Ethical concerns of generative AI in academic libraries: A cross-regional quantitative study of librarians’ risk perceptions, trust, and governance practices

Abstract

This study investigates academic librarians’ perceptions of ethical risk associated with the adoption of generative AI for various libraries in terms of data privacy breaches, threats to intellectual freedom, and algorithmic biases. The study also examines how AI ethics competence and institutional support can lead to responsible use intentions. The study employs a quantitative research approach with a cross-sectional survey gathering data from 305 academic library professionals working in higher education across South Asia (44.3%), Africa (41.6%), and the Middle East (14.1%). The survey instrument consisted of a structured questionnaire with 30 Likert scale variables across six constructs. Data analysis involved descriptive statistics, T-Test, ANOVA, Pearson correlations, and Structural Equation Modeling (SEM). Librarians reported high ethical risk perception (M = 3.41–3.77), strong responsible use intentions (3.77–4.02), and a moderate level of AI ethics competence (M = 3.47–3.77). Formal AI policies showed enhanced guidance perceptions with no robust gender differences across all constructs. Private institutions scored better than the public institutions in terms of equipping the librarians with training and resources. Geographical regional differences showed significant differences across several variables; South Asia scored the highest in awareness level, while Africa scored high in policy gaps. Structural equation modeling results showed that both AI ethics competence and perceived ethical risk positively impacted ethical AI governance, and ethical AI governance significantly increased responsible practice intentions. However, there was no significant direct effect of perceived ethical risk on responsible practice when both governance and competence were controlled. The study adds to the growing literature on librarianship ethics in the context of AI-integrated services in academic libraries, providing quantitative, non-Western evidence on risk, competence, and governance dynamics. The study offers a proposed model/framework for addressing ethical concerns and global gaps in an AI-mediated library environment in higher education. The findings of this study provide theory-driven evidence for ethical AI governance in resource-limited settings, urging targeted training and policies to align innovation with librarianship values.