Ukraine and NATO develop AI system to predict Russian drone strikes

Ukraine and NATO have developed an artificial intelligence system designed to predict where and when Russian drone strikes may occur on Ukrainian territory. The model, labelled Model 13, analyses data from past attacks together with weather conditions and battlefield indicators to identify locations and time windows with higher likelihood of new strikes.

How Model 13 and its companion tools are described

The Ministry of Defense of Ukraine said Model 13 was tested using real operational data and that it will be integrated into military situational-awareness tools. Officials released the two models on 13 October, but they did not publish independent accuracy figures, the forecast horizon (how far in advance it can predict an attack), or detailed assessments of operational impact.

Alongside the predictive model, developers created a second system, Model 32, intended to assist artillery planning. That system takes into account available equipment, ammunition, potential firing positions and other constraints to propose feasible firing plans, eliminate impractical options and evaluate the remainder. A prototype of Model 32 has been presented to Ukrainian authorities for trial use.

The projects involve the Ukrainian armed forces, NATO’s Allied Command Transformation (ACT) and the Joint Analysis, Training and Education Centre (JATEC) established between NATO and Ukraine. Representatives described the effort as a collaboration to adapt data-driven tools for battlefield decision support.

Separately, on 8 October it was reported that the Dutch company Intelic is developing software called Nexus to interlink radar and sensor systems used to detect and counter drones in Ukraine; that work is part of a different project and is not the same programme that produced Model 13 and Model 32.

Talia Beach, a technical adviser at ACT, has highlighted other military uses for AI. In a June 2026 piece published on the command’s website she explained that a model originally developed to monitor populations of rare birds had been adapted to track military movements. She summed up the change this way: "The science remained the same. The context changed." ACT staff emphasised that this particular adaptation is a separate effort from the predictive strike model.

No independent results have yet been published to confirm how accurately Model 13 can predict Russian drone strikes. Although Ukrainian authorities say the system is ready for integration, it is not confirmed whether it is already in operational use; its practical value for the armed forces therefore remains uncertain.

Photo: press material from the event