Small AI models let drones autonomously identify and attack battlefield targets (arstechnica.com)

A NATO-backed startup, Scaleout Systems, is making strides in defense technology by deploying small AI models that enable drones to autonomously identify and engage battlefield targets. Originally founded in 2018 by Uppsala University researchers, the company shifted its focus toward military applications following the onset of the Ukraine conflict. CEO Andreas Hellander emphasized the need for operationalizing edge and sensor data to provide a strategic advantage for NATO allies. By using leaner machine learning models suited for the unique hardware constraints of drones, Scaleout aims to enhance surveillance and attack capabilities. The significance of this development lies in its potential to revolutionize battlefield operations by integrating advanced AI directly into the frontline equipment. Rather than relying on large AI models, Scaleout’s approach allows for real-time computer vision processing even on smaller, embedded systems found in drones and command posts. Selected for NATO's Defence Innovator Accelerator (DIANA) Challenge Program in 2025, the company is actively working on adapting machine learning for edge computing environments, a critical step in enhancing military readiness and operational efficiency in modern warfare scenarios.