Mombasa lab trains African reporters to use AI without surrendering verification

Investigative journalists and media experts spent five days in Mombasa examining how artificial intelligence can assist public-interest investigations without displacing verification or editorial responsibility. The Africa Investigative Journalism Lab, held Nov. 4-8, paired AI training with practical work on evidence, source protection, digital security and storytelling.

KAS Media Africa organised the programme with Africa Uncensored, an independent investigative journalism organisation based in Nairobi. It was intended for mid-career and senior journalists from across Sub-Saharan Africa and included instruction, exercises and individual editorial support, according to the organisers’ account of the event.

Participants worked from initial story ideas towards investigative questions and editorial pitches, with public interest, accountability, evidence and impact presented as the foundations of effective investigations. They developed their own projects throughout the week, ultimately refining angles, presenting pitches and setting concrete action plans.

Responsible AI use was a central strand of the programme. Sessions covered large language models, prompting, technological bias, hallucinations and ethical considerations. In workshops, journalists used AI to organise documents, transcribe interviews and develop tools for their investigations.

The training positioned those uses within a wider reporting workflow rather than as substitutes for reporting judgement. The organisers said the programme explored AI as support for research, analysis and newsroom work, while maintaining that credible journalism depends on verification, ethical judgement and commitment to the public interest.

Other sessions addressed open-source intelligence, data journalism and digital verification. Participants used public databases, online archives and social-media information, and learned methods for checking images, videos, claims and locations. They also covered online influence networks, source protection, covert filming, and the physical and psychological risks faced by investigative journalists.

The programme included ways to communicate investigations through articles, video, podcasts, data visualisations and social media. Its stated aim was to equip participants to develop, investigate, verify and communicate high-impact public-interest stories, in response to the growing importance of investigative storytelling, data journalism, online research and journalist safety.

The focus on capacity and safeguards reflects a broader uneven picture of AI use in African public-interest media. An International Media Support study found use cases across African news organisations, with Kenya and South Africa recording the most, but said adoption remained relatively low and varied by region and type of outlet. Large, well-resourced organisations had invested in premium systems and custom tools, while many smaller outlets had either not adopted AI or relied mainly on open-source tools.

That study identified knowledge and resource gaps, fears of algorithmic harm, weak policy and legal infrastructure, data problems and limited collaboration as obstacles. It also warned that biases can be particularly pronounced where African training data are lacking, and that AI could widen the digital divide. It recommended capacity building, support for local AI innovation and stronger policy frameworks.