<![CDATA[The use of generative artificial intelligence (GenAI) among university students is growing rapidly, yet evidence on Indonesian students' ethical understanding and verification behaviour remains limited. This study maps GenAI use patterns, understanding of five key AI literacy statements, and the relationship between use frequency and that understanding among students who attended an interactive AI literacy training. A post-session cross-sectional survey was used, with Mentimeter data from 70 unique users at the "AI Amazing" training on 29 June 2025; item-level respondents ranged from 48 to 67. Analyses included proportions with Wilson 95% confidence intervals, exact binomial tests, the Friedman test with Holm-corrected Wilcoxon post hoc tests, Spearman correlations, Fisher's exact tests, and thematic coding of open responses. In total, 78.8% of respondents used AI at least several times a week and 86.2% used ChatGPT. Delegating assignments ranked on a par with learning new things (mean ranks 1.94 and 2.03). Most respondents answered in line with the key that AI will not fully replace teachers (85.4%) and recognised learning benefits (92.6%), but only 45.3% recognised ethical risks, 30.0% rejected the idea that using GenAI to write is automatically plagiarism, and 40.7% chose cross-checking to reduce hallucinations. Higher use frequency was associated with lower recognition of risks (rho = -0.36), although this exploratory association does not clearly survive multiplicity correction. AI literacy training should emphasise verification and ethical use, not only tool familiarity]]>
