South Africa needs to move beyond its longstanding “wait-and-see” approach to artificial intelligence (AI) regulation and establish a binding, constitutional framework that protects human rights while allowing technological innovation to continue.
This is according to University of Cape Town (UCT) PhD graduand Nokuthula Olorunju, in her thesis Artificial Intelligence, Bias and the Rule of Law: A Constitutional Approach to the Regulation of AI in South Africa.
While South Africa has previously had more pressing regulatory priorities, Olorunju argues that the rapid growth and adoption of AI mean the country can no longer delay establishing an effective regulatory framework, even if AI remains difficult to fully understand.
A constitutional framework for AI regulation
Olorunju’s study used theoretical, doctrinal and comparative research methodologies. Drawing on a constitutional theoretical framework, she examined AI regulation through law and technology, human rights, constitutional law and international law.
Her research examined several regulatory approaches and concluded that there is no perfect way to regulate AI. Countries should instead start with what they can implement, given their regulatory constraints.
In South Africa, she identified fragmentation, lack of implementation and longstanding legal silos as factors that can allow AI misuse to thrive. Legislative fragmentation can result in jurisdictional overlap, confusion over authorisation processes, conflicting mandates and uncertainty about which bodies are responsible for particular aspects of governance.
“In essence, if a gap can be exploited, it most probably will be,” she says.
Olorunju proposes a constitutional approach as a hybrid framework that could help close these gaps while addressing practical challenges in enforcement and implementation. Her proposal includes clarifying conflicting mandates, establishing an interdisciplinary AI oversight body and avoiding unnecessary duplication of governance structures.
Why existing laws are not enough
Olorunju argues that existing legislation, non-binding ethical principles and self-regulation are insufficient to address the risks posed by AI.
While AI principles and self-regulation have helped fill some gaps, they lack the force of law. Self-regulation also allows organisations to choose principles that are easier to integrate into their missions and values.
She argues that government therefore has a duty to ensure enforceable laws are in place to address grey areas and loopholes that could allow rights to be exploited.
The widespread use of AI also requires an interdisciplinary approach, bringing together experts from multiple disciplines to inform legislation and support its practical implementation.
Protecting South Africans from AI-driven bias
The human rights implications of AI are central to Olorunju’s argument. These include AI-enhanced cybercrime, data breaches, privacy violations through deepfakes, AI systems not trained on South African data, potential racial and gender bias in lending and employment, copyright infringement and mass surveillance.
She argues that AI systems can reproduce existing inequalities when the information used to train them reflects historical biases.
“Given South Africa’s history, it’s the perpetuation of racial and gender bias, the exponential increase of the digital divide and lack of access, the exploitation of the rights of children and vulnerable members of our society, environmental degradation, theft and misuse of creative works, maladministration of resources and the perpetual lack of accountability,” she argues.
Putting the Constitution at the centre
Olorunju believes the Constitution should provide the foundation for regulating AI because it is the supreme law of South Africa.
As technology evolves, constitutional rights must also be considered in light of those developments. She argues that AI-related frameworks should therefore respect and protect the rights and values enshrined in the Constitution.
Existing constitutional institutions and sector-specific regulators could play a role, provided they are properly equipped and their mandates are clarified.
“Honestly, all of them,” she says, regarding which existing constitutional institutions could play a role in AI regulation. “The regulation of AI requires collaboration.”
Accountability and awareness
Her proposed framework would require organisations developing or deploying AI systems to demonstrate compliance with human-rights standards and to accept accountability throughout an AI system’s lifecycle.
Olorunju argues that compliance should be more than an annual “tick-box” exercise. For ordinary South Africans, this could mean clearer rules governing AI tools, assessments of AI systems, accountability measures and processes for lodging complaints about AI misuse.
She also believes public awareness about AI should be accessible in all 12 of South Africa’s official languages.
“Authentic information is power,” Olorunju says, stressing the importance of digital literacy and awareness as AI becomes increasingly embedded in everyday life.
Regulation without stifling innovation
Olorunju acknowledges that regulation must not unnecessarily discourage technological innovation or investment.
She argues that South Africa can draw on international best practices rather than simply copying regulatory models from countries such as the United States or members of the European Union. Instead, she emphasises the importance of regulatory adaptability and sustainability.