South African musicians, including Kabza De Small, Spirit of Praise, and Thomas Chauke, have been identified in artificial intelligence (AI) training datasets without their consent, raising significant concerns over intellectual property rights and the future of creative works. The widespread inclusion of their material without explicit permission has ignited a debate within the music industry. Kabza De Small, a prominent figure in the Amapiano genre, appeared 360 times in these documented datasets, noting the extensive nature of the alleged scraping. Spirit of Praise, a renowned gospel ensemble, was present 212 times, further illustrating the breadth of artists affected. Similarly, Thomas Chauke, a celebrated Tsonga disco musician, was identified 183 times within these same datasets. This alleged scraping of music for AI training could significantly impact artists' livelihoods, potentially undermining their ability to earn fair compensation from their creative output. The International Confederation of Societies of Authors and Composers (CISAC), a global network representing creators' rights, estimates that unlicensed generative AI could divert up to 25% of creators' royalties globally, potentially costing musicians substantial income and threatening the sustainability of creative professions. The implications extend beyond individual artists, posing a challenge to the entire music ecosystem.
Industry Responds to AI Threat
In response to these growing concerns, the sixth Africa Rising Music Conference (ARMC) convened at Constitution Hill, a historic site in Johannesburg, to address creator rights amidst the rapid rise of automation and artificial intelligence technologies. The conference served as a key platform for stakeholders to discuss strategies for protecting artistic integrity and economic interests in an evolving digital landscape. In a strategic move to expand discussions on artificial intelligence and its impact on the music industry, the Berlin AI Think Tank established a significant presence in Johannesburg. This expansion resulted from a collaborative partnership involving several key organizations: Paradise Worldwide, a leading music distribution and services company; AIxchange, a specialist in AI innovation; the Association for Electronic Music (AFEM), representing the global electronic music community; the Fraunhofer Institute for Digital Media Technology (FIDMT), a renowned research institution; the Southern African Music Performance Rights Association (SAMPRA); the Composers, Authors and Publishers Association (CAPASSO); and the Southern African Music Rights Organisation (SAMRO). This multi-faceted collaboration shows the international and inter-organizational effort required to tackle the complex challenges posed by AI.
To tackle the complexities of AI influence and provide concrete solutions for rights management, AIxchange and FIDMT jointly developed Creative Weight Attribution (CWA). CWA is a patent-pending system specifically designed to quantify the degree to which an artist's work influences AI models, offering a potential mechanism for tracking and attributing usage. This innovative tool aims to provide transparency and accountability in the use of copyrighted material by AI systems. The Southern African Music Rights Organisation (SAMRO) is actively supporting initiatives like the Berlin AI Think Tank, recognizing the urgent need for a unified approach to AI governance in music. SAMRO is collaborating with lawmakers to formulate full regulations for AI compensation to creators, advocating for policies that ensure artists are fairly remunerated when their work is utilized in AI training. A SAMRO spokesperson articulated the organization's stance, stating, "Legislative guidelines will be instrumental in ensuring the appropriate protections and safeguards are in place for the benefit of those utilising these services." This emphasizes the critical role of legal frameworks in safeguarding artists' rights in the AI era.
South Africa's existing copyright framework is governed by the Copyright Act of 1978, which predates the advent of advanced AI technologies. Entertainment lawyers across the country are closely monitoring the pending Copyright Amendment Bill, a key piece of legislation that seeks to update the nation's intellectual property laws. This bill includes specific provisions regarding the use of 'fair use' or 'fair dealing' by AI developers, which could have profound implications for how AI systems are legally permitted to ingest and process copyrighted material. The debate around these provisions centers on balancing innovation with the protection of creators' rights.
Protecting Content and Navigating Legalities
The issue of unauthorized use extends to other prominent artists. DJ Maphorisa, another influential figure in the South African music scene, appeared 270 times in documented AI training datasets, further illustrating the extensive and widespread use of artists' work without explicit permission or compensation. Nasty C, a highly successful rapper, was also identified 193 times within these same datasets, indicating that artists across various genres are being impacted. Rights holders contend that commercial AI training should necessitate licenses because the content generated subsequently competes directly with original music created by human artists. They argue that allowing AI models to learn from and then produce content that directly rivals human creativity without proper licensing constitutes an unfair commercial advantage and undermines the value of original artistic expression. This perspective notes the economic threat posed by unlicensed AI use.
In response to these challenges, Gallo Records, one of Africa's oldest and most respected music labels, is implementing proactive measures to safeguard its content from AI scraping and unauthorized training. The label is establishing what it terms a 'walled garden', a strategic approach aimed at controlling access to their extensive catalog and preventing its unauthorized use by AI systems. This strategy involves technical and legal barriers designed to protect their intellectual property. Businesses that utilize AI-generated content without first verifying its ownership face several significant risks, including potential infringement claims from rights holders, substantial legal liability for copyright violations, and severe damage to their brand reputation. The legal and ethical implications of using AI-generated content without proper diligence are becoming increasingly clear, prompting businesses to exercise caution and ensure compliance.
Case Study: Credo V. Daniels
The complexities surrounding AI and music creation were starkly noted in the case of Credo V. Daniels, a TikTok sensation whose creative practices came under public scrutiny. Daniels faced widespread criticism following a live television performance on eNCA, where observers noted a significant discrepancy between his live vocals and the polished quality of his studio recordings. The incident sparked a public discussion about authenticity in music. Daniels later acknowledged that he utilized AI-assisted tools in his creative process, a revelation that further fueled the debate about the role of technology in artistic production. Adding to the controversy, allegations also surfaced claiming that Daniels' song 'Ngafa' copied elements from Hozier's critically acclaimed track 'Cherry Wine'. These accusations raised questions of originality and potential plagiarism in the context of AI-aided creation. Following these developments and the ensuing public and industry discussion, Apple Music took action by removing Credo V. Daniels' debut album, 'Still Where We Were', from its platform, showing the serious implications for artists when questions of AI use, originality, and copyright infringement arise.
