Artificial intelligence as a moral agent: regulatory implications and a relational-contextual extension of Moor's classification
Artificial intelligence as a moral agent: regulatory implications and a relational-contextual extension of Moor's classification

Summary/Abstract: This paper reassesses the regulatory value of James Moor's four-level typology of machine morality, considering the European Artificial Intelligence Act (AI Act) and the forthcoming European Union (EU) liability directives. It asks whether Moors categories--ethical impact, implicit, explicit, and full moral agents--still capture the morally relevant properties of today's generative, adaptive AI, and, if not, whether adding a relational−contextual dimension can better anticipate responsibility gaps. To address this gap, we introduce a novel relational-contextual dimension and a three-factor Responsibility Index (RI₃) that refines Moor's typology by cross-classifying AI systems according to complexity, autonomy, and behavioural predictability for regulatory use. Adopting a strictly conceptual design, the study combines analytic philosophy with illustrative comparisons drawn from recent EU policy debates and high-profile incidents. It refines key terms, tests their coherence against statutory risk tiers, and distils the analysis into a three-factor matrix--complexity, autonomy, and predictability--that can be operationalised by lawmakers. The evaluation confirms that Moor's typology remains a valuable baseline for distinguishing between passive and decision-making artefacts. Nevertheless, it also highlights how moral accountability is distributed within socio-technical networks. The proposed relational−contextual dimension, in conjunction with the regulatory matrix, aligns more closely with the AI Act's risk logic and highlights scenarios in which moral agency is effectively delegated to the system. Moor's framework should be retained but augmented: only by integrating relational criteria can legislators close emerging accountability gaps surrounding large-scale, autonomous AI. The matrix offers a pragmatic tool for aligning philosophical insight with concrete legal duties.

• Details
• Contents