<![CDATA[Artificial intelligence (AI) plays a critical role in supply chain systems by enabling predictive analytics and data-driven decision-making. However, increasing reliance on AI exposes systems to significant data-related vulnerabilities that may compromise reliability and trust. This study investigates three major threats to AI governance: data poisoning, data drift, and data integrity. Using a qualitative literature-based analysis supported by synthesized empirical evidence, this study evaluates the impact of these threats on model performance and operational outcomes. The results show that data poisoning can significantly reduce model accuracy (from approximately 95% to below 75%) and introduce bias, while data drift leads to gradual performance degradation over time due to changing data distributions. In addition, data integrity issues—such as incomplete, corrupted, or unauthorized data—undermine decision reliability and amplify the effects of poisoning and drift. To address these challenges, the study proposes a multi-layered AI governance framework integrating technical safeguards (e.g., adversarial detection and continuous monitoring), organizational controls, and policy-level compliance mechanisms. The findings provide practical insights for improving AI robustness, operational resilience, and trust in supply chain environments, contributing to the development of effective and responsible AI governance.]]>
Copyrights © 2026
