Bridging Ethics and Law in AI Systems with a Structured Rule Language

Artificial intelligence (AI) systems often need to follow both ethical and legal rules. Sometimes, these rules can conflict. For example, a healthcare AI may need patient consent (ethical rule), but the law might allow data sharing in emergencies (legal rule). This paper introduces a domain-specific language (DSL) to help represent and solve such conflicts. The DSL uses simple and readable syntax so that people without technical training—like ethicists or legal professionals—can write rules clearly. The DSL is automatically translated into formal Web Ontology Language (OWL) and Semantic Web Rule Language (SWRL) rules, which work with existing reasoning tools. We tested this DSL in a case study where a medical AI system had to decide whether to access patient data. The DSL helped define when consent is required and when emergency access is allowed. We compared the DSL with writing rules directly in SWRL. The DSL was easier to use, less error-prone, and just as accurate. It also made it easier to update the rules when policies change. This shows that a DSL can support better teamwork between technical and non-technical people. Beyond simplifying rule logic, the DSL improves the transparency and adaptability of normative reasoning in AI. Future work will focus on extending the language’s capabilities to address more nuanced conflicts and scaling its application across broader deployment scenarios.