The team turned to an unusual source of evidence: the OECD’s catalogue of AI ethics tools and trust or quality mark schemes. As of 17 July 2025, that repository listed 938 tools, of which 24 were formal certification or quality marks intended to label AI systems as trustworthy. Rather than sampling, the researchers analyzed the full set, using a descriptive and comparative approach that mapped each tool and framework against the ethical objectives it supports, the type of intervention it represents — technical, educational, or procedural — the stage of the AI lifecycle it addresses, the stakeholders it targets, and the skills it demands. The authors are candid about the limits of this method: they adopted the OECD’s own classifications without independent recoding, and their findings describe the curated repository rather than the entire global landscape. Even so, the patterns that emerge are striking enough to matter.

The first finding concerns which ethical principles get built and which get talked about. Transparency emerged as the single most widely supported objective in the catalogue, followed closely by fairness and robustness. This is the good news: the AI community has invested heavily in bias-mitigation toolkits, model documentation standards, and stability testing, translating the most-cited principles into software libraries, checklists, and assessment frameworks. But the distribution is sharply non-uniform. Explainability — the capacity to interpret and comprehend why an AI system made a particular decision — is addressed far less frequently than transparency, likely because producing genuinely human-interpretable explanations for complex models remains a formidable technical challenge, and because the boundary between “transparency” and “explainability” is itself contested. Digital security tools, which defend against adversarial attacks and data breaches, are similarly scarce. And environmental sustainability is almost an afterthought: very few tools explicitly target the energy consumption or carbon footprint of AI systems.

This asymmetry has a telling structure. Tools proliferate where problems are quantifiable, where regulatory and public pressure is strongest, and where solutions can be generalized into reusable libraries. Measuring demographic bias in a dataset is tractable; explaining the internal reasoning of a deep neural network in a way that satisfies a regulator, a doctor, and a defendant is not. The authors argue that the implementation ecosystem does not simply execute pre-given moral commitments — it filters them through organizational feasibility, technical legibility, and institutional incentives. Values that can be measured, audited, and formalized get infrastructure; values that resist formalization, such as sustainability or genuine interpretability, remain rhetorical. The result, they warn, is a risk of achieving only a “verisimilitude of trustworthiness” — an appearance of ethical completeness that is incomplete at its core.

The second major finding concerns the types of interventions being deployed. Technical tools — software libraries, algorithms, and evaluation platforms — dominate the catalogue, concentrated overwhelmingly on transparency, robustness, and fairness. Procedural tools, such as governance checklists, risk-management frameworks, and documentation templates, come next, and they too cluster around transparency, fairness, and privacy. Educational tools — training programs, courses, and best-practice guides designed to build ethical competence among practitioners — are the least common category of all, and the few that exist focus on broad principles while neglecting explainability, data governance, digital security, and sustainability. This points to what the researchers call a significant capacity-building gap: organizations are buying technical fixes and adopting process documents, but they are not systematically educating the people who must apply them.

The authors connect this gap to a deeper philosophical worry. A governance ecosystem centered on tools and procedures risks externalizing moral responsibility into artifacts of compliance — the implicit assumption that ethical adequacy can be achieved through the correct use of instruments alone. But trustworthiness, they argue, is not an inherent property of a technical system; it arises from socio-technical relationships among designers, institutions, affected communities, and the normative assumptions embedded in practice. If developers and managers lack ethical literacy, checklists may be implemented perfunctorily or audited workarounds found — the phenomenon critics have labeled ethicswashing or ethics theater. Citing research arguing that AI development often proceeds in an “ethically empty milieu,” the study suggests that cultivating judgment, reflexivity, and practical wisdom remains structurally undervalued, and that future certification schemes could mandate or incentivize accredited ethics training for AI teams rather than treating training as optional.

Perhaps the most consequential finding concerns timing. When the researchers examined where trust and quality mark frameworks intervene in the AI lifecycle, they found a heavy skew toward the late stages: the “Verify & Validate” and “Operate & Monitor” phases are by far the most frequently addressed, while the early “Plan & Design” and “Collect & Process Data” stages receive comparatively little attention. In other words, most certification schemes assume the AI system is already built and then audit its behavior — a final compliance check rather than a design discipline. The authors liken this to the history of software security, which long relied on penetration testing after development until “Secure by Design” philosophies took hold. If ethical flaws — biased training data, an unsafe architecture, opaque model logic — are embedded early and only discovered at validation, fixing them may be too late or prohibitively expensive.

