South Africa’s rewritten national artificial intelligence policy will not reach cabinet until March 2027. That is about four months later than the government told parliament in May, and the public will see the document only once cabinet clears it for comment.
Jeanette Morwane, acting deputy director-general at the Department of Communications and Digital Technologies (DCDT) and project manager for the policy, set out the timetable last week at a South Africa-Finland AI dialogue. The Finnish embassy hosted the event at the National Research Foundation in Pretoria. The drafting team must hand the revised policy and a discussion document to Communications Minister Solly Malatsi by 30 November, and consultations within government will follow from December to February.
Cabinet approval in March would mainly authorise gazetting the policy for public comment. The department would then consolidate submissions and return the document to cabinet for final sign-off, a step for which no date has been set. Morwane put the longer timeline down to the amount of government process needed before the policy can go to cabinet.
| Date | Milestone or target |
|---|---|
| April 2024 | Draft national AI plan discussion document released |
| August 2024 | National AI policy framework published |
| 25 March 2026 | Cabinet approves first draft for comment |
| 10 April 2026 | First draft gazetted for 60-day comment |
| 26 April 2026 | Draft withdrawn over fictitious references |
| May 2026 | DCDT tells parliament: cabinet in November, comment in January 2027 |
| July 2026 | Morwane: cabinet in November, January at the latest |
| 30 November 2026 | Revised draft due to minister |
| March 2027 | Revised draft due before cabinet |
The new date is the latest in a series of shifting targets. In September, Malatsi said at GovTech that he aimed to have a draft before cabinet by the end of the financial year, which closes in March.
The revision follows the collapse of the first draft. Malatsi withdrew it 16 days after it was gazetted, when News24 reported that its reference list contained fictitious sources, some credited to journals that never published the work. The minister said the failure went beyond a technical lapse, and two officials were suspended. The withdrawn version had proposed a National AI Commission, an AI Ethics Board and an AI Regulatory Authority, along with tax breaks, grants and subsidies.
In May, Malatsi appointed a seven-member independent expert panel chaired by Benjamin Rosman of Wits University’s Machine Intelligence and Neural Discovery Institute. The other members are Vukosi Marivate, Alison Gillwald, Heather Irvine, Tshepo Feela, Jabu Mtsweni and Lufuno Tshikalange. Each wrote advisory notes in response to a framing document from the minister. A DCDT-led drafting team is now analysing those notes; it includes the Presidency, the Department of Science, Technology & Innovation, the State Security Agency and the Department of Trade, Industry & Competition. An implementation road map is being drafted alongside the policy so that work can start soon after approval.
The redraft centres on compute and data sovereignty, and on what officials call model sovereignty: large language models that work in South Africa’s official languages. Morwane said investment in computing infrastructure was a major need.
The delay matters for business as well as for regulation. MarketsandMarkets values South Africa’s AI market at $2.85-billion (about R50-billion) in 2026. It forecasts growth to $17.2-billion (about R301-billion) by 2031, a compound annual rate of 29.3%. Until a final policy is in place, companies and state bodies using AI rely on self-regulation.
Regional peers are moving ahead. A slide shown at the Pretoria event said South Africa leads the region in AI readiness, but that its neighbours are implementing faster. The slide cited Egypt’s Karnak, a national large language model unveiled in February. Egypt launched it with a refreshed AI strategy that targets a $42.7-billion (about R747-billion) contribution to GDP by 2030. Kenya, Nigeria and Ghana are also pushing ahead with national AI strategies.