Teaching English as a Foreign Language (EFL) in developing contexts like Vietnam is frequently constrained by two systemic deficits: the absence of an authentic language environment due to a scarcity of native speakers, and the prohibitive cost of copyrighted international teaching materials. These constraints often force teachers into a difficult compromise between providing high-quality input and adhering to legal or budgetary limits, leading to widespread reliance on unauthorized or outdated resources. This study investigates how secondary school teachers are strategically employing Generative AI (GenAI) not merely as an administrative tool, but also as a cost-effective solution to these specific environmental and economic challenges. Drawing on qualitative data from in-depth interviews with 15 EFL teachers across diverse socioeconomic settings, the research analyzes teacher agency through the lens of "resource bricolage."

The findings reveal that teachers are actively constructing a "low-cost authentic environment" by deploying AI-driven text-to-speech technologies as "synthetic native speakers," thereby providing students with consistent, high-quality listening exposure previously accessible only in affluent international schools. Furthermore, the study demonstrates a significant shift in material development, where teachers move from relying on copyright-infringing content to engaging in "personalized creation." By using GenAI to generate original, linguistically accurate reading materials and assessments tailored to local contexts, educators are effectively bypassing the need for expensive commercial textbooks. The study concludes that the primary value of GenAI in resource-scarce contexts lies in its ability to democratize access to authentic language input and provide a legally sustainable, ethical model for material design. These insights suggest that AI integration should be framed as a matter of educational equity and economic strategy rather than just technological innovation.