Cybersecurity in the supply chain (SC) is a paramount issue as organizations have grown to rely on digitally connected, multi-tiered vendor ecosystems. While AI can enhance the ability to detect anomalies, create predictive risk scores, and respond quickly, the danger lies in relying too heavily on automated decision-making, which can lead to opacity, automation bias, and governance challenges if there is insufficient human oversight. This paper presents and tests an AI-assisted cyber threat intelligence (CTI) framework to support multi-tier supply chain cyber defense. The study has an empirical approach with a multi case design using five focal firms and extended vendor ecosystems in logistics connected environments within India, Singapore and the United Arab Emirates. The framework being proposed combines anomaly detection, graph-based threat propagation analysis, a dynamic Trust Score Index, and an Explainability layer to enable analysts to review the events and vendors to reclassify their trust. The empirical evaluation is based on 3.2 million pseudonymized log and event records as well as vendor compliance and threat intelligence data. Study is based on standardized measures of performance to ensure consistency across studies and a twelve-month baseline prior to deployment. The results highlight the ability of the framework to lower the number of high severities confirmed cyber incidents per 1,000 vendor linked transactions by 46 percent compared to the baseline period, across a 487-node ecosystem comprising Tier 1, Tier 2 and Tier 3 entities. The mean time to detect and mean time to respond also increased significantly, and dynamic trust reclassification revealed multiple events at high-risk suppliers that weren't found with traditional monitoring, or cross tier propagation pathways. Further robustness experiments showed that explainability is not enough, and that the robustness of adversarial resistance, drift monitoring, and human feedback loops is required for safe deployment. The study provides an empirically supported and resource-oriented framework on how to enhance digital trust in supply chains. The results must simultaneously be understood as pilot evidence on a bounded multi case basis, and not as evidence of effectiveness in all sectors and geographies across the globe. The paper thus proposes that AI can have a major role in enhancing the cyber-resilience of supply chains, but only in conjunction with governance controls, the context analysis and human oversight.