Â鶹Éç

Artificial intelligence is increasingly part of conversations about the future of journalism. In our office in Zambia, we began asking ourselves an important question: how prepared are our partner radio stations to engage with AI - and to use it responsibly?

To find out, we reached out to our media partners and conducted a small survey. We wanted to understand what journalists already knew about AI, whether they were using any AI-powered tools, and whether they had guidelines in place to support responsible use. The answer was clear: while there was growing awareness and curiosity, most partners had limited practical knowledge, and almost none had developed formal guidance.

Building readiness, not replacing people

Journalists were keen to learn. Our study showed some were worried about job losses and what it might mean for their roles. Our conversations with partners and editors specifically revolved around strengthening generative AI oversight in the newsroom, and addressing the challenges that come with it, including how it may negatively impact audience trust, and accuracy in content. Across the board, there was a shared desire to be prepared - and to stay relevant in a rapidly changing industry.

With support from Â鶹Éç Media Action’s internal EDIT team, we designed a training programme focused on understanding generative AI tools, how they process, analyse, and generate data and information; what drawbacks they have and how these may impact accuracy, objectivity, and ethical standards. We also wanted to build confidence and ethical awareness for the participants. Before delivering the sessions, we ran the training ourselves to ensure we fully understood both the opportunities and the risks.

Training and co-creating guidelines

Our first pilot training was delivered with the national broadcaster, involving more than 40 participants from Lusaka and regional studios. We later extended the training to 8 other partners, and this is ongoing.

The sessions were formed around three core elements:

• AI essentials and Responsible AI: introducing what AI is, where it is already appearing in journalism, and the potential risks involved.
• Experimentation with existing AI tools: where participants experimented with AI tools they already had access to. This helped surface issues such as factual errors, bias, and cultural mismatch - including AI-generated images that did not reflect local values - reinforcing the importance of audience awareness and editorial oversight.
• Co-designing responsible AI guidelines, where partners defined their own principles, including transparency, accuracy with human oversight, and respect for privacy. Practical questions were also addressed, such as approval processes, internal accountability and future training needs.

Rather than adopting a one-size-fits-all approach, each station developed guidelines based on its own priorities and realities - making them more meaningful and more likely to be used.

Thirteen partner radio stations have now formally established and adopted their own AI guidelines; each tailored to what they thought would be their priority. For stations like Icengelo, the top priority was that AI aligns with the teaching of the catholic church included in the guidelines, while for Zi FM, compliance with cybersecurity and cybercrime laws and with the IBA came at the top.

All stations, however, integrated human oversight and verification as a priority when using AI in the newsroom.

Keeping people at the centre

We continue to roll out the training in the coming months and by supporting our partners to understand generative AI tools we help ensure that innovation strengthens trust with their audiences.

The outcome is exactly what we had hoped for: partner-led guidelines that place human judgment firmly at the centre of any future use of AI. As the media landscape continues to evolve, our focus remains that technology serve journalism and audiences, but most importantly, the public interest.

--