The emergence of generative artificial intelligence (AI), particularly large language models (LLMs), has rapidly reshaped the educational practices of university students. While these tools offer unprecedented support in academic activities among computer science (CS) students, they also raise critical concerns regarding academic integrity, over-reliance, and the decline of problem-solving skills. Despite growing discussion on the pedagogical and ethical implications of generative AI, there remains a lack of empirical evidence on how CS students engage with these technologies. This study addresses that gap by presenting the results of a survey conducted among 266 undergraduate CS students across multiple universities. The research investigates adoption patterns, practical applications, and self-assessed proficiency related to LLM use. The findings offer insights into how generative AI is integrated into academic workflows, providing a foundation for designing policies that support responsible and effective AI-assisted learning in CS education.