As artificial intelligence (AI) content permeates social media, government communications, and corporate messaging, South Africans are facing a growing crisis: the inability to distinguish fact from fiction in a digital landscape cluttered with misinformation. The challenge is particularly acute for speakers of indigenous languages, as existing fact-checking technologies remain heavily biased toward English, leaving millions of citizens vulnerable to manipulated images, fabricated messaging, and digital deception. This linguistic gap has prompted a critical push to develop more inclusive technological safeguards.
Stepping into this void, Dr. Seani Rananga, a PhD student at North West University and a researcher at the University of Pretoria, has pioneered a groundbreaking study aimed at leveling the digital playing field. Her research focuses on developing AI systems capable of detecting misinformation not just in English, but specifically in isiZulu and Sepedi. By addressing the lack of localized tools, Dr. Rananga’s work seeks to ensure that South Africa’s rich linguistic diversity does not become a liability in the face of increasingly sophisticated online threats.
The urgency behind this research stems from the real-world consequences of unchecked misinformation, which can erode public trust, fuel panic, and manipulate democratic processes. Dr. Rananga’s study utilized the Covid-19 pandemic as a critical test case, demonstrating how rapidly false claims can spread and the profound damage they cause when local-language barriers prevent fact-checking. Her findings indicate that the primary hurdle in this effort is the lack of high-quality machine translation and the systemic underrepresentation of African languages in global AI datasets.
Despite the technical hurdles, the research has yielded a successful framework capable of identifying misleading content across all three languages. By proving that multilingual AI can be adapted to recognize falsehoods in isiZulu and Sepedi, Dr. Rananga has provided a scalable model that could be implemented nationally. This framework is designed to serve as a defense mechanism during sensitive periods, such as election cycles or public health emergencies, where the rapid spread of misinformation can have life-altering repercussions for the population.
The significance of this work has already garnered notable international recognition, including a prestigious Google PhD Fellowship and a Best Poster award at the 2025 Deep Learning IndabaX South Africa. As her academic profile grows, Dr. Rananga is preparing to present these findings at the 2026 Deep Learning Indaba in Lagos, Nigeria. This platform will allow her to advocate for a global shift toward more inclusive AI development, emphasizing that the future of digital safety must be rooted in local languages and cultural contexts to be truly effective.
Looking toward the future, Dr. Rananga intends to expand her model to encompass a broader spectrum of South African languages while tackling the next frontier of digital threats: deepfakes, hate speech, and coordinated disinformation campaigns. Her ultimate vision is to ensure that African languages are fully represented within the next generation of artificial intelligence. By building more inclusive and reliable systems, she hopes to contribute to a safer, more transparent digital space that protects all South Africans, regardless of the language they speak.
