Artificial Intelligence (AI) has penetrated English language learning and teaching, and this penetration parallels a surge of research on its potential and challenges for learners and teachers altogether. This paper provides a glimpse of the status quo of generative (Gen) AI in contexts where English is not used natively. Adopting a holistic approach to researching these GenAI tools, the study takes three large language models (LLMs) as its focus: ChatGPT, Gemini (formerly Bard) and Copilot (formerly Bing Chat). The discussion focuses on how such chatty LLMs bolster learners' command of English and support teaching it innovatively. It is organized around aspects of effectiveness and challenges in four parts. The first is about LLMs and how they shape new pathways for more interactive English learning. The second sketches how these language models are invested for enriching formal learning, self-directed learning, interactive content, feedback, and learning materials. The third part beds on teachers, including planning teaching, assessment, and professional development. The fourth part culminates in concerns associated with GenAI in English programs. The study charts territories for further research to paint both sides of a complete picture of GenAI in language education.