AI can affect everything from delivering better public services to promoting economic growth. It can improve healthcare and enhance education. Thus, AI influences public services, health, education, and growth. Governments are testing the waters with the technology. They look at how it can aid development and support public services and economic growth.
South Africa is one of the nations that is actively participating in these debates. South Africa is attempting to define a national AI strategy. It participates in a broader continent-wide debate on AI development and regulation. By participating in the work of the AU and engaging with worldwide partners, South Africa helps shape AI frameworks. These frameworks foster innovation and ensure safety, transparency, accountability, and responsible use of the technology.
Yet the country’s progress has revealed a problem. Crafting artificial intelligence policies is about more than inserting new technology into a wider system. Governments must establish appropriate policies. These policies must be implemented correctly. They must protect the public’s safety and robustness.
One example is the recent controversy over South Africa’s draft national AI policy. It was removed after AI-generated references and incorrect sources were found.
South Africa’s Role in Shaping Africa’s AI Governance
Growing African-owned AI governance in South Africa?! If you’re asking yourself this question, you’re not alone! How can Africa use artificial intelligence to attain its lofty aspirations? AI development in Africa is not just a matter of foreign policies. It is essential to the wider discussion of the continent’s AI development.
That’s because for the African Union, AI is an enabler for the continent to accomplish long-term goals, such as Agenda 2063 and the UN SDGs. But the deployment of AI poses distinct challenges for African countries: digital infrastructure isn’t always available; technology is difficult to access and expensive; the talent shortage and data protection concerns create hurdles to consensus around the technology.
What’s happening? The African Union published its Continental Artificial Intelligence Strategy in 2024. The goal of the AI strategy is to enable the member states of the union to develop a harmonized AI ecosystem that will accelerate economic, social, and research productivity.
Globally, in May 2025, there was a high-level policy dialogue on Africa’s AI development and regulation. This forum was attended by government representatives, global organizations, civil society groups, academics, and industry players. The result? A demand for a continent-wide AI policy and governance framework that will promote collaboration, ethical AI development, and safeguarding of human rights.
Participation by South Africa in these conversations allows us to inject our African perspective on AI. We need to ensure that international norms on AI are contextual to our continent’s social, economic, and technological conditions, as opposed to importing constructs from elsewhere.
What Was the Controversy Surrounding South Africa’s Draft AI Policy?
One of the most widely discussed developments in South Africa’s AI governance journey occurred in April 2026, when the government released its draft National Artificial Intelligence Policy for public comment.
The document was intended to establish a clearer direction for how AI should be developed, regulated, and used across the country. It covered several important areas, including technological innovation, economic development, ethical AI practices, human rights, and the responsible use of data.
The draft also proposed creating new institutions to oversee AI development and regulation, including a National AI Commission, an AI Ethics Board, and an AI Regulatory Authority. These bodies were intended to strengthen oversight, encourage responsible innovation, and help address the risks associated with AI technologies.
However, the policy soon faced an unexpected problem.
After the document was released, concerns emerged that several references in its bibliography pointed to sources that did not actually exist. An internal review confirmed that the document contained fictitious references, with AI-generated citations included without proper verification identified as the likely explanation.
On 26 April 2026, Communications and Digital Technologies Minister Solly Malatsi announced that the government was withdrawing the draft policy. He explained that the issue was not simply a technical error but one that affected the credibility and integrity of an important government document.
The withdrawal was subsequently formalized through a government notice published in June 2026.
The incident attracted attention because it involved a government policy specifically designed to guide the responsible use of artificial intelligence. It raised an important question: if AI-generated information can enter official government documents without being properly checked, how can institutions ensure that AI is being used responsibly in other areas of public administration?
Why the Incident Matters for AI Governance
The withdrawal of the draft policy highlights a challenge that governments and organizations around the world are beginning to face: AI-generated information cannot automatically be treated as accurate or reliable.
AI systems can produce convincing explanations, research summaries, statistics, and references. However, they can also generate incorrect information, invent academic sources, or present unsupported claims as facts. This behavior is commonly known as AI hallucination.
When such errors appear in everyday content, they can cause confusion and misinformation. But when they enter government policies, legal documents, public reports, or official research, the consequences can be much more serious.
Government policies influence decisions that affect millions of people. They guide public spending, establish regulatory responsibilities, influence business activities, and determine how new technologies are introduced into society.
If these documents contain inaccurate information, they can weaken public trust and create uncertainty among the institutions and communities expected to follow them.
The South African case demonstrates that using AI in government work requires clear responsibilities and strong human oversight. AI can assist with research, writing, summarizing information, and analyzing large amounts of data, but human experts must verify important information before it becomes part of an official document.
It also shows why governments need clear guidelines for the use of AI in public-sector institutions. These should cover source verification, fact-checking, transparency, data protection, accountability, and the responsibility of officials who approve AI-assisted work.
