ChatGPT has been increasingly used in computer science, offering efficient support across software development tasks. While it helps students navigate programming challenges, its use also raises concerns about academic integrity and overreliance. Despite growing interest in this topic, prior research has largely relied on surveys, emphasizing trends over in-depth analysis of students' strategies and ethical awareness. This study complements existing work through a qualitative investigation of how computer science students in one UK institution strategically and ethically engage with ChatGPT in software development projects. Drawing on semi-structured interviews, it explores two key questions: How do computer science students ethically and strategically report using ChatGPT in software development projects? How do students understand and perceive the ethical issues associated with using ChatGPT in academic and professional contexts? Findings reveal a shift in students' learning models, moving from traditional "independent thinking-manual coding-iterative debugging" to "AI-assisted ideation-interactive programming-collaborative optimization." Importantly, many use ChatGPT conversationally to deepen understanding, while consciously reserving creative and high-level decision-making tasks for themselves. Students tend to cap ChatGPT's contribution to roughly 30%, and evaluate its output to mitigate overreliance. However, only a minority thoroughly analyze AI-generated code, raising concerns about reduced critical engagement. Meanwhile, students reject uncredited use, highlight risks such as privacy breaches and skill degradation, and call for clear usage guidelines set by their teachers. This research offers novel insights into the evolving learner-AI dynamic and highlights the need for explicit guidance to support responsible and pedagogically sound use of such tools.

Full paper oral presentation at the European Society for Engineering Education (SEFI) 2025 Annual Conference (September 2025)