AI and ethics最新文献_第3页

How close is too close? Digital replicas and protection of the “unfixed” voice 多近才算太近？数字复制和“不固定”声音的保护

AI and ethics Pub Date : 2026-10-07 DOI: 10.1007/s43681-026-01323-9

Jennifer Chubb, Peter S. Harrison, James Tompkinson, Emily Thompson

{"title":"How close is too close? Digital replicas and protection of the “unfixed” voice","authors":"Jennifer Chubb, Peter S. Harrison, James Tompkinson, Emily Thompson","doi":"10.1007/s43681-026-01323-9","DOIUrl":"10.1007/s43681-026-01323-9","url":null,"abstract":"<div><p>Human voices provide information relating to identity, social traits and emotional markers. Scholars emphasise the intrinsic connection between voices, human cultural habits and self-expression. The development of AI-generated digital replicas means that the voice is no longer a uniquely human characteristic. This paper considers the legal, social and linguistic challenges posed by digital voice replicas. Owing to their prevalence in the creative industries, we draw on qualitative focus group data comprising the views of researchers, creatives and IP practitioners. Our findings illustrate a range of responses including skepticism, uncertainty and fear about the loss of authenticity. These responses serve as a provocation as to what happens to authenticity and ownership of the human voice. We find digital replicas evoke a strong emotional response. The voice means a lot to people. Against this backdrop, we highlight the urgent need to address the legal vulnerability of the “un-fixed” voice, (a voice not permanently captured in traditional recordings), and the inadequacy of current United Kingdom (UK) Intellectual Property (IP) frameworks to protect it. This highlights a significant regulatory lacuna which calls for a more robust protection of a ‘personality right’ in UK law.</p></div>","PeriodicalId":72137,"journal":{"name":"AI and ethics","volume":"6 6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149098075","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Semantic indeterminacy in agentic negotiation: agreement after the recession of the principal 代理谈判中的语义不确定性：委托人退避后的协议

AI and ethics Pub Date : 2026-10-07 DOI: 10.1007/s43681-026-01436-1

Shun Okuhara

{"title":"Semantic indeterminacy in agentic negotiation: agreement after the recession of the principal","authors":"Shun Okuhara","doi":"10.1007/s43681-026-01436-1","DOIUrl":"10.1007/s43681-026-01436-1","url":null,"abstract":"<div><p>Agentic AI systems increasingly do not merely respond within negotiations but participate in constituting them: they can add issues, instantiate participant agents, and modify procedural rules while an interaction is under way. Human action can be compositionally produced without thereby becoming authorless, and legal institutions routinely attach consequences to conduct without identifying an author of every contribution. Our claim is therefore limited: agentic generation can separate three relations that institutions often treat as aligned: causal production, intentional expression, and institutional attribution. We call the resulting condition <i>semantic indeterminacy</i>. An added element may result from several generative influences, including principal intention, agent-side optimization, training distributions, and interactional contingency, while information about its causal production does not, by itself, determine whose intention it expresses or how it is institutionally attributed. We show how this condition can appear across three sites of bargaining: issues, participants, and procedural rules. We then trace two implications. First, causal analysis alone does not determine who is responsible for a generated element or on what basis, although contract, agency, and product-liability doctrines can assign legal consequences under specified conditions. Second, as a limiting case, agentic interactions may occur in <i>post-principal agentic spaces</i>, where no human provides move-specific direction and no identifiable human or institution occupies a legally or institutionally recognized role. Existing governance frameworks remain grounded in identifiable human authority and accountability; their limits become visible where that ground is no longer reliably available. The paper provides a conceptual framework for identifying this residual problem rather than claiming to solve it.</p></div>","PeriodicalId":72137,"journal":{"name":"AI and ethics","volume":"6 6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s43681-026-01436-1.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149098074","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Animal welfare in the world of digitalization: Computer Vision in the context of foundational structures 数字化世界中的动物福利：基础结构背景下的计算机视觉

