Amy McAuliffe
A battlefield dominated by fully autonomous drones capable of selecting and engaging targets without human intervention sounds like a dystopian sci-fi plot. While that scenario has not fully arrived, the Russia-Ukraine conflict has accelerated modern warfare down that road.
Both nations have deployed semi-autonomous drones in combat and are rapidly racing towards fully autonomous unmanned aerial vehicles (UAVs).
Given current production rates, a battlefield with thousands of fully autonomous drones operating simultaneously is plausible within a few years. Yet, the technology underpinning autonomous warfare risks outpacing legal and operational frameworks designed to maintain human control over weapons.
As a weapons technology expert, I fear battlefield imperatives will push aside vital human targeting decisions regarding proportional force, avoiding non-combatant casualties, and limiting collateral damage.
Both sides have prioritised drone production since February 2022. Unable to mass-produce one-way attack drones initially, Russia sought Iranian assistance before building domestic capacity. Ukraine expanded its nascent drone sector through private defence firms and rapid innovation.
Both militaries have added autonomous features to existing systems while developing new platforms. Specifically, Russia and Ukraine deploy drones using AI to recognise targets, navigate, and self-guide to objectives.
In November 2025, a Ukrainian first-person-view drone used an AI guidance system to strike a Russian tank after its communications link failed. A human selected the target, but the drone guided the final attack phase independently.
Reliable numbers remain unavailable, but reports suggest these systems represent a small percentage of the millions of drones deployed by both sides to date.
At the advanced end, Russia fielded a fully autonomous UAV in early 2025. Ukrainian analysis of recovered Russian V2U drones revealed an Nvidia computer running AI decision-making software. Late 2025 variants lacked external communications hardware, indicating zero operator control.
The V2U enables autonomous target selection and engagement post-launch. Human operators traditionally select targets for moral and ethical accountability, highlighting the grave shift marked by autonomous execution.
Onboard AI uses computer vision to analyse sensor images, recognise objects, track targets, navigate, and select strike points.
In July, Russia deployed an AI drone that demonstrated partial autonomy in Zaporizhzhia, killing three civilians at a gas station. Human operators picked the general target area, but software selected the final strike point.
Russia is also integrating similar technology into its Geran series UAVs, derived from Iran’s Shahed-136. Some Gerans feature seeker guidance packages that allow onboard computers to direct final flight phases after human target selection.
Ukraine has focused on modular AI packages — combining chips and cameras — for installation across various platforms. These modules use onboard AI to identify targets and guide drones during final flight stages.
One package guides drones through the final 500 metres. Manufacturers advertise advanced variants promising complete operational autonomy.
Ukraine claims further autonomous successes. Former Defence Minister Mykhailo Fedorov noted that Ukraine deployed a drone capable of autonomous target selection without pilot communications during strikes in Crimea, though it remains unmass-produced.
Ukraine also reportedly used autonomous quadcopters near Bakhmut. However, reports that these systems were pre-programmed raise questions over whether targets were genuinely selected independently.
Autonomous drones have already impacted the battlefield, even if the most advanced systems are not yet deployed in massive numbers.
The tactical benefits are clear. Autonomous systems bypass electronic jamming, maintain operation without radio links, and enable deep strikes.
Final-phase AI guidance improves accuracy against moving targets. Crucially, scaling fully autonomous drones reduces the operator manpower needed for massive drone fleets.
Yet, ethical concerns remain severe. Rapid deployment risks outstripping human control. Ukrainian Defence Ministry guidelines require humans in the loop to verify targets before final interception.
Conversely, Russia defines human control as the ability to set parameters or deactivate systems, without requiring human verification prior to individual strikes.
As worrying as that sounds now, the next step in AI drone warfare is even more concerning: Swarming drones that can hunt in packs of a few to thousands and autonomously alter their behaviour based on communicating with one another present a frightening challenge to maintaining human control.