ISO/IEC 42001 - Aligned Governance Evidence for High-Stakes AI Decision Systems: A Systematic Literature Review

Abstract

Artificial intelligence (AI) is increasingly deployed in high-stakes sectors such as healthcare, finance, and autonomous systems, creating a need for governance approaches that address risks related to bias, opacity, accountability, and compliance. This systematic literature review (SLR) examines how the literature aligns AI governance for high-stakes decision systems with ISO/IEC 42001:2023, with particular attention to evidence generation, auditability, and the operationalization of compliance. Using a structured search and multi-stage screening process, 39 studies were selected for analysis from an initial pool of 2,918 records. The review finds growing convergence toward risk-based, lifecycle-oriented AI governance, supported by mechanisms such as traceability, continuous monitoring, and machine-readable evidence artefacts. The literature also indicates increasing interest in automation-enabled compliance, including blockchain-based logging and agentic support for governance processes. However, evidence remains fragmented, with limited large-scale validation, inconsistent evidence formats, and uneven treatment of regional regulatory requirements. The review concludes that ISO/IEC 42001-aligned governance for high-stakes AI requires evidence-centric, auditable, and context-sensitive oversight that combines automation with human judgment.