Higher education institutions face urgent decisions about integrating generative AI tools such as ChatGPT into Computer Science (CS) curricula, yet most policy responses remain reactive rather than evidence-based. Applying the Unified Theory of Acceptance and Use of Technology (UTAUT), this study tests whether Performance Expectancy (PE), Effort Expectancy (EE), Social Influence (SI), and Facilitating Conditions (FC) predict CS students' Behavioural Intention (BI) to use ChatGPT as a learning copilot. A cross-sectional survey of undergraduates in CS, Cyber Security, Data Science, and Software Engineering programs (N = 325) was analysed using PLS-SEM with bootstrapping. The model explained 84.2% of variance in BI. PE was the dominant predictor (β = 0.627, p < .001) and FC a meaningful secondary driver (β = 0.324, p < .001); EE and SI were non-significant in this post-adoption sample. Findings inform a four-pillar policy framework for sustainable institutional AI integration in CS programs.