AI and journalism in southern Africa: editors are using it but balanced with human expertise and editorial judgement

J Studios/Getty Images Artificial intelligence (AI) is becoming part of everyday newsroom work across Africa. It has entered quietly through routine tasks such as transcription, headline writing, translation and content preparation.

In southern Africa, where AI adoption is steadily growing, its application in journalism is raising critical questions from policymakers and governments. While technology offers gains in speed and efficiency, its use remains contested due to ethical concerns and fears about job losses.

As communication and media scholars researching data and digital communication, in our study we examined its influence on production processes, ethical guardrails and job security. Interviews with senior editors revealed that, while AI improves efficiency and, in some cases, quality, it still requires human expertise and editorial judgement.

Senior editors described efficiency: faster turnaround, transcription, summarisation, headline generation and story drafting. Large volumes of information can be processed within tight newsroom deadlines. Most editors do not see AI as an immediate threat to jobs. Ethical concerns remain, prompting some newsrooms to adopt internal guidelines.

AI is already helping journalists sharpen headlines, summarise reports, generate illustrations, transcribe interviews and clean up copy under pressure. In some Zimbabwean newsrooms, AI-powered presenters are already reading weather bulletins and assist with news delivery.

Yet caution prevails. Editors are experimenting with AI because newsroom pressures demand efficiency, but they remain determined not to surrender editorial judgement to machines.

This caution reflects broader structural pressures. Print circulation has declined, advertising revenue remains fragile and newsroom staffing has shrunk. In South Africa, newspaper circulation declined by 17.3% in 2024, with several major titles reducing operations or shifting to digital-first models. Journalists are expected to produce more content, across multiple platforms, at greater speed.

AI, however, introduces its own risks, including factual inaccuracies, hidden bias embedded in training data and weak contextual understanding. For example, AI systems may reproduce racial, gender, political, or cultural biases while struggling to interpret satire, local idioms and politically nuanced African contexts. As a result, editors emphasise that AI must remain under firm human control.

The first newsroom functions being reshaped by AI are repetitive tasks. Editors described using AI for headline optimisation, summarising, transcription and minor editing. These are labour-intensive processes but do not determine editorial direction.

In Zimbabwe, experimentation is more advanced in selected organisations. AI avatars, AI-powered digital news presenters capable of delivering human-like news bulletins through synthetic voices, facial expressions and automated script reading, are presenting weather updates and selected content.

South African newsrooms remain more restrained. AI is mainly used in editing, reporting and headline optimisation. Full article generation remains limited because editors insist on rigorous human verification.

For now, AI functions as an assistant rather than a substitute.

The central issue is credibility. Generative AI produces fluent language, but fluency does not guarantee accuracy. It predicts plausible content rather than verifying truth. So it can generate convincing but incorrect information. One example is in the saga involving the development of the South African AI strategy by government. It was found to contain several fictitious academic references likely generated by AI hallucinations.

Editors in both countries highlighted this risk. Zimbabwean editors noted that AI often draws from online sources without distinguishing between verified reporting and misinformation. South African editors raised concerns about plagiarism, weak attribution and unverifiable sourcing.

This creates a paradox: AI speeds up writing but also creates more work, as journalists must verify machine-generated content before publication.

Accuracy is only part of the problem. Many AI systems struggle with African linguistic and cultural contexts. Editors reported issues with pronunciation of indigenous names and poor handling of local nuance.

Most AI systems are developed in the global north and trained on western datasets, leaving African languages underrepresented. This calls for greater investment in African-centred AI research, local language datasets, and inclusive digital innovation policies.

African newsrooms are adopting tools that do not fully recognise their communication environments. Editors argued that locally grounded AI systems will be necessary to reflect African realities and avoid deepening technological dependence.

While fears about job losses are widespread, editors offered a more measured view. Most do not expect journalists to disappear but anticipate pressure on technical roles such as sub-editing and layout.

Some acknowledged that media owners may eventually use AI to justify leaner staffing. But high costs remain a barrier. In Zimbabwe, in particular, expensive subscriptions and infrastructure challenges limit adoption.