A reframing of AI unaccountability as a constitutive property of sociotechnical systems, an extension of the four barriers to accountability, and a practical instrument for identifying accountability voids in specific AI deployments are contributed.
Abstract
The increasing deployment of autonomous, agentic AI systems challenges traditional accountability mechanisms. Existing research predominantly frames AI accountability gaps as barriers that can be overcome through better standards, transparency, and institutional reform. We argue that this framing is insufficient: certain configurations of actors, systems, and institutions render AI accountability conceptually unachievable regardless of effort. We introduce the concept of constitutive AI unaccountability to capture these configurations. Through a three-stage qualitative study comprising a concept-centric literature analysis, a secondary analysis of 27 expert interviews with AI professionals from technical, legal, and sociotechnical backgrounds, and an illustrative framework application to the open-source agentic AI system OpenClaw, we identify nine categories and 20 themes of constitutive AI unaccountability. These are organized across structural, technological, and normative clusters and reinforce one another through eight directed interdependencies. Our framework is operationalized as a diagnostic instrument of 20 questions, which detected 17 of 20 conditions when applied to OpenClaw, including an inverted anthropomorphism configuration in which the AI agent was the only identifiable actor. We contribute a reframing of AI unaccountability as a constitutive property of sociotechnical systems, an extension of the four barriers to accountability, and a practical instrument for identifying accountability voids in specific AI deployments.
The widespread adoption of Artificial Intelligence (AI) has led organizations to establish formal governance frameworks aimed at mitigating ethical, legal, and operational risks. Despite these efforts, AI governance frequently fails in practice, as evidenced by the growing prevalence of Shadow AI the unsanctioned use of AI tools by employees. Existing scholarly and practitioner discourses predominantly frame this phenomenon as a compliance failure or security vulnerability, thereby emphasizing stricter controls and enhanced employee training as primary remedies. This conceptual study challenges that prevailing view by arguing that Shadow AI represents a structural manifestation of policy–practice misalignment rather than a problem of individual deviance. The study develops a diagnostic framework that identifies three constitutive dimensions of misalignment: temporal gaps (mismatches between governance processes and operational speed), utility gaps (misalignment between sanctioned tools and task-specific needs), and autonomy–control gaps (tensions between professional discretion and standardization). Drawing on a theory-driven conceptual methodology integrating sociotechnical systems theory with policy–practice analysis, and illustrated through structured synthetic organizational scenarios, the study demonstrates how governance designs that overlook the realities of situated work systematically generate Shadow AI practices. The analysis further suggests that adaptive governance models incorporating structured flexibility such as curated AI tool marketplaces and expedited approval pathways are theoretically more effective than highly rigid governance regimes. The primary contribution lies in advancing a practice-aware AI governance model that reframes Shadow AI as a diagnostic signal of systemic design flaws and provides a foundation for more legitimate and responsive AI governance.
Mia Wilson, Ethan Moore· Journal of Management and In...· 0 citations
The Qualitative Model of Ethics (QME), a naturalistic and teleological framework grounded in a single evaluative scalar: the generative capacity of the Whole Living System (WLS), provides a unified and operational framework for AI governance at scale.
Existing AI governance frameworks address technical alignment, data protection, and regulatory compliance, but not the interior developmental capacity of the humans who govern AI-influenced decisions: the human readiness problem. This paper presents the Emergence Mirror Framework (EMF), an architecture that integrates Kegan’s constructive-developmental theory (1994), Kohlrieser’s secure base theory (2006), and Scharmer’s Theory U (2009) into a unified seven-layer developmental scaffold oriented toward discernment as the apex governance capacity. The EMF advances a specific, testable claim: Kegan’s fourth-order self-authoring mind constitutes the minimum developmental threshold for effective AI stewardship, operationalising what human rights and legal scholarship has theorised but not rendered actionable. The framework was developed through a seven-month phenomenological inquiry and a structured proof-of-concept study with 22 participants across four cohorts (n = 22; descriptive, not statistically generalisable), using multi-source triangulation of developmental surveys, AI dialogue transcripts, and facilitator observations. Participants ranged from late third-order to mid fourth-order profiles, testing the framework across the readiness gap it identifies. Nineteen of 22 participants retained decision sovereignty, 18 of 22 demonstrated early interrogation of AI outputs, and 20 of 22 articulated sovereignty boundaries, consistent with the moral performance indicators the EMF predicts (Kristjánsson et al., Rev Gen Psychol 25:239-255, 2021). A participant at the third-to-fourth order transition experienced the predicted developmental friction whilst governance integrity remained intact. Individual posture profiles, visualised through a Leadership Posture Heat Map (Sect. 4.3.4), reveal variation aggregate statistics alone cannot capture. The paper further demonstrates that an AI mirror as reported by Vallor (The AI Mirror: How to Reclaim Our Humanity in an Age of Machine Thinking, Oxford University Press, Oxford, 2024) can function as a developmental instrument: relationally grounded engagement oriented toward resonance, rather than replacement, can deepen discernment and strengthen sovereign decision-making rather than producing moral deskilling.
Arulnageswaran Aruleswaran, Thaatchaayini Kananatu· AI and Ethics· 1 citation
The paper’s central argumentative shift is to change the narrative from bias mitigation to bias management—treating bias not as a defect to be corrected but as an ongoing condition to be governed.
Gabriela Arriagada-Bruneau· Science and Engineering Ethi...· 0 citations
AI governance instruments are proliferating, and so are their difficulties. Across major jurisdictions and international bodies, reform efforts built on substantially different premises encounter a recognizably similar pattern of failure. I argue that anticipatory regulatory governance rests on three
operational premises
—categorical stability, epistemic accessibility, and manageable pace—and that AI’s emergence, opacity, and velocity violate all three in compound. These premises form a distinct layer of operational preconditions, not a complete theory of governance. Reform within the existing premises reproduces the violations they produce. I call this configuration the
reform trap
: a paradigmatic lock-in at the level of operational preconditions, distinct from path dependence and policy paradigm rigidity. The pattern is convergent across five strategies in active reform—categorical regulation, process-based management, information disclosure, normative guidance, and adaptive experimentation—and persists even in the most adaptive of them. I propose three
premise-level substitutions
—outcome observability, causal attributability, and enforcement capability—each replacing a violated precondition with a weaker, design-addressable one. These differ from outcome-based and performance-based regulation, which swaps instruments within an architecture whose premises remain stable.
There is a rising tide of initiatives countering predominant AI practices and narratives. The social and environmental consequences of AI-dedicated data centres; the death, injury and human displacement involving AI-enabled lethal autonomous weapons; and the corrosion of educational institutions via the adoption of generative AI tools are just a few targets of diverse initiatives taking a stance against AI. However, these initiatives are generally siloed; they do not constitute a single movement but a fragmented patchwork. In this paper, I draw on Gramsci’s Prison Notebooks to make the case for unity among those running such initiatives, whom I call “AI subalterns.” Drawing on literature on epistemic injustice, I confront the AI subaltern with “ruling AI authorities”; those from the ruling classes with the ability to speak about AI with authority and thus perpetrate epistemic injustice. I argue that epistemic activism constitutes a key weapon in the AI subaltern’s arsenal, and that recognising how it is wielded by peers may help develop a sense of political consciousness. I show how the resulting unity among AI subalterns meets the criteria Gramsci sets out when describing how subalterns have historically coordinated successful political uprisings.
Ismael Kherroubi Garcia· AI and Ethics· 0 citations