Wits project pits African creators against AI music’s blind spots

Africa’s cultural contribution to global music is not reflected in the data being used to train generative AI systems, and the continent must move quickly to secure a “seat at the table” before it becomes a consumer rather than a shaper of the technology, music industry veteran Charles Goldstuck has warned.

Goldstuck, founder of GoldState Music and a Wits alumnus, was delivering the keynote address at a showcase hosted jointly by the Wits Innovation Centre and the Wits Mind Institute on Thursday. The event concluded a six-month pilot project, the AI and African Music Project, pairing African musicians with AI engineers to build bespoke technical tools for the continent’s creative sector.

Goldstuck painted a sobering picture of the state of play for generative AI in the global music market, arguing that the gap between rapid commercial development and lagging public policy means the industry’s future will be shaped by commercial settlements rather than legislation.

“Commercial dynamics are really so far ahead of legislation,” he said.

He pointed to the diverging strategies of the two best-known AI music platforms. Suno, valued at US$2.45-billion following a $250-million series-C funding round led by Menlo Ventures in November 2025, is currently defending copyright litigation from major rights holders on a “fair use” basis. Udio, by contrast, has opted for a “walled garden” approach, striking licensing deals with Universal Music Group, Warner Music Group, Merlin and Kobalt.

Goldstuck noted that while detection tools are becoming more robust, the technology for “attribution” – tracing an AI-generated song back to the specific copyrighted works used to train it – remains in its infancy.

Immense

“It is not clear whether this technology can be effectively developed and deployed,” he warned, adding that current compensation models often rely on general market share rather than precise usage data.

The scale of the challenge is immense. While streaming giant Spotify removed more than 75 million “spammy” tracks from its platform in the past 12 months, French service Deezer has revealed it receives more than 60 000 fully AI-generated tracks every day. Deezer estimates that up to 85% of streams on AI-generated music are fraudulent, leading the platform to demonetise and remove them from the royalty pool.

Read: AI vs artists: who owns the future of music?

During a Q&A session, the conversation turned to “creative sovereignty” for African artists. Goldstuck observed that while Africa has been largely spared from the initial “ravages” of generative AI – primarily because platforms have focused on Anglo-American and Chinese repertoires – the continent still lacks a significant “seat at the table” in terms of technical infrastructure and funding.

“Africa’s cultural contribution to the global music scene is not represented in all the scraping that has happened,” Goldstuck said. He called for concerted activism from the creative community to force government participation in funding tech-music programmes.

Prof Christo Doherty, who leads the initiative via the Wits Innovation Centre, echoed this sentiment. “AI offers tremendous possibilities for African musicians, and not only challenges that need to be controlled through regulation,” Doherty said.

Prototypes

The showcase, held at the Chris Seabrooke Music Hall at Wits, featured live demonstrations of five pilot projects developed by teams from seven African countries. The projects focused on archiving, preservation and tool localisation.

The winning teams and their prototypes are:

• Zazi: A “musical digital twin” enabling real-time voice, rhythm and storytelling interaction. It was developed by South African multidisciplinary artist Umlilo and Gideon Gyimah, a Ghanaian AI engineer specialising in multilingual speech technology.
• The Bɛ̀bɛ̀i Engine: A performative AI instrument co-created with the Baka community to preserve endangered polyphonic traditions. It was built by Cameroonian artist Joshua Kroon and Emmanuel Apetsi, a Ghanaian AI/ML engineer.
• Bina.ai: An AI children’s music and storytelling platform rooted in African genres. It was created by Nigerian music strategist Ehinome Ogbeide and DRC-based technologist Muhigiri Ashuza Albin.
• Heritage in Code: A digital archive and AI-fusion engine that preserves African instrumental heritage while enabling contributor royalties. It was developed by Kenyan deejay Linda Nyabundi and Ethiopian AI researcher Gebregziabihier Nigusie.
• TIMah AI: A secure archive documenting Kikuyu traditional music with transcript workflows and community-centred consent governance. The project was led by Kenyan producer Tora Nyamosi and machine intelligence engineer Lawrence Moruye.

“The future of music technology doesn’t have to be imported. It can be homegrown, collaborative and unmistakably African,” said Apetsi during the presentation.

The projects are currently at the prototype stage, with ongoing technical support and mentorship provided by the Mind Institute and the Wits Innovation Centre.  – (c) 2026 NewsCentral Media