This temporal imbalance, the study argues, reflects a particular moral chronology in which ethical reflection is displaced downstream, where it is weaker, costlier, and less transformative. The “Ethics by Design” literature has long maintained that values are not appended to systems at the end of development; they are already inscribed in problem formulation, data selection, and assumptions about users and harms. The researchers recommend that next-generation governance frameworks incorporate design-phase requirements: algorithmic impact assessments conducted before model development, participatory design with affected stakeholders, bias-aware data collection strategies, and documentation of how ethical considerations shaped design decisions — not merely how the final model was evaluated. The hard part, they acknowledge, is creating concrete standards for design practices that can actually be verified.

The third axis of asymmetry concerns audience. The most frequently targeted users of trust mark frameworks are data scientists, developers, and business leaders, with many schemes also addressing all employees of an organization — an internal, corporate-compliance orientation. Policymakers, regulators, the broader public sector, and the general public are markedly under-targeted. This skew means current trust marks function largely as industry-led self-governance: companies assessing their own adherence to ethical standards, with little external oversight or public empowerment. The authors warn that when the power to define what counts as a trustworthy system remains concentrated among those closest to production and deployment, the harms, dependencies, and exclusions experienced by affected communities — often invisible from a technical or managerial standpoint — go underrepresented. They call for co-regulation, independent audits, and public–private partnerships that bring regulators and civil society into the definition and verification of trustworthiness, noting that the required skills listed in the catalogue — programming, data management, IT competencies — reinforce the technical gatekeeping.

The study closes with a set of recommendations that read as a roadmap for the field: broaden the ethical objectives of tools and certifications to include environmental sustainability, explainability, and other neglected principles; embed ethics earlier in the lifecycle through design-stage assessments; expand multi-stakeholder participation to include policymakers, end-user representatives, and interdisciplinary experts; invest in ethics education as a core organizational competence; and strengthen enforcement by linking voluntary marks to regulatory or contractual requirements with independent audit capabilities, updated through regular structured reviews. The overall diagnosis is one of significant progress coupled with significant imbalance. The AI community has demonstrably moved beyond principles on paper — bias audits, transparency documentation, and certification schemes now exist at scale. But the ecosystem, the authors conclude, selectively stabilizes the values that are easiest to formalize while marginalizing the rest. The future of AI governance, they suggest, will depend not on producing more principles, but on transforming the institutional, epistemic, and design conditions under which any principle can become materially operative — and on ensuring that the AI systems increasingly woven into society earn the trust of the people they affect.

Subject of Research: Empirical analysis of trustworthy AI tools and trust mark frameworks using the OECD catalogue

Article Title: A critical analysis of trustworthy AI tools, mark frameworks, and the implementation chasms

Article References: Papademas, M., Karpouzis, K., Ziouvelou, X., & Karkaletsis, V. (2026). A critical analysis of trustworthy AI tools, mark frameworks, and the implementation chasms. AI & SOCIETY. https://doi.org/10.1007/s00146-026-03364-4

Image Credits: AI Generated

DOI: 10.1007/s00146-026-03364-4

Keywords: trustworthy AI, AI ethics, OECD catalogue, AI governance, explainability, algorithmic fairness, ethics by design, AI certification, ethicswashing, AI lifecycle, sustainability, AI auditing

The study’s central conceptual move is a distinction that sounds simple but carries enormous consequences: the difference between the governance of AI and governance by AI. The first phrase describes the familiar task of regulating, auditing and ethically constraining algorithmic systems. The second describes something far stranger, in which AI systems themselves increasingly supply the inputs, and sometimes the outputs, of decisions that shape public life. As AI shifts from an object of regulation into an active participant in administration, accountability becomes shared between human officials and algorithmic actors, and oversight mechanisms designed for purely human bureaucracies begin to strain. Kaiser argues that this shift turns governance into a hybrid human-machine decision environment, demanding new institutional arrangements rather than incremental policy patches.

