NPR details how a Ukrainian drone flies itself to a human-selected target

A pilot selected a moving car, waited for a target lock and let go of the controls. The drone flew toward it on its own. In an October 29 report, NPR’s Joanna Kakissis describes The Fourth Law’s unarmed demonstration in Ukraine—a close look at a weapon that separates human target selection from AI-controlled flight.

From pet cameras to battlefield guidance

The company’s founder, Yaroslav Azhnyuk, worked in Silicon Valley before turning to defense after Russia’s full-scale invasion in 2022. One of his earlier companies, Petcube, makes connected cameras that let owners watch and entertain pets remotely. He told NPR that, living in Ukraine when the invasion began, he decided to leave that work to help protect his country.

He now runs Ukraine-based defense firms The Fourth Law and Odd Systems. The Fourth Law says dozens of Ukrainian military units use its Lupynis-10-TFL-1 drone. The aircraft combines a small first-person-view drone, which a pilot steers using its camera feed, with an onboard computer that handles the final approach to a target.

The handoff in an overgrown field

NPR visited a test team in a field outside Kyiv. The pilot, identified by the call sign Professor, began the flight with a control stick. A red flash on his monitor signaled a lock on the car, which an employee was driving along a dirt road. Onboard software then kept the aircraft on course. This test drone carried no explosive; a battlefield version would.

The autonomous terminal-guidance unit takes over during the final 500 meters of flight. The test and integration chief, using the call sign Stab, said its image-recognition system uses a deep-learning neural network trained with hours of combat footage to recognize vehicles, tanks and flying objects. He also said AI drones are useful because they are hard to jam.

And in the end, it's the human that decides how to use that tool.

Yaroslav Azhnyuk, speaking to NPR

The boundary is who chooses the target

For the current drone, Stab identified human target selection as a safeguard against the aircraft acting on its own. The demonstration showed the aircraft following a person’s chosen target; it was not a demonstration of a machine independently deciding what to attack.

Azhnyuk rejects comparisons between his current drones and fictional killer robots. But he acknowledged that future drones could make their own decisions about designating targets. He presented that possibility as a future development, not a capability demonstrated in the field.

His argument for further automation is rooted in survival. NPR reports that Russia is also deploying AI-enabled drones against Ukraine. Azhnyuk argues that refusing AI decision-making for defense, while an adversary uses it offensively, risks defeat. He contrasts machine decisions made in microseconds with human control taking minutes, hours or days.