AI in Advertising and Newsrooms South Africa: The Newsroom Reality

Per press statements from recent years, artificial intelligence is no longer a looming future; it is already pervading South African newsrooms and advertising agencies. The Press Council of South Africa’s (PCSA) stark reminder, reiterated in its November 2023 AI guidance, conjures robots poised to take even more control from journalists. In a press statement, the PCSA stated that member organisations would remain responsible for every bit of published content, regardless of any AI intervention. Human eyes and hands must still verify any AI-generated content.

Yet for all the talk of AI-powered newsroom and agency workflows, few of the South African media outlets actually have explicit AI policies in place; only one systematic review tracking the transparency and AI usage of major outlets in 2025 found none of the sampled newsrooms to have a single publicly visible policy on AI. Corporate and marketing ’bots may be integrating with journalist workflows, but editorial oversight remains: most newsrooms, like the major broadcaster SABC, exclude sensitive stories from AI/script generation.

Instead, South African journalists report that AI tools are primarily used for assistive, task-specific roles in everyday production. A 2023 study ranked ”basic AI tools” and “transcription robots” among the most popular tools with one CNBC Africa journalist saying, “We use robots to transcribe our interviews. We actually can use a transcribed interview in real time.” A later KAS briefing reported AI integration for journalist research, summarisation, transcription, translation and drafting content such as headlines and social media posts. These tools are typically accessed through cloud apps, with little formal training among newsroom staff.

Advertising agencies, meanwhile, have already begun using AI to optimise campaign execution and track broadcast and digital coverage. News monitoring service Ornico claims to use “state of the art AI technology” – without detailing which models or hardware – to monitor “the most comprehensive list of South African and African media across all channels” and . Newsclip and similar services are linked to corporate marketing strategies to surface coverage for brands.

But the raw information in AI NLP models is not raw, but pre-processed by next-generation humans. While Ornico touts AI outputs, they also highlight human editors in the loop, screening and contextualising the AI-generated data for clients. Human editors filter and contextualise the AI-captured data for clients.

These human systems extend similarly into newsrooms, where exclusive automation is far from the reality. In its 16 February 2026 AI Policy guidelines, for example, The Witness’ approach to AI is journalist-centric, focusing how AI might assist journalists, rather than how it might replace them: ”AI doesn’t replace our journalists,” the statement reads. The Witness’ AI Policy says the paper may use AI tools to assist in headline and social media posts, summarising long texts, assist in editing for grammar and style, and support planning and story development, but every article is signed off by a human journalist before publishing. The publication offers to offer transparency to readers about AI’s role.

This will be a key theme of the years to come: how South African AI guidance is localised to fit with the PCSA’s human-centric ethos for journalism. The Moxii Africa guidelines, for example, recomends editorial independence, transparency disclosures, and training, to be matched as closely as possible with editorially-agreed guidelines from each newsroom or agency. The PCSA has been monitoring AI’s impact on the Press Code of Ethics and Conduct since August 2026, as guardian of the Press Code of Ethics and Conduct.

This does not mean that human editors and journalists are immune from AI’s impact. Newsrooms will need to protect themselves and their practices, and step up their training – and decision-making – competency. The year’s research from SANEF and Civic Signal shows that even though AI tools have quietly gained a foothold in South African newsrooms, the extent of AI’s impact (let alone any explicit AI policies) is not consistent across organisations, and staff training is often lacking. A tailored specialisation would require bypassing traditional journalism education pathways, and integrating non-traditional AI resources. This will be essential not simply for editorial credibility, but to enforce the Press Council’s key advice: that due to the potential of AI to produce content with an “unverified” character, any AI-generated content should be verified by a human and, where possible, provided with a byline, with disclosure of AI’s role.

This approach now looks backward to the history of AI and journalism, long before the latest wave of generative AI training broke, when a news outlet declined to publish articles produced by an algorithm, calling it a breach of its journalistic standards.

As South African newsrooms and agencies navigate the next wave, we may learn more about effective practices for integrating AI tools.