Text-generative AI, such as ChatGPT, is transforming higher education. This exploratory research used a survey to gain deeper insights into the nuances of educator perceptions at South Africa’s universities. A conceptual model is proposed and tested to serve as a starting point for future research. Key findings indicate that educators are most concerned about the perceived risks of cheating, the spread of misinformation, and the impact on academic integrity among students. Conversely, the highest perceived opportunities are for increased productivity for both educators and students. Text-generative AI is seen as a valuable language tool, brainstorming aid, summary generator, and research facilitator for students, while its use in exams and tests is met with strong opposition. Recommendations predominantly revolve around the need for educational institutions to embrace text-generative AI while simultaneously establishing regulatory frameworks and best practices. Notably, only a small fraction of South African educators expressed outright opposition to the technology, reflecting educators' recognition of the evolving technological landscape and the imperative for education to adapt. It is believed that some of these findings can be validated in other contexts and will contribute to shaping policies and practice around the use of text-generative AI in higher education.