Artificial intelligence (AI) is reshaping academic writing by offering both challenges and opportunities. Numerous studies have discussed this topic and are worth compiling and synthesising. Based on the topic, several existing studies published from January 2023 to March 2025 were reviewed systematically using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) framework. After searches in Scopus and Google Scholar yielded 1,024 records, 34 titles were selected based on defined inclusion and exclusion criteria. The findings highlight the two sides of AI in academic writing: 1) presenting ethical issues, training needs, Critical thinking degradation, AI misinformation, AI addiction, weak AI detectors, AI’s rhetorical patterns, users’ negative perceptions, and other challenges, 2) facilitating several opportunities, such as writing quality improvement (accuracy, organization, and style), writing productivity and efficiency, idea development, AI awareness, research support, new insights, writing engagement, and usability. This study contributed to developing guidelines to address ethical, effective, and prudent use of AI in academic writing. It is a foundation for upcoming researchers who delve into AI in academic writing and other related skills in the EFL teaching context. Moreover, investigations across disciplines and educational levels are worth exploring.