Hallucination in Generative AI: Challenges to data integrity, ethical concerns and implications for knowledge management

This study examines the phenomenon of hallucination in generative artificial intelligence (AI) and analyzes its implications for data integrity, ethical responsibility, and knowledge management systems. A qualitative research design was adopted using a systematic review of literature published between 2017 and 2025. Relevant studies were retrieved from academic databases including DOAJ and Google Scholar using targeted keywords related to AI hallucination, data reliability, ethics, and knowledge management. A total of 34 eligible studies were analyzed using thematic analysis to identify recurring patterns, divergences and emergent themes across the literature. The findings reveal that generative AI hallucinations significantly undermine data reliability by producing plausible but inaccurate information, thereby increasing the risk of misinformation and flawed decision-making. Ethical challenges including bias, transparency deficits, accountability gaps, and intellectual property concerns are consistently reported in the literature. While generative AI enhances efficiency, creativity, and information retrieval within knowledge management systems, the persistence of hallucinations reflects deeper structural limitations in large language models which weaken user trust and compromise knowledge accuracy and institutional credibility. This study is among the first to provide a comprehensive and integrated examination of generative AI hallucination in relation to data integrity, ethical concerns, and knowledge management systems.