OSPRE AI Threat Detection Achieves 89% Predictive Accuracy in Field Trials

The latest upgrade to OSPRE's artificial intelligence threat detection module has achieved 89% predictive accuracy in an independent field validation exercise conducted across 15 states over a six-month period. This result significantly exceeds the 70% benchmark established by Nigeria's National Security Framework for AI-assisted early warning systems.

The upgraded module, developed in partnership with the African Centre for Technology Studies (ACTS) and two Nigerian university computer science departments, combines natural language processing of open-source intelligence with structured data from OSPRE's community reporting network to generate conflict risk scores at the LGA level.

The system processes over 40,000 data points daily — including social media signals, community reports, security incident logs, market price data, and weather information — to generate daily risk scores for all 774 LGAs. When risk scores exceed defined thresholds, the system automatically escalates alerts to relevant state and federal response units.

Field testing across the North-East, North-West, and North-Central zones demonstrated that the module correctly predicted 89% of significant security incidents at least 48 hours in advance. False positive rates were maintained below 12%, which operational teams described as acceptable and manageable.

OSPRE plans to expand the module's integration with the National Emergency Management Agency (NEMA) and the Office of the National Security Adviser (ONSA) during the second quarter of 2026.