AI and ethics Pub Date : 2026-10-06 DOI: 10.1007/s43681-026-01417-4

Mariska Thalitha Bosschaert, Bart Gremmen

{"title":"Animal welfare in the world of digitalization: Computer Vision in the context of foundational structures","authors":"Mariska Thalitha Bosschaert, Bart Gremmen","doi":"10.1007/s43681-026-01417-4","DOIUrl":"10.1007/s43681-026-01417-4","url":null,"abstract":"<div><p>Computer Vision is an emerging technology within Precision Livestock Farming (PLF), combining camera systems with artificial intelligence to monitor animal welfare. While animal ethicists recognize its potential benefits, they also raise concerns about possible negative implications for animal welfare. This paper argues that addressing ethical concerns cannot rely on solutions targeting isolated ethical risks; rather, potential solutions must account for the technology’s embeddedness within society-wide structures. The focus in this paper will be on foundational structures, these structures concern how the world is experienced and understood, what is taken as self-evident. Over centuries, humans have increasingly delegated tasks to machines and standardized the world in doing so. This process has fostered growing trust in the outcomes of (digital) technologies. These developments have changed human experience and understanding of the world in various ways. Farmers, as members of society, are unlikely to be exempt from this trend. Therefore, reflection on Computer Vision must go beyond the farming practice and take into account these foundational structures that are currently reshaping society at large. Computer Vision examined as embedded within foundational structures demonstrates that ethical issues are also embedded within these structures, consequently, potential solutions must take account of them to increase their effectiveness.</p></div>","PeriodicalId":72137,"journal":{"name":"AI and ethics","volume":"6 6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-10-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s43681-026-01417-4.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149098665","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Whose responsibility? Deconstructing global responsible AI from the African lens: a review 谁的责任?从非洲视角解构全球负责任的人工智能：综述

AI and ethics Pub Date : 2026-10-06 DOI: 10.1007/s43681-026-01418-3

Damian Eke, Makuochi Nkwo, Ibifuro Robert Jaja, Muhammad Adamu, Simisola Akintoye, Michael Ugom, Sarah Kiden

{"title":"Whose responsibility? Deconstructing global responsible AI from the African lens: a review","authors":"Damian Eke, Makuochi Nkwo, Ibifuro Robert Jaja, Muhammad Adamu, Simisola Akintoye, Michael Ugom, Sarah Kiden","doi":"10.1007/s43681-026-01418-3","DOIUrl":"10.1007/s43681-026-01418-3","url":null,"abstract":"<div><p>This paper uses the theory of decoloniality to critically examine how African scholars have interrogated the dominant, global narratives surrounding responsible Artificial Intelligence (AI). Through a multivocal literature review, the paper reveals a range of critical perspectives that expose systemic exclusions and imbalances in the current responsible AI discourse. Key themes identified include the marginalisation of African epistemologies, values, and principles; the absence of representative African datasets, lack of locally grounded regulatory frameworks; the limited inclusion and or availability of African expertise and funding; and the persistent power asymmetries and coloniality embedded in global responsible AI ethics and governance. The review highlights emerging efforts towards reconstructing responsible AI from an African contextual lens. Central to this constructive turn is the proposal of a decoloniality framework that puts emphasis on African knowledge systems, reclaims agency in AI development, and reimagines responsible AI in ways that are contextually grounded, pluriversal, and socially acceptable. This review not only contributes to critical AI scholarship but also charts pathways for more inclusive and equitable AI futures on the continent.</p></div>","PeriodicalId":72137,"journal":{"name":"AI and ethics","volume":"6 6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-10-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s43681-026-01418-3.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149098072","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A pluralistic view on AI ethics: perspectives from Buen Vivir 人工智能伦理的多元观：来自《幸福生活》的视角

AI and ethics Pub Date : 2026-10-06 DOI: 10.1007/s43681-026-01403-w

Dorine Eva van Norren

引用次数: 0

Objective blindness in agentic systems: a taxonomy of behavioral signatures 代理系统中的客观盲目性：行为特征的分类

AI and ethics Pub Date : 2026-10-05 DOI: 10.1007/s43681-026-01441-4

Fabio Massimo Conte

{"title":"Objective blindness in agentic systems: a taxonomy of behavioral signatures","authors":"Fabio Massimo Conte","doi":"10.1007/s43681-026-01441-4","DOIUrl":"10.1007/s43681-026-01441-4","url":null,"abstract":"<div><p>Can a system be held accountable for a harm it was structurally incapable of representing as a reason for action? This paper argues that a specific class of AI alignment failures—termed objective blindness—creates a critical evidentiary gap that complicates this question. Objective blindness occurs when a system faithfully optimizes its formal objective, but its representational architecture omits contextual variables that human observers regard as normatively relevant. Unlike specification gaming, this failure produces outputs that are formally flawless, masking the underlying omission from standard behavioral oversight. We argue that this creates a distinctive epistemic accountability gap: while moral responsibility remains traceable to human designers in principle, the invisibility of the omission practically shields them from scrutiny, especially in complex “many hands” institutional settings. The paper further argues that regulatory environments are especially exposed to this failure. To address this, we propose a taxonomy of four behavioral signatures for detecting objective blindness in practice, illustrated with reference to Anti-Money Laundering compliance. We close by examining the ethical constraints that any detection infrastructure must satisfy, including risks of false attribution and the privacy costs of authorship-profiling techniques, and outline directions for future empirical validation.</p></div>","PeriodicalId":72137,"journal":{"name":"AI and ethics","volume":"6 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132349","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Beyond one-size-fits-all: how adaptability and autonomy shape responsible AI governance 超越一刀切：适应性和自主性如何塑造负责任的人工智能治理

