The FCSA Industry Conference 2026 placed artificial intelligence in finance firmly at the centre of industry transformation, with a high-level panel featuring Ayanda Ngcebetsha, Partner Development Director for Global System Integrators at Microsoft; Dr Lavina Ramkissoon, Co-Chair of the African Union on Science, Research and Innovation; Nolwazi Hlophe, Senior Fintech Specialist at the Financial Sector Conduct Authority; Darren Franks, Chief Executive Officer of the Fintech Association of South Africa; and Fatos Koc, Head of the Financial Markets Unit at the Organisation for Economic Co-operation and Development. The discussion focused on AI adoption in financial services, the role of regulation, and the global frameworks shaping responsible innovation.
A key anchor of the conversation was the OECD AI Principles, first adopted in 2019 and updated in 2024, which continue to serve as a global benchmark for trustworthy AI in finance. These principles establish a foundation for responsible AI by emphasising transparency, accountability, fairness, robustness, and human-centric values. As financial institutions accelerate AI deployment, these guidelines are becoming critical in ensuring that innovation does not outpace governance.
Artificial intelligence is rapidly transforming the financial services sector in South Africa and globally. Financial institutions are using AI to reduce operational costs, improve productivity, and enhance efficiency across the value chain. From automating customer support and generating code to delivering advanced analytics, strengthening fraud detection, and improving anti-money laundering processes, AI is now embedded in real-world financial operations. It is also being used for marketing compliance and data-driven decision-making, highlighting its role as both an operational and strategic tool.
South Africa’s fintech ecosystem reflects this momentum. Around 41 percent of fintech companies currently have live AI production use cases, while 32 percent are scaling AI solutions more broadly. An additional 14 percent have fully embedded AI into their products. However, despite this strong adoption curve, governance frameworks remain underdeveloped. While 86 percent of organisations believe AI will drive business growth, only 5 percent report having mature AI governance structures in place. Approximately 20 percent of companies have no formal AI governance, 23 percent rely on ad hoc decision-making, and 27 percent lack consistent policies, creating a significant risk gap in the market.
The conference highlighted the urgent need to balance innovation with effective supervision. AI introduces new complexities into financial systems, including speed, opacity, autonomy, and scale. These characteristics challenge traditional regulatory frameworks, requiring a shift toward more adaptive and collaborative supervisory models. Regulators are increasingly focusing on strengthening their understanding of AI deployment and operational contexts while enhancing their ability to identify and manage emerging risks.
Public-private cooperation emerged as a critical solution. Sustained collaboration between regulators and industry players is essential to align supervisory expectations with real-world implementation. This includes initiatives such as model testing, regulatory experimentation, and shared learning environments that help bridge knowledge gaps and build trust between stakeholders.
Global examples demonstrate how this approach is being implemented. The UK Financial Conduct Authority’s AI Live Testing initiative enables real-time model testing in controlled environments, supporting validation and fostering collaboration. Supercharged sandboxes are also gaining traction, allowing multiple stakeholders to test AI systems using advanced datasets, enhanced computing capabilities, and cutting-edge AI tools. In Japan, public-private forums are facilitating continuous dialogue between regulators and industry, promoting mutual understanding and more effective oversight.
Another major focus was the intersection of artificial intelligence and Open Finance. AI is acting as a powerful accelerator for Open Finance by enhancing interoperability and improving the utility of financial data ecosystems. At the same time, Open Finance enables AI by providing the large datasets required for model training, fine-tuning, and output generation, including hyper-personalised financial services. However, this interplay also introduces heightened risks. The combination of AI and Open Finance can amplify cybersecurity threats, increase data governance complexity, and reshape competition dynamics across the financial services value chain.
Supervising AI in finance requires translating regulation into practical oversight. This includes addressing critical issues such as explainability, transparency, auditability, robustness, and fairness. Financial institutions are facing challenges in model risk management and validation, particularly in determining acceptable levels of explainability and fairness within complex AI systems. Questions such as how fair is fair enough, or what level of explainability meets compliance standards, remain key areas of debate.
Governance and data management continue to be central to responsible AI adoption. The concept of “human in the loop” must be clearly defined and implemented to ensure meaningful human oversight in automated systems. At the same time, organisations must strengthen their data governance frameworks to address issues related to data quality, representation, and third-party data usage.
Risk amplification is one of the most significant concerns associated with AI in finance. AI systems can reinforce bias, introduce new cybersecurity vulnerabilities, and amplify existing risks at scale. In high-speed financial environments such as algorithmic trading, AI systems can react instantly to market signals, potentially intensifying volatility and creating systemic risks. Additional threats include fraud, impersonation, and inaccuracies resulting from AI-generated outputs, often referred to as hallucinations.
Industry data presented at the conference highlighted key risk priorities for financial institutions. These include data privacy and protection, cybersecurity, data policy, data security, data representation, and the use of third-party data. There is also growing awareness of the need to mitigate bias and ensure that AI systems do not disproportionately impact vulnerable groups.
Despite these challenges, there are clear signs of progress. Over 60 percent of organisations have adopted some form of AI policy, strategy, or regulatory framework within the past 18 months. However, scaling responsible AI requires addressing broader structural issues, including infrastructure limitations, skills shortages, and policy gaps. Current AI utilisation in South Africa remains largely focused on natural language processing, indicating that the market is still in an early but rapidly evolving stage of maturity.
Globally, regulatory approaches to AI continue to diverge. The United States is generally adopting a pro-innovation approach, while the European Union is taking a more precautionary stance. Increasingly, there is a call for a hybrid regulatory model that balances innovation with oversight. This includes adapting regulatory frameworks to match the pace of technological advancement and incorporating supervisory experimentation to create continuous feedback loops. Regulators are also shifting their focus toward the impact of AI systems rather than solely who is using them.
The overarching message from the FCSA Industry Conference 2026 is clear: artificial intelligence is already transforming the financial services sector, and its influence will continue to expand. The focus now is on responsible AI adoption at scale, ensuring that systems are transparent, fair, and accountable while still enabling innovation. In an increasingly complex and fast-moving financial environment, trust will be the defining factor in long-term success.
