The proliferation of generative artificial intelligence (AI) technologies has introduced challenges related to the use of automatically generated texts within foreign language learning settings. This study is aimed at developing clear-cut principles of incorporating large language models (LLM) into the English teachers’ workflow. Our approach employed examining research papers and conducting a small-scale study designed to obtain and evaluate texts produced by state-of-the-art LLMs in response to the General English course writing assignments. The experiment revealed peculiarities and limitations of the chosen LLMs in generating texts for study purposes. The educational potential of this technology as well as suggested conditions for its effective integration into instructional practices were also presented. We concluded that the transformation of writing assignments is necessary, with a focus on fostering critical thinking skills. Furthermore, fostering students’ information and communication technology (ICT) skills while engaging with an LLM chatbot is considered of paramount importance.