The Future of AI Governance in South Africa: What’s Happening?
Did you read about the new developments in South Africa’s AI management? The government isn’t giving up after the draft policy papers have been withdrawn. They’re taking another shot at developing a robust AI national strategy.
In May 2026, Communications and Digital Technologies Minister Solly Malatsi announced to Parliament that an independent panel of seven experts had been appointed to review the draft. They’re tasked with examining what’s wrong with the document, resolving the wording, and making the necessary changes ahead of public consultation.
For the record, the government has a schedule. They want to bring the new policy to cabinet in November 2026, and they want a new version to be ready for public input in January 2027. But it’s worth noting those are rough targets; the new policy isn’t ready yet.
Why should this concern us? It would be helpful to have a national framework in SA that encourages the development of tech but manages the risks associated with AI. The updated policy could give organizations, government departments, and businesses that use AI tools a more precise idea of how they should operate. In essence, our problem now is to formulate policies that are flexible and flexible enough to operate in a quick-paced world while still putting human safety, fairness, and accountability first.
What This Means for NGOs and Nonprofit Organizations
Developments in South Africa’s AI governance are also relevant to NGOs, nonprofits, social enterprises, and humanitarian organizations working across Africa.
Many NGOs are beginning to explore AI tools to improve their everyday operations. These technologies can help organizations analyze project data, prepare funding proposals, identify communities in need, translate information into local languages, and improve program planning.
However, greater AI adoption also creates new responsibilities.
For organizations working with vulnerable communities, inaccurate AI-generated information can affect important decisions. For example, an incorrect analysis of community needs could influence how resources are distributed. Poorly verified research could weaken a funding proposal, while unsafe handling of personal information could expose beneficiaries to privacy risks.
As AI becomes more common in the nonprofit sector, organizations should consider a few important practices:
- Verify AI-generated information: Always check statistics, research findings, references, and other important information before using them in reports or proposals.
- Protect beneficiary information: Avoid entering sensitive personal details into AI tools without understanding their privacy and data protection policies.
- Maintain human oversight: AI should support NGO professionals rather than replace their experience, community knowledge, and decision-making.
- Promote transparency: Organizations should establish clear guidelines explaining when and how AI is used in their operations.
- Build AI awareness: Staff members should receive basic training to understand AI’s limitations, recognize inaccurate outputs, and use these tools responsibly.
For smaller NGOs with limited resources, these practices are particularly important. Adopting AI should not simply mean working faster; it should also mean making informed, responsible, and trustworthy decisions.
The Challenges Facing Africa’s AI Future
Although South Africa and other African countries are making progress in developing AI strategies, several challenges still need attention.
1. Limited Digital Infrastructure
Many African communities continue to face limited internet access, unreliable electricity, and insufficient digital infrastructure. Without addressing these issues, the benefits of AI may remain concentrated in larger cities and economically developed areas.
2. Shortage of AI Skills and Expertise
Developing and managing AI systems requires trained professionals, researchers, and technical experts. African countries need greater investment in education, digital literacy, and professional training to strengthen their local AI capabilities.
3. Data Privacy and Security
AI systems depend heavily on data. Governments and organisations must ensure that personal information is collected, stored, and processed responsibly, particularly when dealing with sensitive information belonging to vulnerable communities.
4. Dependence on International Technology Providers
Many advanced AI systems and infrastructure are developed by companies outside Africa. Building local research capacity, supporting African technology companies, and encouraging regional partnerships will be important for reducing technological dependence.
5. Building Public Trust
People need to understand how AI is being used in decisions that affect their lives. Governments must ensure that AI systems are transparent, accountable, and subject to appropriate human oversight.
Addressing these challenges will require cooperation between governments, universities, technology companies, civil society organisations, and international development partners.
The Road Ahead: Responsible AI Development in Africa
What is occurring in South Africa is not an isolated case but part of a larger trend of government utilizing AI in the public sector. This offers chances for governments to provide more effective and innovative public services, provided they are implemented properly with strong policies.
Other AU initiatives aimed at establishing a unified strategy for AI management indicate the potential for regional collaboration. African countries need governance strategies that align with their development objectives, protect their populations, and enhance their access to the global AI economy.
The choice to halt South Africa’s draft policy due to errors, accountability, and human involvement highlights the importance of these principles in policy formation. It also serves as an important lesson for organizations introducing AI: while technology can enable good outcomes, ultimate responsibility lies with human users.
It will be intriguing to observe how South Africa’s AI policy develops as the nation works on a new draft of its national AI policy. The choices made today will influence not just the continent but also shape future governance of AI.
Ultimately, the speed of Africa’s adoption of artificial intelligence is not the only factor shaping the continent’s AI future. The responsible development, governance, and deployment of technology for the benefit of its people are equally crucial.
