Do not let AI steal our culture

South Africa’s National AI Policy Framework is an important statement of intent. It places artificial intelligence within a broader programme of digital transformation and recognises the need for ethical, inclusive and responsible development.

As the country moves from framework to implementation, the creative economy should be treated as a governance priority, not an afterthought.

Music, film, broadcasting, publishing, performance and design do more than generate commercial value. They preserve language, memory and identity.

They also supply the recordings, images, text and cultural patterns on which generative AI systems can be trained.

If South Africa wants an AI ecosystem aligned with its people, the people who create its culture must have a meaningful role in determining how that material is used.

The challenge is not to stop experimentation. AI can help independent artists develop ideas, manage catalogues, reach audiences and reduce administrative burdens.

The challenge is to distinguish authorised assistance from extraction and impersonation.

The gap between principle and remedy

Most responsible-AI frameworks contain familiar principles: transparency, fairness, accountability, privacy and human oversight.

Those principles matter but a creator facing an unauthorised voice clone needs more than a statement of values.

The person needs to know who generated the material, what account distributed it, whether relevant evidence will be preserved, how to request removal and where to appeal if the first decision is wrong.

That is the difference between ethics as aspiration and governance as an operating system.

South Africa’s implementation work should be judged by whether people can exercise rights in practice, especially when they lack the resources of a major label, broadcaster or technology company.

Four protections to build now

The first is meaningful consent for commercial digital replicas.

Permission should be specific about purpose, duration, territory, modification, transfer and compensation. Contracts should not bury synthetic-use rights in vague language that a performer cannot reasonably evaluate.

The second is provenance.

Providers and commercial deployers should support reliable records that identify when media has been generated or materially altered by AI. Provenance will never be perfect but it can make authorised uses easier to verify and deceptive uses easier to investigate.

The third is accessible redress.

Platforms and distributors should maintain a standardised process for complaints concerning impersonation and unauthorised use.

The process should include evidence preservation, human review, deadlines, an appeal mechanism and safeguards against fraudulent takedown requests.

The fourth is cultural-sector AI literacy.

Artists and small creative businesses need practical education that joins innovation with rights management. This includes understanding licensing terms, data privacy, content credentials, collective-management options and the points at which professional legal advice is necessary.

Government procurement and public funding can turn these protections into market expectations. AI services used by departments, broadcasters, schools and cultural institutions should be required to document data practices, security controls, human oversight and complaint procedures.

Grants supporting AI-enabled cultural projects should budget for rights clearance, community participation and provenance rather than treating those activities as optional administration.

South Africa could also establish a small number of supervised pilots across music, broadcasting and multilingual publishing.

The purpose would be to test consent records, authorised voice models, content credentials and rapid complaint handling with real creators and small enterprises.

Independent evaluation should ask whether the tools work for people with limited money, bandwidth and legal support — not only whether they satisfy large institutions. Results should be published in plain language so creators, policymakers and developers can see which safeguards succeed, which fail and what must change before wider deployment.

Africa-centred does not mean technology-centred

The African Union’s Continental AI Strategy calls for a people-centred, development-oriented and inclusive approach while preserving African culture.

South Africa can help demonstrate what that commitment looks like in a sophisticated creative market with multiple languages, histories and unequal access to power.

Implementation should include independent creators, unions and guilds, collecting societies, broadcasters, universities, cultural institutions, technologists and representatives of language communities.

Their task should be practical: develop model consent clauses, data documentation expectations, provenance guidance, complaint procedures and indicators that show whether smaller creators can actually use the system.

Unesco’s 2026 global creativity report warns that generative AI could deepen inequalities and contribute to significant revenue losses for music and audiovisual creators. Those forecasts are not destiny. They are a reason to act before market practices harden into norms.

South Africa can pursue AI growth while refusing a model in which cultural value is collected from many and controlled by a few.

The strongest framework will be one that makes innovation possible, consent visible, accountability enforceable and participation broad enough to reflect the country itself. Michael Neely is a veteran music executive, songwriter, author, educator and keynote speaker with more than four decades of experience in music and entertainment. As CEO of Global Music Conference and XS10 Magazine, he focuses on responsible AI, creator rights, copyright, education and creative industry transformation.