AI and ethics Pub Date : 2026-10-05 DOI: 10.1007/s43681-026-01439-y

Mohammad Hossein Jarrahi, Vern L. Glaser, Joel Gehman

{"title":"Beyond one-size-fits-all: how adaptability and autonomy shape responsible AI governance","authors":"Mohammad Hossein Jarrahi, Vern L. Glaser, Joel Gehman","doi":"10.1007/s43681-026-01439-y","DOIUrl":"10.1007/s43681-026-01439-y","url":null,"abstract":"<div><p>Artificial intelligence (AI) governance is a critical safeguard against risk and ethical harm. However, organizations need clearer guidance on how oversight should vary across an AI portfolio. Building on characteristics recognized in scholarly work, we organize governance around two dimensions: adaptability, the degree to which a system’s behavior changes through learning after deployment; and autonomy, the degree to which it selects or executes consequential action without contemporaneous human direction. Adaptability can make earlier validation and documentation progressively less representative of the system in use, while autonomy enables consequential actions without immediate human discretion and raises questions about intervention and accountability. We develop a two-dimensional model that distinguishes these governance demands and identifies when they become recursively compounded: system action shapes what is subsequently observed, those observations inform post-deployment learning, and the resulting adaptation changes subsequent actions. In these situations, the governance of action and learning can no longer proceed independently. We extend this analysis to portfolio interdependence, showing why organizations need to trace shared dependencies, assign responsibility, and coordinate review when a change or failure in a shared component affects multiple AI applications. Our model presents responsible AI governance as an organizational capacity for matching oversight to changing systems, delegated action, and interdependencies across an AI portfolio.</p></div>","PeriodicalId":72137,"journal":{"name":"AI and ethics","volume":"6 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132350","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The ethical half-life of AI: moral decay outpaces technical obsolescence 人工智能的伦理半衰期：道德衰退超过技术过时

AI and ethics Pub Date : 2026-10-05 DOI: 10.1007/s43681-026-01428-1

Lu Chao, XiaoXi Ma

{"title":"The ethical half-life of AI: moral decay outpaces technical obsolescence","authors":"Lu Chao, XiaoXi Ma","doi":"10.1007/s43681-026-01428-1","DOIUrl":"10.1007/s43681-026-01428-1","url":null,"abstract":"<div><p>AI ethics has been built on an assumption it rarely defends: that a system’s ethical properties are fixed at deployment and remain valid as the world moves on. This paper argues that the assumption fails, that the ethical standing of a deployed model has a measurable half-life, and that this moral decay typically arrives before technical obsolescence. We integrate four strands of literature on model drift, value drift, incident monitoring, and lifecycle governance into one account built on five decay mechanisms: normative drift, input drift, exposure erosion, institutional neglect, and model implosion. A formal model defines ethical integrity as a decaying stock, derives the half-life, links it to an estimable hazard, yielding four testable hypotheses. Using 1597 incidents from the AI Incident Database (2012–2026), we estimate an ethical half-life of 2.95 years (95% CI 2.46–3.68). An event study around the launch of ChatGPT shows normative harms rising 1.44 log points with no detectable pre-trend while technical failures declined, evidence that moral decay outpaces technical decay. Temporal holdout tests show decay-aware models improve forecasts and, more importantly, produce better audit decisions: risk-based scheduling captures more than twice the ethical loss of uniform allocation when capacity is scarce. Reporting frequency and accuracy can serve as an early warning signal, and we treat them as the first instrument of a re-certification regime. The results survive placebo breaks at alternative dates across the series, weight permutations, and clustered inference. We close with a governance agenda built around ethical re-certification rather than birth certification.</p></div>","PeriodicalId":72137,"journal":{"name":"AI and ethics","volume":"6 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132340","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

