Purpose – This study aims to examine the comprehensive role of generative artificial intelligence (AI) tools and smart learning devices in enhancing knowledge acquisition among accounting students in Ghanaian private universities, as well as the impact of perceived usefulness, perceived ease of use, and institutional support on students' AI adoption behaviour in accounting education. Design/methodology/approach – Anchored by the Unified Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh et al., 2003), this study employs a qualitative research design using in-depth interviews with accounting students purposively selected from four private universities in Ghana. Thematic analysis was used to analyse the data gathered, using UTAUT's four constructs, performance expectancy, effort expectancy, social influence, and facilitating conditions, as sensitising concepts. Findings – The study reveals that generative AI tools, particularly ChatGPT and smart learning devices, positively enhance accounting students' knowledge acquisition in Ghanaian private universities. Perceived usefulness and ease of use emerged as the dominant drivers of AI adoption, while institutional support and digital literacy moderated how effectively students integrated these tools into their learning. Ethical concerns, over-reliance, and uneven access to devices remain significant challenges. Research limitations/implications – This study focuses exclusively on private universities in Ghana and therefore limits the generalisability of findings. Future research should extend to public universities and across African contexts. Practical implications – The findings encourage Ghanaian private universities to invest in AI-driven pedagogical infrastructure, develop institutional policies governing responsible AI usage, and build the digital competencies of accounting faculty and students to maximise the educational benefits of generative AI. Originality/value – This paper makes three distinct contributions. Primarily, it demonstrates that UTAUT's four constructs are not static cognitive judgments but are dynamically negotiated through institutional, infrastructural, and ethical contexts in developing-country higher education settings, thereby extending UTAUT’s explanatory scope. Furthermore, it introduces the concept of an institutional legitimacy vacuum to account for how the absence of formal AI governance policies drives covert and unguided student AI use in Ghanaian private universities, a phenomenon that existing UTAUT-based models do not theorise. Finally, it provides the first qualitative, multi-university evidence base on generative AI adoption among accounting students in Ghanaian private higher education, an under-researched context with distinct structural challenges including large class sizes, resource constraints, and infrastructure deficits.