The urgency is not hypothetical. The United Arab Emirates has announced plans to use AI to support drafting, reviewing and amending legislation, while Albania introduced Diella, an AI-generated digital minister, as part of procurement and anti-corruption reform. These early, largely political experiments signal a growing willingness to invite AI into decision-relevant spaces once reserved for human officials. Meanwhile, AI policy initiatives now span more than 80 countries and jurisdictions, and technologically advanced states are making billion-dollar investments to secure strategic leadership. The question, the paper suggests, is no longer whether AI enters governance, but whether governance can absorb AI without losing accountability, legitimacy and public trust.

To ground the analysis, the study synthesizes cross-sector risks that recur wherever AI is deployed. Algorithmic bias tops the list: systems trained on non-representative datasets can reinforce existing disparities, as when diagnostic algorithms underperform in minority populations. Real-world failures illustrate the stakes. A wrongful arrest in Tennessee was linked to an AI facial recognition error, and Michigan’s MiDAS system falsely accused thousands of citizens of unemployment fraud through automated determinations. Beyond bias, the review identifies weak oversight, regulatory gaps and capture, vendor dependence, geopolitical and socio-political risks, environmental burdens from energy-hungry data centers, cybersecurity vulnerabilities, and persistent legitimacy and trust deficits. Because a single incident can trigger several of these concerns at once, the author treats them as systemic and interconnected rather than isolated technical problems.

Vendor dependence receives particular attention as an emerging political-economy risk. Procurement arrangements often give public agencies limited access to source code, model documentation, training data and audit trails, weakening data portability and institutional autonomy. Over time, functions that should remain under public authority can become dependent on a small number of dominant technology firms, shifting power, expertise and infrastructure ownership toward the private sector. The paper also flags normative drift, in which the ethical principles embedded in AI systems subtly shift over time without detection, and accountability erosion in multi-stakeholder settings where responsibility is so dispersed that no one can be held answerable when systems fail.

On the theoretical side, the framework integrates three traditions: digital era governance, which explains the movement toward data-driven, reintegrated public services; polycentric governance, drawn from Elinor Ostrom’s work on multiple overlapping centers of authority; and adaptive governance, which emphasizes adjustment under uncertainty. The author extends digital era governance for the AI era by treating hybrid human-AI judgment as an institutional design issue rather than a technical choice. Evidence cited from recent research suggests that human-AI ensembles can improve decision-making and reduce bias, but only when humans retain decision authority and AI inputs remain explainable, reviewable and contestable, supported by audit trails, bias checks, human override and appeal mechanisms.

The practical core of the paper is a six-step adaptive, ethical and responsible AI governance framework that embeds risk management and ethics-by-design across the AI lifecycle. It begins with policy coordination, aligning goals across agencies, levels of government and stakeholders to prevent fragmented rulemaking. Resource allocation follows, building the technical infrastructure, civil-service expertise and independent testing capacity that many public agencies currently lack. Implementation then deploys AI systems with appropriate human oversight and sector-specific standards, such as clinical validation and bias testing in healthcare. Monitoring and evaluation institutionalizes audits, transparency reports and impact assessments, while feedback and learning introduces regulatory sandboxes, periodic reviews and participatory mechanisms that let governance evolve with the technology. The cycle closes with policy adjustment, recalibrating legal and organizational arrangements, including procurement rules, to rebalance public-private power.

To show how the framework works in practice, the study applies it illustratively to two very different contexts: New York City as a decentralized smart city and Singapore as a centralized smart nation. New York has built coordination through its Office of Technology and Innovation, an AI Action Plan, and the earlier Automated Decision Systems Task Force, and in late 2025 the City Council passed the GUARD Act establishing an independent Office of Algorithmic Accountability. Singapore coordinates through its Smart Nation structures, the National AI Strategy 2.0 and the Model AI Governance Framework, supported by GovTech’s shared infrastructure. The comparison reveals a striking trade-off: Singapore’s centralized architecture achieves coherence and rapid strategy updates but offers fewer channels for direct citizen participation and independent redress, while New York’s fragmented system produces uneven capacity but denser external accountability through legislation, civil society advocacy and open data. The framework, the author stresses, is structure-agnostic; it helps each model locate its own strengths and gaps.

