<![CDATA[The development of artificial intelligence has driven a fundamental transformation in inventory audit practice, yet systematic syntheses that simultaneously integrate the dimensions of automation accuracy, assurance risk, and frameworks remain very limited. This study aims to identify the accuracy level of AI-based inventory audit systems, map the resulting assurance risk profile, and synthesize an emerging framework to guide implementation. Using a Systematic Literature Review (SLR) approach based on the PRISMA protocol, 30 studies published between 2021 and 2026 were selected from 309 articles identified through Scopus, Web of Science*, and ScienceDirect* databases. The synthesis results indicate that deep learning*, machine learning*, and computer vision-based systems achieve accuracy between 85% and 96.3% in risk classification and inventory verification, significantly outperforming conventional methods. However, the adoption of AI also introduces multidimensional assurance risks that include algorithmic bias, erosion of professional skepticism, and regulatory uncertainty. Emerging frameworks are moving toward a holistic integration of technical, ethical, and governance dimensions, although most have not been validated across industries. These findings underscore the urgency of updating international assurance standards and developing an adaptive human-in-the-loop model.]]>
Copyrights © 2026