“Us” and “them” and the “moral tribes” of AI: new theorizing of the human–machine relationship through a dual-process theory of moral cognition 人工智能的“我们”、“他们”和“道德部落”：通过道德认知双过程理论对人机关系的新理论化

AI and ethics Pub Date : 2026-10-05 DOI: 10.1007/s43681-026-01420-9

Mario D. Schultz, Peter Seele

{"title":"“Us” and “them” and the “moral tribes” of AI: new theorizing of the human–machine relationship through a dual-process theory of moral cognition","authors":"Mario D. Schultz, Peter Seele","doi":"10.1007/s43681-026-01420-9","DOIUrl":"10.1007/s43681-026-01420-9","url":null,"abstract":"<div><p>This paper introduces a dual-process approach grounded in moral cognition theory to theorize the human-AI relationship. As AI systems become increasingly integrated into daily life and their societal and environmental impacts expand, debates over a distinct AI ethics and the moral status of AI systems have intensified. While AI ethics guidelines typically fail to address the complexities of evolving human-AI interactions, the field is characterized by a polarized theoretical landscape: classical ethics theories and post-anthropocentric approaches diverge on how to address these interactions. The paper proposes a dual-process theory of moral cognition as a bridging lens to theorize how humans often intuitively view AI as an ‘other tribe,’ while more in-depth reasoning about AI as a human-made, deployed system tends to arise only later as a secondary reflection. Using the 'dual-process' concept of “moral tribes” from J. Greene to distinguish between automatic/intuitive (mode 1) and manual/deliberative (mode 2) cognition, we highlight that ethical reasoning about AI is neither purely intuitive (e.g., fear or love) nor purely rational (e.g., abstract principle application), but results from a dynamic interplay of the modes, which shape the salience of ethical reflection about AI in a given moment. This dual-process perspective offers a dynamic framework for examining human-AI interactions, underscoring the challenge of transcending tribalistic biases while critically evaluating relationships with AI systems. The paper emphasizes the need for a nuanced understanding of human-AI relations, beyond static perspectives, calling for attention to cognitive limitations and bias exploitation, while outlining pathways for future research across applied ethics domains, including business, military, theological, and medical contexts.</p></div>","PeriodicalId":72137,"journal":{"name":"AI and ethics","volume":"6 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132357","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

AI auditing through professional tools: an exploratory qualitative case study of a digital ethics training program 通过专业工具进行人工智能审计：数字道德培训计划的探索性定性案例研究

AI and ethics Pub Date : 2026-10-05 DOI: 10.1007/s43681-026-01423-6

Mirko Farina, Andrea Lavazza, Sergei Kladko

{"title":"AI auditing through professional tools: an exploratory qualitative case study of a digital ethics training program","authors":"Mirko Farina,  Andrea Lavazza, Sergei Kladko","doi":"10.1007/s43681-026-01423-6","DOIUrl":"10.1007/s43681-026-01423-6","url":null,"abstract":"<div><p>This article examines the applicability of traditional audit tools to the ethical assessment of artificial intelligence (AI) systems through an exercise conducted at the 2024 School of Digital Ethics. Interdisciplinary participant teams applied a range of widely used analytical instruments—including SWOT analysis, risk matrices, risk maps, process mapping, and GAP analysis—to evaluate AI-based scenarios. The study adopted an exploratory qualitative case study design, drawing on participant surveys, facilitator observations, written audit reports, and final team presentations. An interpretive and comparative analysis was conducted to identify recurring patterns across the teams regarding the strengths, limitations, methodological adaptations, and perceived practical applicability of each auditing tool. The findings demonstrate that while these tools provide a useful structural foundation for organizing ethical inquiry, their applicability is constrained by the distinctive characteristics of AI systems, including non-determinism, emergent behavior, and lifecycle complexity. Across the teams, process mapping proved particularly useful for identifying ethical risks at different stages of system development, while SWOT analysis and risk matrices facilitated the initial structuring of discussions but presented difficulties in prioritization and in addressing qualitative ethical values. GAP analysis supported the identification of discrepancies between existing practices and normative benchmarks but remained dependent on the availability of operationalized standards. Rather than evaluating the effectiveness of the educational program itself, the study provides empirical evidence of how traditional auditing tools were interpreted and adapted within an interdisciplinary learning environment designed to simulate AI ethics auditing. The findings highlight the need for adaptive, context-sensitive auditing approaches that combine multiple tools and integrate ethical principles into dynamic, lifecycle-oriented evaluation processes.</p></div>","PeriodicalId":72137,"journal":{"name":"AI and ethics","volume":"6 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132351","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0