Sustainable E-Commerce in MENA: An SOR Analysis of Gen Z Purchase Intentions through Open-Source LLM (DeepSeek) Interactions, Trust, Familiarity, and Privacy Concerns

Purpose: This study models how open-source large language model (LLM) recommender interaction quality (personalisation and responsiveness) affects Gen Z online purchase decisions through trust, AI familiarity, and privacy concerns using the Stimulus-Organism-Response (SOR) framework. Design/Methodology/Approach: We have conducted a Partial Least Squares Structural Equation Modeling (PLS-SEM) analysis of 570 Gen Z online shoppers from Egypt, Saudi Arabia, Jordan, and Qatar. Findings: Interaction quality improved trust (β=0.2585) and AI familiarity (β=0.2943) while increasing privacy concerns (β=0.5499). Purchase intention was positively influenced by trust (β=0.2848) and privacy concerns (β=0.3583). AI familiarity showed no direct impact (β=0.0477). Model explained 47% variance in purchase intention. Originality/Value: Few studies examined trust-privacy dynamics for open-source LLM recommenders across four economies. Research Implications: SOR theory application reveals trust mediates positively while high interaction quality increases privacy concerns among Gen Z. Practical Implications: Retailers should optimise interaction quality while implementing transparency and user control mechanisms to address heightened privacy concerns. Keywords: Open-Source LLMs; E-Commerce Interaction Quality; Trust-Privacy; SOR Model; Gen Z; Middle East Shopping. Citation: Shaheen, A., Omeish, F., AlGhamdi, D. S., Khataan, A. and Awad, A. (2026): Sustainable E-Commerce in MENA: An SOR Analysis of Gen Z Purchase Intentions through Open-Source LLM (DeepSeek) Interactions, Trust, Familiarity, and Privacy Concerns. World Journal of Entrepreneurship, Management and Sustainable Development (WJEMSD), Vol. 22, Nos 1-2, pp. 133-150.