Credibility is the real challenge facing South Africaâs AI policy

This story has significance for readers across Africa and beyond.

On 26 April, just two weeks after South Africaâs draft AI policy was released for public comment after a process that began in 2020, South Africans woke up to astonishing news.

Minister of Communications and Digital Technologies Solly Malatsi said: âFollowing revelations that the draft National Artificial Intelligence Policy published for public comment contains various fictitious sources in its reference list, we initiated internal questions, which have now confirmed that this was the caseâ¦

âThis failure is not a mere technical issue, but has compromised the integrity and Âcredibility of the draft policy. As such, I am withdrawing the ... policy.â

In an effort to manage the crisis, Malatsi hand-picked seven experts to review the draft. Although this might seem like a good move, it raises concerns that are not necessarily about the competency of the individual experts.

Rather, these concerns are centred on the challenge of assigning seven professionals to revise a complex technology policy that is evolving faster than any regulatory system can adapt and cuts across all disciplines, with greater societal implications than any other technology humanity has ever invented.

Because of AIâs rapid evolution, which far outpaces the bureaucratic timelines of traditional policymaking, conventional policy cycles and governance procedures are increasingly inadequate. Also, AI operates in a globally interconnected ecosystem of data, algorithms, computing power and cloud infrastructure.

To govern AI effectively, there is a need for an adaptive governance mechanism, continuous multi-stakeholder engagement and regular policy updates to ensure the governance frameworks remain relevant and effective.

AI is increasingly becoming integral to all our daily lives, from algorithms curating our news feeds to the systems approving our loan applications. The stakes extend beyond technical implementation: fundamentally, AI is redefining humanity in the 21st century and beyond. It is reconfiguring how we work, learn and make decisions, and how we relate to one another.

For this reason, effective AI governance is urgently required, and if South Africaâs AI policy is to serve the more than 65 million people who live here, it must be built on credibility and trust.

Credibility through process

The credibility of any AI policy depends on two pillars. First, on who controls the infrastructure underpinning the policy. Second, on the integrity, transparency and inclusiveness of the process that produces it.

South Africaâs AI policy is facing a credibility challenge that goes beyond the withdrawal of the policy draft. The key question now is whether the government can rebuild public trust through a transparent, participatory and accountable process. The appointment of seven experts in good faith does not automatically restore the legitimacy lost when fictitious AI-hallucinated citations were first exposed in a News24 report.

Given AIâs complexity, an effective policy-drafting process requires multidisciplinary expertise and multi-stakeholder collaboration that extends beyond AI applications to the broader ecosystem. A closed circle of seven experts, regardless of their qualifications and competencies, cannot provide the diverse perspectives that can emerge from the open and inclusive dialogue needed to address a complex technology such as AI.

Towards a peopleâs policy

In July, I contributed to the South African National AI Policy Dialogue at the University of Johannesburg. Legal minds, policy gurus, media practitioners and researchers came together to grapple with these challenges.

Public trust in the development of AI policy depends on the government taking people into its confidence about the processes and procedures it follows. The mirage of âblind trustâ, presented as sparing people the complexity of the subject, is easily dispelled by societyâs daily contact with, and the impact of, this technology.

Furthermore, its fast-evolving nature means that no single party is fully attuned to the nature of AI. Thus, the policy and its process are a peopleâs policy, as people are the one constant in the differentiation of AI engagement.

Malatsi has committed to these timelines: by September 2026, the panel is to present the revised version of the AI policy draft to the minister; by November 2026, the minister will submit the revised version to the Cabinet; and by January 2027, the revised draft is to be published for a new round of public consultation â in other words, back to where we left off in April.

Even though meeting these deadlines matters, meeting them with a closed panel of seven will not restore the credibility that has been lost.

Widening scope for input

Public trust is earned through meaningful and continuous engagements from all sectors of society. This certainly includes input from tech and legal experts, but because of AIâs far-reaching societal impact, also from philosophers and psychologists.

To rebuild trust, Malatsi should adopt a multi-stakeholder, multidisciplinary drafting process for the policy. This entails regular progress updates, clearly communicating what is happening and who is responsible for each stage.

In doing so, a meaningful opportunity for participation will be created throughout the drafting process, not just during formal consultations. This process requires that the minister leverage multiple platforms such as national television, community radio stations, social media and other digital channels to ensure broad public engagement and avoid future shocks.

The withdrawal of the draft was not just a flawed document being discarded, but also a loss of public trust in the governmentâs ability to govern AI. Therefore, in future, the minister and his team must prioritise process as much as outcome, ensuring that South Africaâs AI policy is well written and technically sound, as well as democratically legitimate.

The clock is ticking. However, credibility and trust depend on demonstrated commitment, not quick fixes. DM

Dr Rejoice Malisa-van der Walt is a postdoctoral research fellow at the Centre for Information Integrity in Africa at Stellenbosch University.

This story first appeared in our weekly DM168 newspaper, available countrywide for R35.

Reporting originally appeared via Daily Maverick. Read the full source for additional context.