US lawmakers are calling for the brakes to be applied to AI development, warning that increasingly powerful systems could escape human control.
The data behind the warning is specific rather than abstract. A 2026 comparison of AI exposure across African labour markets, built on Statistics South Africa's Quarterly Labour Force Survey and ILO methodology, gives South Africa an exposure score of 4.84 out of 10 , the highest in the region with disruption classified as imminent within one to three years. The reason is structural: 15.4% of employed South Africans work in clerical support roles and 12.5% are managers, both categories with well-documented automation exposure internationally. This is the flip side of the same 2026 Global Outsourcing AI Readiness Index that ranked South Africa eighth globally and first in Africa with a score of 66.5, driven by unusually strong enterprise adoption and workforce literacy scores. Readiness and exposure are, in effect, two readings of the same underlying fact: AI is already embedded deeply enough in South African workplaces to move fast, for better or worse.
South Africa's response , anchoring the transition in NEDLAC-style social dialogue between government, business, labour and civil society has a real precedent, and it is not the one usually invoked in African policy debates. Denmark's "flexicurity" model, developed from the 1990s onward, paired relatively easy hiring and firing with strong income support and aggressive, continuously funded retraining, explicitly so that automation and trade shocks would shift workers between jobs rather than out of the labour market entirely. It has kept Danish unemployment structurally low through multiple waves of technological disruption precisely because retraining infrastructure was built before mass displacement occurred, not after. The cautionary counterexample is the wave of manufacturing automation that swept the US Rust Belt from the 1980s onward, where job losses arrived faster than retraining systems could absorb them and entire regions never fully recovered economically. Ramokgopa's insistence that "AI and new digital technologies must not leave SA's workforce behind" is, in substance, a bet that South Africa can build its version of the Danish model before it needs it, rather than discovering, as Detroit did, that the retraining infrastructure should have existed years earlier.
