Monday 28 September, 2026
A rise in fraud incidents associated with illegal investment schemes, commonly known as Ponzi schemes, was observed in one of the West African markets in MTN Group Fintech. These schemes exploited the Mobile Money network to collect funds from victims and facilitate fraudulent cash-out transactions through agent networks, which could themselves be complicit. The scale and relational complexity of the activity made it resistant to conventional transaction-level detection. Using AI-powered graph analytics capabilities, MTN Group Fintech, initiated a programme to develop a dedicated detection capability targeting this fraud typology.
Solution
MTN Group Fraud and Forensics developed an Illegal Investment Scheme detection feature, leveraging Graph Analytics to map and analyse the relationships between participants in suspected fraudulent networks including agent network. Rather than examining individual transactions in isolation, the solution modelled the connections between victims, agents, and transactional counterparties, enabling the identification of structured patterns characteristic of scheme-based fraud. This approach surfaced agents linked to fraudulent cash-out activity that would not have been detectable through conventional transaction monitoring alone. Where active investor victims were identified within a scheme network, a proactive intervention process was initiated to protect affected customers before further financial harm occurred. As automated blocking is not yet supported, protective action currently depends on manual intervention: the investigation team suspends implicated Mobile Station International Subscriber Directory Numbers (MSISDNs) or accounts and escalates confirmed cases to local law enforcement upon identifying a victim. The solution was developed and validated in a West African market, where it has continued to operate in this validated form, with production deployment and an enhanced alert management interface planned for July 2026 to enable a faster, more automated response.
Impact
Over the past few months, the detection feature identified and disrupted multiple multi-agent investment schemes, protecting 2,180 unique customers across 4,803 associated fraudulent transactions and mitigating an estimated 36 million West African CFA francs (XOF) in exposure. Of the cases the model flagged, all were confirmed on investigation as linked to illegal investment scheme activity, a one hundred per cent confirmed fraud hit rate, with all flagged cases validated as linked to illegal investment scheme activity. The Graph Analytics approach successfully identified agents involved in fraudulent cash-out transactions within scheme networks, demonstrating the value of relationship-based detection over transaction-level analysis alone. In one confirmed case, a victim actively investing in a fraudulent scheme was identified through the network model and protected through proactive intervention, preventing further financial loss. The solution established a new detection capability within MTN Group Fintech for a fraud typology that had previously operated below the detection threshold of existing controls, and has already delivered measurable customer protection ahead of its planned production rollout.
Additional information
Cost: Medium. The feature was developed internally, building on existing data infrastructure and extending it with Graph Analytics capabilities. Primary cost drivers included graph model development, data pipeline integration, together with the ongoing operational cost of manual case investigation and account suspension while automated blocking remains unavailable. Further investment is planned for the alert management interface enhancement accompanying the July 2026 production deployment.
Participating organisations: MTN Group Fintech – Group Fraud and Forensics
Location: Sub-Saharan Africa
