The rise of generative AI as a writing tool constitutes an unprecedented threat to academic integrity in L2 composition classes, leading to a new form of academic integrity violation, AI-assisted plagiarism. In the face of this groundbreaking challenge, L2 educators can only rely on automated commercial AI detectors to identify AI-generated texts; however, exploratory studies suggest they are not always reliable, especially when examining obfuscated AI-generated texts. What complicates the issue further is that L2 educators usually lack the technical expertise to evaluate the underlying mechanisms behind these detectors, which leaves them with minimal guidance in (a) identifying reliable detectors, (b) interpreting AI detectors' reports, and (c) utilizing available resources to inform academic integrity decisions. To address this challenge and explore the uncharted territory of AI-assisted plagiarism in L2 writing contexts, the present study offers a review of the literature on AI-content classification approaches grounded in empirical machine learning research. The study provides a critical review of the current literature on the strengths and limitations of each approach, guiding L2 educators in selecting and using commercial AI content detectors. Practical and research implications follow the literature review and discussion.
Karim Hesham Shaker Ibrahim (2026) studied this question.