Ethics-Based Auditing of Automated Decision-Making Systems: Nature, Scope, and Limitations.

Abstract: Switzerland’s long-standing policy of neutrality, anchored in the 1815 Congress of Vienna and codified in the 1907 Hague Conventions, is facing unprecedented tests in the age of artificial intelligence (AI). The integration of AI into military data analysis, cyber defense, and logistical operations raises profound questions about how a neutral state can maintain its non-aligned status while safeguarding national security. This study examines the tension between Switzerland’s neutrality doctrine and its expanding investments in AI-driven defense capabilities, with a particular focus on the 2022 Swiss Federal Department of Defence AI Strategy. It analyses the regulatory, technological, and diplomatic dimensions of neutrality in the digital era, drawing on international humanitarian law, cybersecurity standards, and case studies from NATO cooperation, the Russia–Ukraine war, and dual-use technology exports. The paper proposes a three-pillar framework for reconciling innovation with neutrality: (1) Technological self-regulation through a “Red-Line AI” model that prohibits offensive algorithms and mandates quantum-resistant encryption; (2) Diplomatic leadership via a digital extension of the Geneva Conventions and a United Nations “Digital Neutrality” initiative to codify responsible AI use by neutral states; and (3) Societal engagement through public oversight boards and annual neutrality impact reports. The findings suggest that neutrality in the 21st century must evolve from a purely geographic principle to a functional and technological ethic, embedding legal and ethical safeguards directly into AI system design and governance. By adopting these measures, Switzerland can maintain its defensive readiness without compromising its non-alignment, while contributing to the development of international norms for (Nuredin, 2016) responsible AI in military contexts. This approach reframes neutrality as an active form of global stewardship in emerging security domains, positioning Switzerland as both a beneficiary and a custodian of stability in the AI era.

TL;DR: This study examines the European law's applicability to algorithms used in public sector automated decision-making, highlighting discrepancies in EU and ECHR jurisprudence regarding access to information and official documents.

TL;DR: Previous efforts to define explainability in Machine Learning are summarized, establishing a novel definition that covers prior conceptual propositions with a major focus on the audience for which explainability is sought, and a taxonomy of recent contributions related to the explainability of different Machine Learning models are proposed.

TL;DR: In this paper, a taxonomy of recent contributions related to explainability of different machine learning models, including those aimed at explaining Deep Learning methods, is presented, and a second dedicated taxonomy is built and examined in detail.

TL;DR: In this article, a global convergence emerging around five ethical principles (transparency, justice and fairness, non-maleficence, responsibility and privacy), with substantive divergence in relation to how these principles are interpreted, why they are deemed important, what issue, domain or actors they pertain to, and how they should be implemented.

TL;DR: It is suggested data controllers should offer a particular type of explanation, unconditional counterfactual explanations, to support these three aims, which describe the smallest change to the world that can be made to obtain a desirable outcome, or to arrive at the closest possible world, without needing to explain the internal logic of the system.