Governing trustworthy AI in critical communications infrastructure: an ethical and regulatory framework for 5G/6G Open RAN and network automation

Purpose This paper aims to examine how trustworthy artificial intelligence (AI) principles can be operationalised where AI is becoming an operating control layer in fifth-generation (5G) and emerging sixth-generation (6G) communications infrastructure, and to propose a governance framework bridging horizontal AI rules and telecom-specific assurance. Design/methodology/approach It uses a structured documentary review and cross-jurisdictional gap analysis of AI governance instruments, Open Radio Access Network (Open RAN) assurance literature, telecom-security standards, and policy developments in the United States, the European Union, the United Kingdom, Japan and Singapore, without primary empirical data. Findings The telecom standards stack is mature on architecture and baseline security but weaker on explainability, post-deployment AI assurance, model provenance and human-override regimes, while horizontal AI laws lack telecom-specific operational depth. The paper develops a risk taxonomy, a cross-jurisdictional gap map and TRACE-5G/6G (Tier, Register, Assure, Control, Evaluate). Research limitations/implications The framework is analytical rather than empirically validated because the study draws on documentary sources rather than primary field data or a formal systematic-review protocol; future work should test it against operational deployments. Practical implications It offers regulators, operators, integrators and standards bodies an analytical synthesis, a gap map, a mapping table and directions for future testing. Originality/value The paper organises dispersed trustworthy-AI obligations into a telecom-tailored lifecycle checklist and provides a transparent cross-jurisdictional gap map as an analytical tool for future profiling exercises.