South Africa’s Gambling Rules Lag as AI Expands the Market

What changed

Artificial intelligence is widening gambling’s reach in South Africa while the country’s gambling laws remain rooted in a bricks-and-mortar world. The technology can tailor advertising, games and services with remarkable precision, but South Africa has no AI-specific law governing those uses.

Why it matters

The same system that spots risky gambling can also make gambling harder to resist.

Algorithms can study player behaviour and spending patterns, then tailor content to keep people engaged and reduce churn. That may improve the experience for some players. It may also sharpen the tools that encourage harmful behaviour.

There is a safer use. Researchers working with France’s national gambling authority in 2015 and 2016 built machine-learning models that classified online players by risk using demographic and behavioural data. Those models showed how tracking could support targeted, personalised interventions.

But AI is not a neutral referee. It can reproduce bias, make mistakes, “hallucinate” information and change as its data shifts. In simulated gambling environments, large language models repeatedly made irrational, high-risk bets, sometimes reflecting the gambler’s fallacy they had learned from human data.

South Africa already has rules covering data protection, consumer protection, competition, copyright and cyber law. Those safeguards matter, but they were not designed to answer a basic new question: what level of testing, oversight and transparency should apply when an algorithm influences a player’s decisions?

Malta offers one possible direction. On 18 September 2026, its gaming authority launched a voluntary AI Gaming Charter with a risk-based approach. Higher-impact systems face greater expectations around documentation, testing, governance, transparency, human oversight and monitoring.

That is the useful lesson for South Africa: the choice is not simply between banning AI and letting operators experiment freely. A workable framework could allow predictive tools that help identify harm while demanding much stronger scrutiny of systems designed to personalise the gambling experience. Until that distinction is made, the technology will move faster than the rules meant to contain it.