Abstract
This study examines the transition from traditional Machine Translation (MT) technologies to Generative Artificial Intelligence (GenAI) in website localization workflows, with particular attention to Arabic localization challenges. It argues that localization has evolved from a string-based, efficiency-driven process toward a contextaware and integrative practice enabled by Large Language Models (LLMs). By analyzing technological developments, localization procedures, and linguistic constraints, the paper demonstrates how GenAI improves discourse coherence, cultural adaptation, and technical accuracy while reducing risks such as tag corruption and layout errors. The study concludes that hybrid workflows combining MT reliability with GenAI contextual intelligence represent the emerging paradigm in contemporary digital localization.