In April 2025, South Africa’s National AI Policy framework will be sent for official evaluation as part of a gazetted process – the culmination of months of soliciting and reviewing public input. Officials acknowledge there is still work to be done, but are confident of the framework’s value proposition.
This was communicated to delegates at the ITWeb AI Summit 2025 hosted in Bryanston, Johannesburg yesterday.
In his opening keynote presentation, Dumisani Sondlo, acting director: Information Society Evaluation and Impact Assessment, Department of Communications and Digital Technologies (DCDT), said: “We have the contribution of all stakeholders and the public… we are not going to have politicians dictate how things need to be done. So, we’ve brought questions directly to stakeholders.
“This is based on a policymaking cycle. So instead of taking potshots at government, we’d rather have people contest ideas.”
As expected, the impact of AI on jobs was a major focus.
Sondlo cited the World Economic Forum Future of Jobs report and pointed out that according to the research, 22% of jobs globally will face disruption by 2030, 170 million new roles will be created, and 92 million roles displaced.
He referenced the role of organisations such as the AIISA (AI Institute of South Africa) and PC4IR (Presidential Commission on the 4th Industrial Revolution), which can help to streamline awareness of developments in AI, including technology hubs at academic institutions.
He emphasised that both challenges and opportunities exist with regard to with AI, including regulation battles to keep up with advances in technology, social demand, social equity, sustainable development, the digital divide, historical inequities and global leadership.
“If you don’t work out how to govern AI today, you are then playing by other people’s rules. Africa’s voice cannot be ignored when it comes to AI,” Sondlo said.
Strategic pillars
Sondlo added that SA’s AI policy framework comprises several strategic pillars, including professional responsibility, promotion of cultural and human values, safety and security, as well as human control of technology.
“We’ve adopted a human-centric approach in AI systems,” he added.
According to the DCDT, this approach focuses on human-in-the-loop (HTL) – especially GenAI – and specific thresholds for human intervention, decision-making frameworks to prioritise human judgment, and the option to engage with real humans. The other pillars include transparency and sufficient explainability, fairness and mitigating bias.
The framework places a premium on research, development and innovation, he said. This covers regulatory sandboxes, research into AI’s technical complexities, export of IP and highly capable AI foundation models.
Sondlo also underlined government’s strong focus on public sector implementation of AI.
This incorporates entrepreneurial public sector, incentives for key AI research, AI maturity investment, transparent AI procurement, digitalisation, public value creation and pilot projects, along with citizen engagement.
“We’ve got to use regulation to preserve democracy,” said Sondlo, who added that the GNU envisages multiparty collaboration and alignment between several regulatory bodies, including the Information Regulator, ICASA and the Competition Commission.
The DCDT said privacy and data protection are priorities, with public input emphasising a risk-based privacy approach, secure data handling and connecting with the Cybercrimes Act, Consumer Protection Act, ECT Act as well as global regulation governing data sovereignty and data exchange.
Share