Is Your Music Training an AI? What Tanzanian Artists Must Do Before It’s Too Late
Earlier last week, Atlantic reporter Alex Reisher exposed four major music datasets used to train AI models and launched a fully searchable public database. The discovery revealed one data set containing 12 million tracks, while another holds 9 million. The remaining two datasets, though smaller, still represent millions of songs worth of training material. Nigerian IP and entertainment lawyer Rita Chindah published a post that stopped the African music industry mid-scroll. She described how artist Kimani Victoria had discovered through no action of her own that 22 of her songs had been included in The Atlantic’s AI training database. She had not submitted them. She had not been asked. She had not paid.
That discovery is not an isolated case. It is a symptom of a global problem that is arriving in Africa faster than the legal frameworks designed to address it.
What AI Training Databases Are and Why They Matter
Generative AI music tools are platforms that produce original-sounding compositions, vocals, and arrangements using machine learning and are trained on large datasets of existing music. To teach an AI system how music sounds, how harmonies work, and how lyrics are structured, developers feed it enormous libraries of recorded and written works.
The legal question at the heart of this crisis is whether using copyrighted music to train an AI model without the rights holder’s permission and without compensation constitutes infringement. In the United States, over 100 AI copyright lawsuits are currently pending. The Sony v. Suno/Udio case, which is expected to produce a ruling this summer, will be among the most significant copyright decisions in decades. Whatever outcome it reaches will have downstream effects on how AI companies operate globally, including in Tanzania.
The Tanzanian Legal Context
Tanzania’s Copyright and Neighbouring Rights Act protects creators’ rights in their original works from the moment of creation. The exclusive rights granted under this framework include the right to reproduce, adapt, and communicate a work to the public. Whether the use of a protected work to train an AI system falls within one or more of these categories remains a question not yet definitively resolved in Tanzanian courts.
That absence of precedent cuts both ways. It means Tanzanian artists cannot yet rely on a court ruling to protect them. But it also means the question remains open and the legal arguments available to rights holders are far from exhausted.
What Tanzanian Artists Can Do Now
The most important immediate step is to understand your exposure. Artists who distribute their music through digital platforms Boomplay, Spotify, Apple Music, YouTube, and others should review the terms of those distribution agreements carefully. Some platforms have updated their terms in ways that affect how they respond to AI data requests. Some have not.
Beyond distribution terms, artists should consider these practical steps. First, search publicly available AI training database records to determine whether your work appears. Second, document your ownership rights clearly: registration with the Copyright Society of Tanzania (COSOTA) provides evidence of authorship and ownership that will be relevant in any future dispute. Third, consult a legal professional before signing any agreement with an AI platform, a music tech company, or any entity that seeks a broad license to use your works.
Quinn Law Chambers is monitoring the AI copyright litigation landscape closely and advising artists on how to audit their existing licensing arrangements for AI exposure. The legal window to act is open, but it is unlikely to remain open indefinitely.
The fundamental question every Tanzanian artist should be asking right now is simple: if your music is making an AI smarter, should you not at least know about it?
For more information and services, contact us via info@qlc.co.tz.
