<p>The rapid integration of generative artificial intelligence (AI) in academic writing has raised concerns regarding authorship authenticity, yet limited research addresses the efficacy of AI text detection tools in identifying nuanced forms of AI involvement, particularly AI-edited content. Existing studies primarily focus on distinguishing fully human-produced and AI-generated texts, neglecting hybrid forms of authorship. This study addresses these gaps by evaluating the accuracy and reliability of ZeroGPT and SciSpace in detecting student-produced, AI-edited, and AI-generated essays. Using a cross-sectional quantitative approach, the study analyzed 450 essays across three categories using these detection tools. Results indicate that ZeroGPT significantly outperforms SciSpace in identifying student-produced and AI-generated texts, with consistently high accuracy and reliability. Both tools, however, demonstrate limited effectiveness in detecting AI-edited essays. The findings suggest that while ZeroGPT provides robust detection capabilities, further refinement is necessary for the comprehensive identification of hybrid authorship. Enhanced training datasets and algorithmic adjustments are recommended to improve the tools’ performance in increasingly complex academic writing scenarios. Implications for educational practices, tool enhancement, and future studies are discussed.</p>