As artificial intelligence gains prominence in education, its application in designing and generating language tests has generated renewed interest in assessment. This systematic review examines the perception of EFL educators regarding AI-generated exam questions and further investigates classroom judgments of AI-assisted assessment tools in terms of quality, relevance, and efficacy. Hence, through peer-reviewed studies published from 2023 to 2025, the review synthesizes empirical findings concentrating on the views of teachers concerning AI-generated questions. Along the lines of the PRISMA guidelines, the review incorporated studies found in ERIC, Scopus, Web of Science, and other academic databases. Some of the key themes explored are: trust in AI, alignment with language learning goals, perceived bias or limitations, and the changing role of the teacher in designing assessments. Results indicated mixed feelings regarding AI in that teachers feel AI can assist in improving efficiency and creativity in assessments but remain worried regarding aspects of accuracy, context appropriateness, and pedagogical value of AI-generated content. The review points out some recommendations, such as teacher training, tools that are more transparent, and further investigations to foster the responsible integration of AI in EFL education.