The rapid expansion of AI in South Africa’s media sector offers news outlets significant opportunities to improve operations, but the country’s linguistic diversity and uneven digital access require an ethical and context-sensitive approach to adoption. Feedback from the South African media ecosystem highlighted that many journalists were navigating AI on their own due to a lack of formal policies and training opportunities. This aligns with findings from the Foundation’s report on AI use in the Global South, which revealed that most of the 200+ journalists surveyed reported their newsrooms had no established AI policy despite widespread usage. The absence of guidance increases the risk of AI-related errors, such as unintentional bias or factual inaccuracies, which could undermine reader trust and damage reputations. Ethical concerns consistently emerged, emphasizing the need to support newsrooms in responsible AI integration.
In response, with support from Microsoft, the Foundation implemented a four-month program for four South African newsrooms—The Mail & Guardian, AmaBhungane Centre for Investigative Journalism, Briefly News, and Pondoland Times—to leverage AI while minimizing risks through robust editorial policies and guardrails. Participating newsrooms received bespoke mentoring from AI and journalism experts to develop strategies and policies, laying the foundation for responsible AI adoption and innovation.
The newsrooms identified practical AI applications tailored to their organizational needs and developed tools to improve editorial efficiency, audience reach, and workflow processes. Briefly News created “Editorial Eye,” an AI tool for proofreading, style, and grammar checking, which increased article output from 80–90 to 150–200 per day and boosted page views by 22% over six months. AmaBhungane developed a tool to transform in-depth investigations into engaging multimedia content, streamlining scriptwriting and making complex stories more accessible. Pondoland Times automated social media posting and is developing an AI-generated avatar for news delivery, resulting in website traffic growth from 2.4 million to 10 million monthly impressions and a revenue increase of 20,000 ZAR in one month. Mail & Guardian implemented a sub-editing tool to automate grammar checks, headline generation, and hashtag suggestions, halving the time spent on these tasks.
Throughout the program, newsrooms gained confidence in using AI tools responsibly, addressing initial concerns about job displacement and ethical issues. The project reinforced the importance of human oversight as a central component of ethical AI practices while demonstrating the potential for AI to enhance journalism in South Africa.