The paper is candid about its limits. The framework is conceptual, derived from an integrative literature review rather than primary empirical research, and the city applications are illustrative rather than evaluative. Jurisdictional variability, bureaucratic inertia and uneven resources mean it requires context-sensitive adaptation, and poorly managed compliance regimes could create procedural burden without improving substantive accountability. Still, the concluding message lands with force: governance arrangements themselves can generate new risks over time, so AI oversight must remain adaptive rather than fixed. Even highly capable systems, including agentic AI, remain forms of narrow intelligence operating within bounded goals, but they are already reshaping decision pathways, and preparing for hypothetical artificial general intelligence should be treated as a stress test for existing institutions. The window for establishing robust guardrails, Kaiser argues, is not a future concern but an immediate one, and the health of democratic governance in smart cities and nations may depend on whether institutions learn as fast as the technologies they are trying to steer.

Subject of Research: Adaptive, ethical and responsible governance frameworks for AI in smart cities and nations

Article Title: Adaptive, ethical and responsible AI governance for smart cities and nations

Article References: Kaiser, Z. R. M. A. (2026). Adaptive, ethical and responsible AI governance for smart cities and nations. Discover Cities, 3(1), Article 172. https://doi.org/10.1007/s44327-026-00349-2

Image Credits: AI Generated

DOI: 10.1007/s44327-026-00349-2

Keywords: AI governance, smart cities, agentic AI, algorithmic bias, digital era governance, polycentric governance, vendor dependence, public trust, ethics-by-design, New York City, Singapore, adaptive governance

The analysis adopts a conceptual, policy-informed approach, synthesizing scholarly literature, European regulatory frameworks, and implementation evidence from international case studies. Rather than cataloguing gadgets, the authors map how AI is reshaping four interdependent pillars of education: curricular content, teaching paradigms, assessment systems, and governance structures. The systemic framing is deliberate. The authors argue that piecemeal adoption, a chatbot here, an analytics dashboard there, produces fragmented benefits and uneven risks, while genuine transformation requires coherent change across all four pillars simultaneously. Where one pillar lags, they warn, the others can amplify existing inequalities instead of reducing them.

The first pillar, curricular content, is undergoing a quiet but profound revision. AI literacy is emerging as a foundational competence that cannot remain the preserve of computer science electives. Drawing on initiatives such as the Informatics for All strategy and the European Commission’s STEM education strategic plan, the review argues that students need not only technical skills but also the capacity to critically interrogate algorithmic systems: to understand how training data shapes outputs, where models fail, and how automated recommendations can encode bias. Media literacy and data literacy, long treated as adjacent add-ons, are repositioned as core components of citizenship in dataf societies. The authors cite scholarship on datafication to stress that curricula must prepare young people to live with, and question, systems that increasingly classify and sort them.

The second pillar concerns teaching paradigms. The review rejects both utopian narratives of AI-powered personalization and dystopian visions of replaced teachers, aligning instead with scholarship that asks whether machines should replace teachers at all. Intelligent tutoring systems and adaptive platforms can tailor pacing and feedback to individual learners, but the authors emphasize that teaching is a relational, ethical, and world-centered practice that no optimization engine can replicate. Evidence from a teacher choices trial on ChatGPT use in lesson preparation illustrates a more plausible near future: AI as a productivity assistant that relieves teachers of administrative burden, freeing time for the human dimensions of pedagogy. Crucially, the paper argues that teacher agency must be protected through human-in-the-loop safeguards, ensuring that educators retain final judgment over instructional decisions and that automation supports rather than supplants professional expertise.

Assessment forms the third pillar, and arguably the one under the greatest strain. The authors draw on the OECD’s multi-volume work on AI and the future of skills, which maps the capabilities of contemporary AI systems against the competencies schools traditionally certify. As generative models demonstrate fluency in essay writing, problem solving, and knowledge recall, the validity of conventional testing regimes comes into question. Educational data mining and learning analytics promise continuous, formative assessment capable of tracking growth in real time, but the review cautions that measurement is never neutral. Big data-driven education carries structural consequences: when algorithms define what is observable and valued, narrower and more quantifiable skills can crowd out creativity, collaboration, and critical reflection. Diagnostic tools face scrutiny too, with the paper noting research investigating the validity and reliability of AI-based testing products, a reminder that procurement decisions in schools carry psychometric as well as ethical weight.

