Responsible AI Governance: A Systematic Literature Review

Most AI governance frameworks are incomplete: Only 5 out of 61 studies analyzed provided comprehensive answers to fundamental questions about who is accountable, what should be governed, when governance should occur, and how it should be implemented—indicating that current governance approaches often lack holistic coverage of critical governance elements.

Accountability remains unclear: The paper identifies "WHO is accountable" as a key unanswered question in AI governance literature, suggesting that organizations and policymakers still lack clear guidance on responsibility assignment for AI systems' outcomes and failures.

Organizational-level governance is critical: The findings emphasize that AI governance at the organizational level plays an especially important role in establishing effective and responsible AI practices, meaning companies cannot rely solely on external regulation and must implement internal governance structures.

Governance timing matters across the AI lifecycle: The research identifies "WHEN governance occurs" as a distinct governance dimension, highlighting that responsible AI requires oversight throughout the entire AI development process—from initial design through deployment and monitoring—rather than only at specific stages.

Multiple governance mechanisms exist but lack coordination: The paper identifies various governance approaches including frameworks, tools, standards, policies, and models, but the fragmentation across these mechanisms suggests a need for more integrated and standardized approaches to AI governance.