With rising costs of living, concerns over the stability of household finances dominate every level of decision-making, from individuals choosing which expenses to prioritise to government policy. Using County Court Judgment (CCJ) data (a record of defaults on unsecured debts acknowledged by the UK legal system) between 2016 and 2023, this paper explores the spatial distribution of households facing legal repercussions for debt arrears in England and Wales. We then provide high-resolution evidence of the shared socioeconomic characteristics of these areas. The results of this analysis demonstrate that areas facing the highest rates of CCJs are often those that are experiencing the greatest social and economic challenges. However, whilst general deprivation measures do broadly align with higher cases per capita, focusing on specific features provides a substantially more accurate picture of where the highest rates are found, with tenure type being found to have the greatest influence on expected indebtedness. We then consider an approach to relaxing the assumptions about spatial uniformity that underpin our initial model. A mixed-effects model demonstrates modest improvements, correcting for spatial concentrations of errors, whilst a regional model suggests significant sub-national variations in driver effects. Overall, the results presented use a novel data source to provide further support for a number of established theories about the drivers of indebtedness in England and Wales, whilst questioning the extent to which these theories apply equally across space.

Housing agencies must prioritize physical inspections across tens of thousands of subsidized properties while lacking systematic condition data for most. We demonstrate that Graph Attention Networks (GATs) can propagate distress signals from inspected to uninspected properties through ownership and spatial networks, substantially improving screening efficiency over conventional approaches. Ownership connections add predictive value beyond geographic proximity. Removing them lowers performance, and properties linked to owners with distressed properties elsewhere face elevated predicted risk, even within the same neighborhoods. This finding is consistent with theoretical claims that maintenance outcomes cluster at the portfolio level rather than merely reflecting neighborhood poverty. Cross-program validation also exposes important limitations. In a separate feature-based experiment, an XGBoost model trained on federally inspected Section 8 properties achieved near-random performance when applied to the available Low-Income Housing Tax Credit sample, with an AUC of 0.515. Because this experiment did not directly evaluate the GAT, it does not determine whether graph-based representations would improve cross-program transfer. The result nevertheless cautions against applying models trained on U.S. Department of Housing and Urban Development-inspected properties to other program contexts without program-specific validation. Integrating predictions with economic pressure measures, we identify properties facing compounded threats, including simultaneous physical distress, market exit pressure, contract expiration, and advanced age. These properties concentrate geographically, creating opportunities for targeted intervention. We contribute methodologically by demonstrating what graph neural networks can accomplish within an administratively coherent setting while identifying important limits to cross-program transfer across fragmented housing policy landscapes.