The rapid integration of GenAI into higher education has raised an ongoing discussion about academic integrity and answer reliability. While it presents many other risks, such as biases and data privacy issues, it also offers opportunities for teachers and learners. AI literacy is therefore now deemed essential for educational stakeholders. This mixed-methods study, conducted in October 2023, involved a survey on ChatGPT uses that included an AI literacy questionnaire, based on an emerging framework of AI literacy. We analyzed semi-structured interviews with teachers and students and performed lexical analysis on open-ended responses. The findings reveal that students do not feel technically knowledgeable about GenAI and offer poor definitions of it in which they themselves have little confidence. The majority also use inappropriate strategies for conversations with ChatGPT. Despite having little technical understanding, however, they feel confident in their ability to use it. Students and educators primarily use GenAI for information retrieval and writing support, with students trusting its output more than teachers do. Although GenAI has become much better for information searches, these results point to a worryingly low student AI literacy level on the technical dimension. Both groups recognize AI’s risks, particularly plagiarism, cognitive skill loss and confabulations. Both students and teachers highlight the need for formal training on GenAI, as well as institutional guidelines that are not overly restrictive. This study emphasizes the urgency of comprehensive AI literacy programs and clear guidelines at the course or institutional level to promote responsible use.