June 2026 – Central banks are among the most data-intensive public institutions, holding vast datasets that underpin financial supervision, payments, and policy. As they adopt AI-enabled analytics and supervisory technology (SupTech), their ability to draw inferences from data expands dramatically. This shift raises profound governance questions, especially in emerging markets and developing economies (EMDEs) where oversight frameworks are still evolving.
AI-driven inference allows central banks to connect information across datasets, revealing insights about legal entities and individuals that go beyond the original purpose of data collection. While this enhances risk detection and supervisory capacity, it also creates new vulnerabilities. Without strong governance, these tools risk undermining financial inclusion, institutional trust, and the legitimacy of digital financial reform.
Central banks often acknowledge processing personal data, but few clearly explain how AI-derived profiling and inference are governed. This opacity is particularly risky in developing countries, where legal protections may lag behind technological advances. For example, Kenya’s courts halted the Huduma Namba Digital ID scheme until proper safeguards were introduced, illustrating how digital infrastructure can outpace the law.
Even institution-level datasets can carry personal implications. The European Central Bank’s AnaCredit dataset, which records loans to legal entities, was found to potentially identify individuals when names and addresses overlap. As AI analytics grow more sophisticated, the risk of “inferred identities” becomes more acute.
The stakes are especially high in digital payments and central bank digital currencies (CBDCs). Systems like Brazil’s PIX fast payment platform give central banks visibility into all transactions. AI can then derive insights beyond the original purpose, such as profiling individuals for AML/CTF compliance. While legally justified, such broad bases can enable extensive analytics that disproportionately affect vulnerable populations.
In Africa, AI-based credit scoring is common due to limited formal credit histories. But when AI flags trigger scrutiny without human review, errors can exclude individuals from financial systems. Weak governance around inference risks deterring participation, undermining the very inclusion goals these systems aim to achieve.
To ensure AI strengthens rather than erodes trust, three governance priorities stand out:
- Institutional transparency: Central banks should maintain inventories of AI tools, detailing their purpose, data categories, and whether they generate inferences about individuals.
- Human oversight: Risk assessments, independent reviews, and contestability mechanisms are essential to prevent unchecked AI-driven decisions.
- Data separation and controls: Clear rules for data sharing, privacy-enhancing technologies, and access controls can prevent misuse and political exploitation.
AI governance for central banks cannot be left solely to data protection laws or emerging AI frameworks. It must be embedded in the rules governing payment systems, supervisory practices, and institutional accountability. With transparency, oversight, and strong controls, central banks in developing economies can harness AI responsibly, fostering financial inclusion while maintaining public confidence in digital reform.
