System level governance of generative AI risks in cross border e-commerce supply chains
Generative artificial intelligence (GenAI) is increasingly embedded in cross-border e-commerce supply chains, where digital platforms support automated content generation, product recommendation, customer interaction, supplier evaluation, compliance automation, and sustainability reporting. While these applications create opportunities for efficiency and innovation, they also introduce ethical, governance, and sustainability risks that extend beyond individual algorithms or isolated organizational processes. This study conceptualizes system-level risk as risk that emerges and propagates across interconnected data pipelines, model architectures, platform functions, organizational actors, supply-chain relationships, and cross-jurisdictional regulatory environments. Drawing on a PRISMA 2020-based systematic literature review of 66 peer-reviewed and policy-relevant publications, the article develops a structured taxonomy of GenAI-related risks and an integrated conceptual governance framework for cross-border e-commerce supply chains. The review identifies five core risk domains: bias and fairness, privacy and data governance, misinformation and manipulation, accountability and liability, and environmental sustainability. The findings suggest that these risks often interact and reinforce one another across data, model, application, organizational, and regulatory layers, rather than operating as separate technical problems. The study further maps these risk domains to governance mechanisms discussed in the literature, including explainable AI, privacy-preserving and federated learning approaches, auditability mechanisms, human-in-the-loop oversight, regulatory alignment, and green AI strategies. However, the review further indicates that these mechanisms are supported by different levels of evidence and should be understood not as universally validated solutions, but as governance enablers. The proposed framework does not seek to replace existing AI governance models, but rather complements and extends them by focusing on GenAI risk propagation in cross-border platform ecosystems, where technical design, organizational responsibility, regulatory fragmentation, and sustainability obligations intersect. The article contributes to the literature by offering a system-level synthesis of GenAI governance risks and by outlining a diagnostic framework that may support researchers, platform operators, and policymakers in assessing and governing GenAI-enabled cross-border e-commerce infrastructures. The limitations of the study include its conceptual nature, its reliance on English-language indexed literature, and the absence of primary empirical validation.