Ukrainian Startup Tests AI Drones That Steer Themselves for Final 500 Meters

Ukrainian defense firm The Fourth Law is testing artificial intelligence systems that allow drones to steer themselves during the final 500 meters of a flight. The technology removes the need for a human operator to manually guide the aircraft through the most dangerous part of its mission. This development marks a shift in how small drones are used on the battlefield in Ukraine.

Yaroslav Azhnyuk, a 37-year-old computer scientist and founder of The Fourth Law, describes the technology as a major force multiplier. He states that the integration of AI can make units 10,000 times more effective. Azhnyuk argues that the speed of machine decision-making creates a gap that human operators cannot close. He notes that machines decide in microseconds, while human control often takes minutes or hours.

The specific model, named Lupynis-10-TFL-1, is already in use by dozens of Ukrainian military units. The system pairs a standard first-person-view drone with an autonomous terminal guidance unit. An onboard computer takes control of the drone once it is within 0.3 miles of its target. Azhnyuk emphasizes that the technology relies on basic image recognition tools rather than the complex artificial intelligence found in self-driving cars or large data centers.

Azhnyuk previously spent years in Silicon Valley and founded Petcube, a company that sold one million AI-powered pet cameras. He explains that the image recognition used in the drones is directly related to the technology used in those pet cameras. The difference is that the drones now deliver explosives instead of treats. The company names its weapons after Ukrainian intellectuals and artists who were persecuted by Russia, with the current model honoring poet and activist Anatoliy Lupynis.

Testing takes place daily in fields outside Kyiv. A team led by the head of test and integration, who uses the call sign Stab, demonstrated the system for reporters. Stab explained that a microcomputer inside the drone runs a deep-learning neural network trained on hours of combat footage. The system is designed to recognize tanks, light-armored vehicles, and flying objects. In the demonstration, a pilot with the call sign Professor guided the drone to a target, which was an old car driven by an employee along a dirt road.

During the test, a red flash indicated that the drone had locked onto the vehicle. Professor released the control stick, and the drone flew itself to the car. The test drone carried no explosive. Stab noted that the drones are difficult to jam and include safeguards to prevent them from going rogue. Humans still select the targets, a point Azhnyuk stresses to counter the idea that these are killer robots. He calls them calculators.

Russia is also fielding AI-enabled drones against Ukraine, creating a high-tech arms race. President Volodymyr Zelenskyy has warned that AI could be making battlefield decisions as soon as next year. Azhnyuk says Ukraine must adopt the technology for defense to avoid losing the conflict. He argues that if adversaries use AI for offensive decisions and Ukraine does not, the outcome is decided by speed. He concludes that if Ukraine loses, there will be no time to debate the ethics of the technology, making survival the primary driver for adoption.