As generative artificial intelligence (GenAI) increasingly permeates educational domains, English Language Teaching (ELT) is undergoing a profound transformation. Drawing on theoretical frameworks, reflective pedagogy, and recent classroom applications, this paper explores the dual nature of AI integration in ELT, presenting both its innovative potential and the challenges posed by AI-generated inaccuracies, or ‘hallucinations’. By examining contemporary AI-driven platforms, such as ChatGPT, Twee.ai, Natural Reader, Gemini, as well as image-generation tools like DeepAI, the article demonstrates their applicability to real-world classroom contexts, including writing support for Cambridge exam preparation, adaptive reading tasks, and multimodal resources for learners with special needs, with a primary focus on how AI can enhance engagement, personalisation, and accessibility in ELT. Simultaneously, it addresses the limitations of AI-generated content, such as inaccuracies, biases, and ethical concerns, emphasising the importance of pedagogical mediation and critical digital literacy, with a call for responsible AI integration and continuous professional development. Ultimately, it argues that AI should serve as a complement – not a substitute – for human judgment, creativity, and contextual awareness in language education, where cultivating critical thinking among learners and educators is essential to balance innovation and enthusiasm with ethical responsibility.
