Ukrainian drone manufacturer Aero Center disclosed on June 12, 2026 that it conducted a one-time battlefield test of quadcopters operating in full autonomous combat mode two years ago. The drones searched for and attacked targets on the front line without any human involvement. It is the first producer-confirmed use in combat of a system meeting the definition of a lethal autonomous weapon system (LAWS).
Key takeaways
- Aero Center CEO Alexander Kokhanovskyy confirmed the test at a press event at the Ukrainian Embassy in London
- Quadcopter drones operated in "Terminator mode" — preprogrammed flight to the front zone, then automatic target search and attack on anything in the designated area
- After the test, drone pilots sent to assess the area found "a couple" of dead Russian soldiers
- The test was a one-time event — Ukraine's military routinely uses semi-autonomous systems with a human in the loop
- CSIS analysts assess that full autonomy remains rare, but partial autonomy is spreading rapidly
"Terminator mode"
Alexander Kokhanovskyy, CEO of Aero Center, disclosed the test at a press event organized by the Ukrainian Embassy in London, in an interview with New Scientist.
The drones were preprogrammed: they flew into a designated front-line area, then activated an AI mode that independently identified and attacked everything within range. There was no video feed to the operator. There was no way to abort the mission remotely. After the mission, human drone crews sent to inspect the area found "a couple" of dead Russian soldiers — which Kokhanovskyy cited as proof of effectiveness.
The absence of mission footage means the test cannot be independently verified. It is unclear what the drones actually targeted, how they identified military targets, whether the strikes were precise or accidental, and whether there was any risk of hitting friendly forces.
How Ukraine is driving autonomy
Ukraine wages drone warfare at an unprecedented scale: more than 5,000 strikes per month on Russian targets at ranges exceeding 20 km, according to the Ministry of Defense. That level of activity would be impossible without AI.
Russian electronic warfare systems jam operator communication links. Russian GPS jamming disrupts guidance. Ukraine's answer is small AI models trained on small datasets — capable of running on cheap onboard chips. The result: the success rate of Ukrainian drone strikes climbed from 10-20% to 70-80%, according to a report by Kateryna Bondar for the Center for Strategic and International Studies (CSIS) in Washington.
The same philosophy of small, specialized AI modules is applied across all platform types: FPV drones, quadcopter bombers, long-range strike drones, and ground robots. AI modules can be installed as a standalone hardware unit — and that is exactly how the "Terminator mode" worked in the Aero Center system.
The Russian side
Russia is accelerating too. Shahed-136/Geran-2 drones — of Iranian and increasingly Russian manufacture — are fired by the hundreds at Ukrainian cities every night. Standard Shaheds follow a preprogrammed flight path with no autonomous decision-making. But upgraded Geran-2 variants have been equipped with smuggled NVIDIA Jetson Orin computers with computer vision modules — capable of autonomous target recognition and in-flight retargeting.
In other words: both sides of the conflict are independently moving in the same direction. Autonomous combat capability is not a question of "if," but "when and how deep."
Law, ethics, and the LAWS definition
There is no universally agreed definition of a "lethal autonomous weapon system." The UN — through its Office for Disarmament Affairs — has been working on norms for years without consensus.
The US Department of Defense defines LAWS as systems that "once activated, can select and engage targets without further intervention by a human operator." The requirement for "meaningful human control" is the key dividing line in ethical debates. The Aero Center test crosses that line.
A Ukrainian military commander quoted by New Scientist said the military routinely uses only semi-autonomous systems that preserve human control and respect international humanitarian law. Kokhanovskyy himself noted the test was a one-time event. The risk of friendly fire and civilian casualties — from a system attacking "anything in the designated area" — is real under full autonomy.
Why this matters
The Aero Center test is likely the first producer-confirmed use of a drone meeting LAWS criteria in actual combat. Earlier there were speculation and reports suggesting such systems had been tested or used by Turkey (Kargu-2) and Libya (Harop) — but without producer confirmation.
This changes the debate. Until now, the discussion about LAWS was academic and political: "should we even allow them?" Now it is concrete and military: "how do we constrain them, since they already exist?" Powers that have blocked a UN LAWS treaty for years now face a dual argument — that norms are urgently needed, and that regulation will slow their own development.
For the robotics and AI industry, the context is different: military autonomy is driving technology transfer. AI modules developed for combat — small, efficient, interference-resistant — will migrate into civilian systems. This is already happening with visual navigation, object recognition, and route planning.
What's next
- The UN is set to resume negotiations on a LAWS treaty — the Aero Center disclosure will likely be cited by both sides at the next session of the Disarmament Committee.
- Ukraine has announced plans to scale drone production to 10 million units per year by 2027 — increased partial autonomy is inevitable at that scale.
- NVIDIA Jetson Orin (identified in upgraded Shahed drones) is subject to US export controls, which did not stop Russia — a signal to dual-use technology regulators.
