Under current interpretations of South African law, the person who prompts an AI tool is generally regarded as the most likely copyright owner of the resulting work. But law firm Wright Rose-Innes has acknowledged this approach is far from settled - and the uncertainty carries real consequences for writers who use these tools professionally.
The firm's analysis hinges on the idea that the user directed the creation process through their input. Yet that reasoning frays when the human contribution shrinks. "Where the user's contribution is minimal, for example, where a simple or generic prompt produces a highly complex output, it becomes difficult to justify full ownership based on human creativity alone," Wright Rose-Innes said.
How copyright works in South Africa
Copyright arises automatically the moment a creative work is written down, recorded, or saved in a tangible form. No registration is required for most works. The owner gains exclusive rights to reproduce, distribute, perform, or license the work - and to stop unauthorised alteration or misuse.
The law covers literary works, artistic creations, musical compositions, films, sound recordings, broadcasts, and computer programs. South African copyright law also recognises that works generated by computers may qualify for protection. The problem is deciding who the author is when a machine does the heavy lifting.
Several candidates exist: the developer who built the AI system, the company that owns or operates it, or the user who supplied the prompt. Wright Rose-Innes said the central challenge is "whether it remains appropriate to attribute ownership to a human user in circumstances where the creative contribution is minimal or indirect."
The tension between AI and traditional authorship
Generative AI tools such as ChatGPT and DALL-E produce text, images, and code by processing prompts against vast datasets and learned patterns. The outputs can be sophisticated, but the creative distance between a short prompt and a complex result tests the boundaries of copyright law, which has always rested on human authorship and originality.
Many AI service providers try to paper over this gap by including terms that say users retain ownership of what they generate. Wright Rose-Innes said these contractual provisions help in practice but "do not fully resolve the underlying legal ambiguity." A clause in a terms-of-service document cannot rewrite statutory law.
"If AI can independently produce complex works, the law must grapple with whether such outputs should be protected at all, and, if so, in whose hands those rights should rest," the firm said. South Africa's current framework leaves those questions unanswered as AI systems grow more autonomous.
Why this matters for writers
If you write for a living and use AI tools to draft, brainstorm, or polish text, the copyright status of that output affects your ability to license it, sell it, or stop someone else from using it. A client who pays for AI-assisted copy may find the ownership chain is weaker than either of you assumed.
Wright Rose-Innes said legal reform or clearer judicial guidance may be needed to keep copyright law "both fair and relevant." Until that arrives, writers should read the terms of any AI tool they rely on and document their creative input - because the difference between a generic prompt and a detailed, iterative creative process may one day determine who actually owns the work.
For writers building a career around words, the tools that speed up production also introduce a legal question mark. The AI for Writers conversation now includes not just craft and efficiency, but ownership - and the answer depends on where you live and how much of yourself you put into the prompt.
Your membership also unlocks: