Nigeria’s Central Bank has introduced a new artificial intelligence-based fraud detection framework designed to identify suspicious financial activity in real time and strengthen oversight of the country’s digital payments ecosystem.

The initiative marks a shift away from traditional rule-based monitoring systems, which typically review transactions after completion. Under the new approach, machine learning tools will analyse payment activity as it occurs, enabling potentially fraudulent transfers to be flagged or interrupted before funds are fully processed. The reform is aligned with the Nigeria Payments System Vision 2028, which sets out a roadmap for modernising financial infrastructure and improving regulatory standards.

Real-time fraud detection shift

A core component of the programme is the transition to predictive monitoring powered by artificial intelligence and machine learning. Instead of relying on post-transaction reviews, the system is designed to assess transactions in real time using behavioural signals such as spending patterns, device identifiers, location data, and historical activity. This approach is intended to improve detection of sophisticated threats, including phishing attacks and authorised push payment fraud, while also reducing the frequency of false alerts generated by older systems.



Implementation and regulatory rollout

The Central Bank plans to establish a National RegTech and SupTech Lab to support the deployment of automated compliance tools and live risk monitoring dashboards. The facility will also act as a coordination hub for developing supervisory technologies across the financial sector.

Under the implementation roadmap, financial institutions, including banks and fintech operators, will be required to integrate AI-driven compliance systems and real-time reporting tools within a three-year period. The Central Bank has set a target of reducing fraud-related losses by up to 70 per cent by 2028, with success dependent on coordinated adoption across the financial ecosystem.