When a country’s financial regulators release their first-ever report on artificial intelligence, you pay attention. It isn’t just another dry government document; it’s a snapshot of the future being built, one algorithm at a time. The recent joint report from South Africa’s Financial Sector Conduct Authority (FSCA) and Prudential Authority (PA) does just that, and it’s a fascinating case study for anyone watching the space of emerging markets AI adoption. It reveals a landscape of stark contrasts, where ambition often runs far ahead of execution.
The report, based on a comprehensive survey of over 2,100 financial institutions, pulls back the curtain on who is actually using AI versus who is just talking about it. The headline figure? A modest 10.6% of the sector has AI systems in production. But dig a little deeper, and the real story emerges. This isn’t a single, uniform market; it’s a tale of two very different worlds.
A Two-Speed Race in AI Adoption
On one side, you have the big banks and payment providers. According to the report cited by iAfrica, a striking 52% of banks and 50% of payment providers are already using AI. They’re the ones with the deep pockets and the scale to justify significant investment. In fact, 45% of banks using AI are planning to spend over R30 million on it this year alone. They’re deploying it for the usual suspects: operational efficiency, fraud detection, and risk management. It’s practical, it’s sensible, and it delivers a clear return on investment.
On the other side, you have almost everyone else. Insurers and lenders are lagging significantly, with a mere 8% adoption rate. For them, the investment story is entirely different. A hefty 62% of investment providers and 41% of insurers are expecting to spend less than R1 million. It’s a classic technology adoption curve playing out in real-time, where the initial phase is dominated by a few well-resourced pioneers while the majority watch from the sidelines, waiting for the costs to come down and the path to become clearer. So, the question isn’t if they will adopt AI, but when and how.
The Playground for Progress: Regulatory Sandboxes
How do you encourage the cautious majority to start innovating without letting them blow up the financial system? This is where the concept of regulatory sandboxes comes in. Think of it as a controlled test environment, a supervised playground for new financial products. It’s like giving a budding chef a corner of a professional kitchen to experiment with a new recipe under the watchful eye of the head chef. They can try bold ideas, see what works, and if something goes wrong, the mess is contained.
For emerging markets AI adoption, this is absolutely crucial. Sandboxes allow fintech startups and even established players to test new AI-driven models—perhaps for micro-lending or personalised insurance—on a small scale. This fosters innovation in areas like financial inclusion tech, helping to create services for those previously unbanked or under-served, all while regulators gather the data they need to build smart, effective rules for the future. Without this safe space to experiment, innovation would likely stall, throttled by the fear of falling foul of regulations that were written long before AI was a boardroom topic.
The Elephant in the Room: Building the Human Infrastructure
You can have the best algorithms and the most supportive regulators, but none of it matters if you don’t have the people to build, manage, and oversee the technology. The report hints at a significant challenge in AI capacity building. It’s one thing for a bank to pour millions into a new system; it’s another thing entirely to find the data scientists, AI ethicists, and governance experts needed to run it responsibly.
This skills gap is the single biggest brake on AI adoption in many parts of the world. It’s not just about teaching people to code. It’s about developing a deep bench of talent with expertise in data privacy, cybersecurity, and the ethical implications of automated decision-making. Right now, there’s a huge disparity between the grand ambitions for AI and the on-the-ground capacity to deliver it. How can institutions in emerging markets bridge this gap without simply importing expensive talent from abroad? The answer lies in long-term investment in local education and training programmes.
Navigating the Minefield of Governance and Privacy
As institutions wade deeper into AI, they’re bumping up against some thorny issues. The report highlights data privacy as a top concern, which is no surprise given South Africa’s robust Protection of Personal Information Act (POPIA). Every piece of customer data fed into an AI model is a potential privacy liability. How do you ensure compliance when the AI itself is a “black box,” its decision-making process opaque even to its creators?
This leads directly to the problem of governance. The survey found that while many firms have some form of oversight, these systems are uneven and lack the rigour needed for high-stakes AI. Regulations like the Financial Advisory and Intermediary Services Act (FAIS) provide a framework, but they weren’t designed for a world where financial advice might come from a machine.
Institutions are crying out for clearer regulatory expectations. They want to know what “good” looks like when it comes to AI governance. This includes everything from ensuring model fairness and mitigating bias to establishing clear lines of accountability when an algorithm gets it wrong. Without this clarity, companies are navigating a minefield blindfolded.
The Path to Responsible Implementation
So, what is the way forward? The future of emerging markets AI adoption hinges on a dual strategy. First, institutions need to move beyond a tick-box approach to compliance. Adhering to POPIA and FAIS is the starting point, not the finish line. True responsibility means embedding ethical principles into the entire AI lifecycle, from design and data collection to deployment and monitoring. Tools that increase transparency, like SHAP and LIME, are a step in the right direction, helping to explain why a model made a particular decision.
Second, the ecosystem as a whole must mature. This means more government and industry collaboration on AI capacity building. It means expanding regulatory sandboxes to fast-track promising innovations, especially in financial inclusion tech. And most importantly, it requires regulators to provide the clear, forward-looking guidance the industry needs. The upcoming Conduct of Financial Institutions (COFI) Bill in South Africa will be a critical piece of this puzzle.
The South African financial sector’s journey with AI is more than just a local story; it’s a blueprint filled with lessons for other emerging economies. It shows that while technology opens the door, it’s the careful construction of skills, governance, and smart regulation that paves the road to sustainable and inclusive progress.
What do you think is the single biggest barrier to AI adoption in your country’s financial sector—is it the cost, the skills gap, or the regulatory uncertainty?
