Research completed through North-West University has developed a multilingual artificial-intelligence framework aimed at identifying misinformation in English, isiZulu and Sepedi.
Dr Seani Rananga, a former NWU doctoral student and now a University of Pretoria lecturer, used misinformation from the Covid-19 pandemic as a pilot case. The work found that multilingual systems can identify misleading content across the three languages, while also showing that machine-translation quality has a major effect on performance in languages with fewer digital resources.
The framework is intended as a foundation for broader tools that could help fact-checking organisations, public institutions, online platforms and communities identify harmful information earlier. Rananga plans to extend the research to additional South African languages and to challenges including deepfakes and hate speech.
The research received a Google PhD Fellowship and a best-poster award at the 2025 Deep Learning IndabaX South Africa. Rananga is due to present the work at the 2026 Deep Learning Indaba in Lagos in August.