South Africans' Caution Culture Shields Against AI Fraud Risks

South Africa logged just 12 entries in the Artificial Intelligence Incident Database between March 2023 and April 2026, out of 1,618 cases collected globally. That fraction, less than 1 percent of all recorded AI incidents, reflects something more than luck.

A culture of caution built over decades of high crime and economic constraint appears to be giving South Africans an unexpected buffer against AI-powered scams. Citizens who routinely navigate a high-crime environment develop a skepticism that, for now, makes them harder targets than their counterparts in wealthier nations.

“We worry about everything; we look over our shoulders when we’re driving down the road. Our level of gullibility is not as high as the rest of the world,” says Alistair Pugin, founder of Celyntis, a cloud technology enabler. “Our level of crime is a blessing in disguise.”

That buffer, though, is not permanent.

As broadband expands, data costs fall, and more South Africans come online, the protective effect of limited connectivity and heightened wariness will erode. The pace of change makes the timeline urgent. Microsoft CEO Satya Nadella has argued that Moore’s Law is becoming outdated, with computing capabilities now doubling every six months rather than the traditional two-year cycle. Pugin puts it plainly: “We’ve never seen a rate of change in any industry in the history of humankind that we’ve seen with AI.”

Meanwhile, the 12 incidents already on record reveal how wide the damage can spread when AI-enabled harm does take hold. The documented cases span criminal fraud, information integrity, institutional trust, and critical public services, with South Africa appearing variously as victim, target, institution, actor, and alleged source of AI-enabled activity.

Public services have borne some of the most direct costs. The South African Social Security Agency’s eLife Certification portal suffered repeated facial-recognition and one-time password failures, leaving grant beneficiaries unable to complete mandatory verification. A parliamentary report confirmed 67,868 grant suspensions in a single quarter. In January 2026, a purportedly AI-generated image of Orlando West High School in Soweto engulfed in flames was shared with emergency services, causing public alarm and diverting resources from genuine needs.

Government institutions have been compromised from within. South Africa’s draft National AI Policy contained fictitious references believed to be AI hallucinations. The Department of Home Affairs withdrew an apparently AI-generated list of fictitious references appended to its Cabinet-approved Revised White Paper on Citizenship, Immigration and Refugee Protection. The department suspended two senior officials and commissioned an independent review of policy documents produced since November 2022.

The legal system has proved particularly exposed. In March 2023, lawyers in a Johannesburg defamation case relied on legal judgements generated by ChatGPT that were entirely fictitious, forcing the court to delay proceedings and resulting in a punitive costs order against the legal team. In the Mavundla case, AI-generated citations submitted to the Pietermaritzburg High Court, many of which did not exist, led to a referral to the Legal Practice Council. In June 2025, another South African legal team acknowledged that non-existent case citations in an urgent licensing dispute had been produced by Legal Genius AI, drawing a similar referral.

Ordinary citizens have been targeted directly. An AI-enabled scam aimed at teenagers in North Carolina deployed nearly 100 malicious websites linked to a centralised operation in South Africa, using fake local headlines and students’ likenesses to lure visitors toward sites hosting harmful advertising or malware. Investment fraud has gone further, using deepfake videos of former finance minister and Old Mutual chairman Trevor Manuel, cloning his likeness and voice to promote fake investment schemes on social media. In May 2025, xAI’s Grok began inserting unsolicited references to “white genocide” in South Africa into unrelated conversations on X, distorting public discourse for users who had no reason to expect it.

Taken together, the incidents show that AI-enabled harm does not respect the boundaries between public institutions, professional services, and private citizens. The caution that has served South Africans well so far is a cultural habit, not a structural safeguard. The question now is whether institutional protections and public awareness can be built fast enough to replace it as connectivity grows and the technology accelerates.