The rapid diffusion of generative artificial intelligence (GenAI) is reshaping higher education by challenging traditional roles of teaching, trust, and academic integrity. This study aims to explore how university staff cognitively and ethically evaluate GenAI by analysing perceptions of its pedagogical replacement potential, practical feasibility, academic integrity risks, and perceived reliability across national contexts. The analysis is based on an anonymous cross-sectional survey of 637 respondents conducted between May and September 2025, using descriptive statistics, correlations, regression models, and exploratory factor analysis. The findings show that perceived replacement potential is low (M = 2.47), with over 51% of respondents rejecting the idea of AI replacing teachers. Academic integrity concerns are the strongest dimension (M = 3.51), while trust in AI accuracy remains low (M = 1.99), indicating widespread scepticism. Perceived cost and complexity do not significantly influence beliefs about replacement (R² = 0.007; p = 0.117), suggesting a weak relationship between feasibility and perceived impact. Finally, moderate positive correlation (ρ = 0.34) and low reliability (α = 0.50; α = 0.45) provide evidence that perceptions of GenAI are fragmented and multidimensional rather than internally consistent.