Imagine \"InnovAI,\" a multinational corporation deploying an AI-powered predictive maintenance system for its global network of manufacturing plants. Initially, the system performed exceptionally well, minimizing downtime and optimizing resource allocation. However, after six months, subtle shifts in operational parameters and environmental conditions across different plant locations led to a gradual decline in the system\'s accuracy. To address this, InnovAI\'s engineering team implemented a series of modifications to the AI model, including retraining with new data and adjusting algorithm parameters. However, due to inadequate documentation of the original model architecture, limited traceability of the changes made, and insufficient post-deployment monitoring protocols, the modifications inadvertently introduced new biases and vulnerabilities. As a result, the system started generating inaccurate predictions, leading to increased equipment failures and operational inefficiencies. According to ISO 42001:2023, which of the following aspects of AI lifecycle management was most critically overlooked by InnovAI, leading to the system\'s performance degradation and increased risks?