The Battlefield AI Revolution Is Not Here Yet: The Status of Current Russian and Ukrainian AI Drone Efforts

Kateryna Stepanenko

Source Link

Russia and Ukraine are engaged in an active technological race to develop and deploy drones with artificial intelligence (AI) and machine learning (ML) capabilities. Russia and Ukraine are competing to advance these AI/ML-powered drones to automate drone interoperability, targeting, and battlefield analysis. The successful integration of AI/ML drones could enable Russian and Ukrainian forces to reduce their reliance on human drone operators and defenders, bypass electronic warfare (EW), including jamming, 

reduce human limitations in target identification, and speed up decision-making processes involved in drone warfare.[1] Russian and Ukrainian forces will seek to operate unmanned systems in multiple domains: unmanned aerial systems (UAVs), 

unmanned surface vehicles (USF), and unmanned ground systems (UGVs).[2] Neither Russia nor Ukraine has leveraged AI/ML drones on the battlefield at scale as of early June 2025.[3] Russia and Ukraine are, however, increasingly integrating ML capabilities with some limited AI adaptations into new drone variants on the path to developing fully AI/ML-powered drones.

This paper uses AI and ML to refer to different implementations and varied degrees of development complexity, although there can be considerable overlap in definitions and discussions frequently lump ML functionality into a general bucket of AI. ML capabilities can be more scalable and easier to implement into drones when these models are trained to perform predictable and specific tasks that do not require significant processing power, memory, or data clouds.[4] Some examples of specific tasks include navigation in a GPS-denied environment and terminal guidance, image and pattern recognition, homing, 

and target locking, although some of these tasks may require AI and other more advanced tools.[5] Drones with ML-powered capabilities still require general guidance and analysis from a drone operator, such as identifying a target or modifying and training the model to operate in new or complex environments, and generally require some communication with the operator.[6] ML capabilities, in other words, can enable drones to perform some pre-programmed and pre-trained tasks, but lack the autonomy and necessary reasoning skills to adapt to battlefield conditions without human intelligence and fine-tuning.[7]