Why the EU’s new AI law matters for South African newsrooms

SA and legacy media can use this world-first legislation to improve their role as guardians of information integrity in an African context.

Published on 20 August 2026 in ISS Today

By

Karen Allen
ISS Consultant and former BBC foreign correspondent

This month, transparency obligations under the world’s first artificial intelligence (AI) law came into force in Europe. The groundbreaking European Artificial Intelligence Act (EU AI Act) has important lessons for South Africa and for legacy media, given their role as guardians of information integrity.

The law carefully regulates the use of AI for content creation, distribution and audience engagement, with various stipulations regarding labelling and transparency. Media companies can be held liable for harmful AI-generated content and must establish ways to demonstrate accountability.

While legacy (traditional) media (radio, television and print) must increasingly compete with social media for audience share, it is bound by professional standards that social media is not.

The Independent Communications Authority of South Africa controls the airwaves and electronic communications, and the Broadcasting Complaints Commission and Press Council of South Africa (PCSA) adjudicate on press complaints. The PCSA acts as guardian of the Press Code of Ethics and Conduct for print and online media and closely monitors AI’s impact on news media operations.

A new CINIA study in South Africa found that few newsrooms have developed clear AI policies

Hendrik Sittig, Director of the Konrad-Adenauer-Stiftung’s Media Programme for Sub-Saharan Africa, says: ‘The EU AI Act does not force newsrooms to stop using AI. Instead, it requires greater transparency, clear labelling and accountable editorial oversight.’ This is relevant for South African media as ‘news organisations that use AI will increasingly have to explain how it is being used and who is ultimately responsible for the content they publish.’

The media increasingly use AI for research, translation, summarising material and content creation. It also helps distribute material at speed and scale. While the EU AI Act imposes limits and assesses the risks of various AI uses in European contexts, African newsrooms must consider regulation and how to make a technology largely shaped in the global north work in an African setting.

A new Centre for Information Integrity in Africa (CINIA) study assessing how the media use AI indicates that South African newsrooms must ramp up efforts to establish the rules of the road. CINIA found that AI use was largely ‘predicated on whether there is an individual in the newsroom with a keen interest in the technology’ and whether they are informally teaching their peers.

Some participants told researchers they felt certain journalists were using AI ‘secretly.’ Others were unclear about what they were and weren’t allowed to do. More experienced staff said they worried about ‘plagiarism, lazy journalism and the inability of some AI tools to deal with African languages.’

Many journalists told CINIA that ‘[AI] mistranslates political or cultural terms’ because training data for AI models is largely drawn from outside the country, and lacks context. To deal with this problem, African developers are seeking to create large language models, which underpin AI, that are better equipped to handle African languages.

Newsrooms must be able to explain why AI was used and whether there was human oversight

The study also found a ‘fragmented’ and ‘nascent policy landscape,’ with few newsrooms having developed clear AI policies. CINIA Director Professor Herman Wasserman said although there was ‘great enthusiasm among South African journalists about the potential and possibilities of AI for journalism,’ he worried that guidelines on how to use AI ethically and responsibly were lacking, and more training was needed.

The PCSA is already considering ways to deal with AI-generated content that isn’t clearly labelled. A 2023 guidance note urged editors and journalists to be ‘thoughtful when they deploy new AI tools.’ It emphasises that AI-generated material must be checked by ‘human eyes and hands,’ fact-checked, and maximum transparency provided to audiences about AI use. The guidance note is not binding, and many agree that a new AI press code is needed.

PCSA Executive Director Phathiswa Magopeni says industry discussions about AI use essentially boil down to media ethics: ‘It’s basically about verification, accuracy and editorial due diligence.’ For her, the two primary focus areas for South African media should be issues of disclosure and explainability.

Images used from external providers are usually labelled with a credit, but AI use may be less obvious if it’s entrenched in news workflows, Magopeni says. ‘We need to determine levels of disclosure because AI may be used at various stages of the production process, for example writing, research and the generation of images.’ A key point is how to adapt and use the technology while maintaining audience trust.

In future, media organisations and newsrooms must be able to explain why AI was used, she argues, and whether there was human oversight.

Newsrooms must be able to explain why AI was used and whether there was human oversight

To date, most instances of ‘inauthentic’ AI-generated content in South Africa have appeared on social media rather than broadcast news networks – such as deepfakes endorsing parties and candidates during elections. This is thanks partly to an energetic but underfunded fact-checking community that calls out the activities of those seeking to bend the truth for political ends, entertainment value, or to boost engagement.

Editors already facing staff cuts and a drive towards automation to compete with online platforms are aware of the risk of eroding trust. That risk increases as AI gets integrated into newsroom operations in the absence of guardrails and rigorous training.

The case of the Communications and Digital Technologies Department’s use of fake citations in its draft National AI Policy shows what happens when AI is followed without caution. More recently, similar AI hallucinations embarrassed a Johannesburg acting judge whose ruling appeared to rely on ‘non-existent’ case law.

Professional news journalism is vital for democracy. And if credibility is the currency of that journalism, South African newsrooms urgently need practical support as AI becomes part of their daily operations.

More generally, CINIA’s Rejoice Malisa-van der Walt argues that AI regulations for the country as a whole need to be dynamic and inclusive, as the technology advances. In a news context, that means consulting audiences about their tolerance threshold for AI use in news, and identifying the red lines.


For permission to re-publish ISS Today articles, please email us. In South Africa, News24 has exclusive rights to republish ISS Today articles. In Nigeria, Premium Times has exclusive rights to republish ISS Today articles.