Governance, the fourth pillar, is where the paper’s policy analysis becomes most pointed. The European Union’s AI Act, Regulation 2024/1689, establishes a risk-based framework that touches education directly, and the review examines how schools can operationalize its requirements alongside the Digital Education Action Plan 2021-2027 and the Commission’s ethical guidelines on AI and data in teaching and learning. The authors argue for ethics-by-design, embedding fairness, transparency, and accountability into systems before deployment rather than patching problems afterward. They also flag a persistent gap: robust mechanisms for algorithmic accountability in schools remain scarce. Studies of automated inequality and data colonization demonstrate that when high-tech tools profile and sort vulnerable populations without oversight, the harms fall hardest on those least able to contest them. Governance must therefore include auditability, contestability, and clear lines of human responsibility.

Equity threads through every pillar. The review documents persistent digital inequalities, in access to devices and connectivity, in the quality of AI tools available to different schools, and in the preparedness of teachers to deploy them well. Case-based insights from systems as varied as China’s national AI education strategy and India’s DIKSHA digital infrastructure show that implementation models diverge widely, and that scale does not guarantee inclusion. The OECD’s policy surveys on schooling in the digital age similarly reveal uneven national readiness. Without deliberate intervention, the authors warn, AI risks becoming an accelerant of educational stratification, benefiting affluent schools with strong digital capacity while leaving under-resourced systems with superficial adoption and weakened oversight.

Teacher preparation emerges as a decisive variable. The paper synthesizes evidence on ethical-digital competencies for educators and on the challenges facing educational leaders navigating AI adoption, concluding that capacity building is not a training afterthought but a precondition for responsible system-level implementation. Teachers need structured opportunities to develop technical understanding, ethical judgment, and the confidence to question vendor claims. School leaders, meanwhile, require support to make procurement and governance decisions grounded in evidence rather than marketing hype. The authors’ practical guidance distills policy-relevant recommendations on professional development, human-in-the-loop protocols, and institutional ethics frameworks that can be enacted without waiting for perfect technology or perfect regulation.

The paper’s central argument is ultimately about values. Rejecting the framing of AI as a mere driver of automation, Mobilio and Guglielmini call for a transformative approach rooted in equity, human agency, and democratic values. If schooling is reconceived around what humans and machines each do best, then AI can expand opportunity: offering personalized support to struggling learners, giving teachers richer insight into student progress, and making governance more transparent and accountable. But that outcome is conditional, not automatic. It depends on coherent policy infrastructure, empowered teachers, and sustained attention to the students most at risk of being left behind.

The conclusions carry a clear warning and a clear invitation. Warning: absent deliberate design, algorithmic systems will quietly redefine learning in ways that serve commercial and administrative interests rather than educational ones, reinforcing the very inequalities schools exist to overcome. Invitation: when technological innovation is aligned with inclusive, ethical, and future-oriented aims, schools can ensure that AI contributes to social justice rather than undermining it. The four pillars, curriculum, pedagogy, assessment, and governance, stand or fall together, and the authors argue that the moment to reinforce them collectively is now, while the technology is still young enough to be shaped by the values of the institutions that deploy it.

Subject of Research: The transformative impact of artificial intelligence on school education across curriculum, pedagogy, assessment, and governance, with emphasis on equity and ethics.

Article Title: Rethinking Schooling in the Age of AI: Equity, Ethics, and the Four Pillars of Transformation

Article References: Mobilio, V., & Guglielmini, G. (2026). Rethinking Schooling in the Age of AI: Equity, Ethics, and the Four Pillars of Transformation. Frontiers of Digital Education, 3(1), Article 7. https://doi.org/10.1007/s44366-026-0081-3

Image Credits: AI Generated

DOI: 10.1007/s44366-026-0081-3

Keywords: AI in education, school transformation, educational equity, pedagogical innovation, educational governance, assessment systems, AI literacy, ethics-by-design, teacher empowerment, digital inequality, education policy, Rethinking