Navigating risks and rewards – How South African journalists use AI in the newsroom

There is a moment in every newsroom where something shifts.

Not loudly. Not with fanfare. But quietly, almost imperceptibly, the way a new tool slips into daily practice. A shortcut here. A quicker turnaround there. A headline drafted in seconds rather than minutes.

Artificial intelligence has arrived in South African newsrooms in much the same way.

Not as a grand strategic rollout. Not as a carefully managed transformation. But as something journalists are already using, often instinctively, sometimes cautiously, and frequently without the support structures they need.

The result is a landscape defined less by disruption and more by improvisation.

The latest research shows AI is embedded in everyday newsroom tasks, from transcription and translation to drafting and research. But this adoption is uneven, informal, and largely unsupported by policy or training.

 

Used widely – but without guidance

Most journalists are teaching themselves.

There are a few formal policies. Little structured training. And almost no consistent frameworks for how AI should be used.

So decisions are being made in real time:

 

• • When to trust AI
  • When to double-check
  • When not to use it at all

This creates risk, not just for accuracy, but for credibility.

 

A technology journalists don’t fully trust

There is a clear tension.

AI saves time. But journalists don’t trust it enough to rely on it.

Outputs are checked manually. Facts are verified again. Translations are corrected.

In practice, much of that efficiency is lost.

At the same time, there are deeper concerns:

 

• • AI produces “bland” copy
  • It misses nuance and context
  • It struggles with African languages

In a country as linguistically and politically complex as South Africa, that matters.

 

The risk to journalism itself

The biggest concern is not job losses.

It’s the erosion of what makes journalism work:

 

• • critical thinking
  • strong writing
  • original reporting

There is a growing fear that over-reliance on AI could lead to shortcuts and ultimately, weaker journalism.

 

Built elsewhere, used here

The tools shaping South African journalism are not built for the realities it operates in. Most tools being used are not designed for African contexts.

They struggle with local languages, idioms, and political nuance.

This creates a structural mismatch between global technologies and local realities – with real consequences for accuracy and representation.

 

The opportunity, if done properly

Despite all of this, journalists are not resisting AI, but tell us they’re using it under the wire.

They see its value:

 

• • speeding up research
  • supporting investigations
  • freeing up time for deeper reporting

But they are clear about what’s needed next:

 

• • practical training
  • clear editorial guidelines
  • tools built for local realities

 

What comes next

The question is no longer whether AI should be used in newsrooms.

It already is.

The real question is whether it will be used responsibly.

Because trust remains the foundation of journalism.

And while AI can support the work, it cannot replace the judgment, context, and accountability that journalists bring.

And that responsibility remains, as it always has, with the journalist.

To explore the findings in full, and what they mean for the future of journalism in South Africa, read the report here: