An Ethical Governance Framework for AI Storytelling in Cultural Narratives

This paper introduces the Right Human-in-the-Loop (R-HiTL) model, arguing that it is not merely a best practice but an ethical imperative for the responsible use of generative AI (GenAI) in cultural contexts. Our central claim is that narrative integrity arises from proactive, structured governance rather than after-the-fact fixes. Using the Named after Nelson (NaN) Podcast Series as a case study, we show that raw AI outputs are often factually unreliable, culturally misaligned, and emotionally unconvincing. We embed subject-matter expertise into the content lifecycle through three targeted ‘repair’ loops, cognitive (fact-checking and source validation), cultural (accent and language authenticity), and affective (tone and narrative resonance), to mitigate techno-linguistic bias, hallucination, and the ''emotional deficit". Across two student workshops, the framework delivered measurable improvements in factual accuracy, cultural authenticity, and audience engagement. Participants rated post-intervention outputs higher and valued the expert feedback loop, providing empirical support for R-HiTL as a practical governance mechanism. Our focus is on institutions and educators integrating AI into teaching and cultural storytelling. However, the same governance needs arise in adjacent domains. We therefore note applicability to the GLAM sector (galleries, libraries, archives and museums) and to community heritage organisations, where responsible AI use similarly depends on context-specific expertise and transparent oversight. Taken together, the findings validate R-HiTL as a workable blueprint for teams seeking to use GenAI responsibly while safeguarding narrative integrity and genuine inclusion.