Is AI looting SA culture? Kabza De Small tops list of copied musicians

<p>Sharing the findings on X, Mokoena wrote: “The biggest export South Africa never authorised may be its culture. </p>

<p>"Our audit of major AI training datasets identified dozens of South African artists appearing across publicly documented training corpora used to develop the next generation of AI music systems.</p>

<p>“As artists, rights holders, and policymakers begin to examine the scale of this practice, it may prove to be one of the most significant data laundering controversies the global music industry has ever faced.”</p>

<p>The findings came after music industry executive Lindiwe Dhlamini pointed to a searchable database published by "The Atlantic", which allows users to search music that was reportedly used by technology companies to train generative AI models.</p>

<p>According to Cybrsec Clinique’s audit, more than 200 South African artists across genres, including Amapiano, Kwaito, Maskandi and Gqom, were identified in the datasets.</p>

<p>“More than 200 South African artists across Amapiano, Kwaito, Maskandi, Gqom &amp; more were sampled. To my shock, even Mashamplane and Mawlies didn’t survive.</p>

<p>“AI companies are looting our culture without consent or pay. Colonial data theft in our lifetime will not be accepted. Urgent policy reform is needed,” Mokoena added.</p>

<p>Amapiano heavyweight Kabza De Small was found to have the biggest presence across the audited AI training datasets, with frequent collaborator DJ Maphorisa also ranking among the most-used South African artists.</p>

<p>Gospel favourites Spirit of Praise and rapper Nasty C were also among the names that appeared prominently in the findings, showing just how wide-ranging the datasets are across different genres.</p>

<p>The audit suggests that music from these artists was used to help train AI systems that can create similar-sounding songs from a simple text prompt. </p>

<p>Other familiar names identified include Cassper Nyovest, Kelvin Momo, Emtee and Sjava, among hundreds of South African artists whose work was reportedly included in the datasets.</p>

<p>Related Topics:</p>

Guardian Sport · Jun 24

Guardian Sport · Jun 24

234 stories