Swedish Drone AI Autonomously Selects and Strikes Target in Demo
A BAE Systems drone used onboard AI to identify and destroy a vehicle without human input during a recent test.
Key points
- A BAE Systems drone autonomously selected and struck an armored vehicle in a January 2026 test.
- The system used onboard AI from Scaleout Systems to detect and target without real-time human input.
- The demo highlights the capability of small drones to operate independently, raising regulatory concerns.
In a recent demonstration on a snow-covered Swedish test range, a small fixed-wing drone successfully identified, ranked, and struck an armored engineering vehicle. The exercise, which took place in January 2026, highlighted the growing capability of artificial intelligence to manage the entire targeting process independently.
According to reports from TechRadar, the system utilized onboard software to detect potential threats and make a kill decision without any real-time human intervention. While operators initiated the mission and set initial parameters, the drone executed the final strike autonomously, marking a significant step in autonomous warfare capabilities.
Autonomous Targeting in Winter Tests
The demonstration was part of BAE Systems Bofors' Affordable Loitering Modular Ammunition program, with artificial intelligence supplied by Uppsala-based Scaleout Systems. During the test, the drone detected multiple objects on the range and selected the armored vehicle as the highest-value target. It then navigated to the location and deployed an explosive payload, completing the mission cycle without pilot input.
Although a pilot remained on standby and a data link was available, the system operated end-to-end using local processing. This approach suggests that future munitions could function effectively even if communications are jammed, provided the initial mission parameters are correctly set by human operators.
Operational Implications for Defense
Military officials view such capabilities as both a strategic advantage and a potential risk. An autonomous drone that does not rely on a continuous backlink is less susceptible to electronic warfare interference, increasing its likelihood of mission success. However, the absence of human oversight during the final phase raises concerns about error rates and the potential for unintended strikes.
This development follows earlier autonomous systems, such as Israel's Harpy loitering munition, which has been capable of independently detecting and attacking radar emitters since the 1990s. The recent Swedish test distinguishes itself by demonstrating the integration of detection, ranking, and targeting within a small airframe using compact AI models.
Regulatory Frameworks Lag Behind Tech
The timing of this demonstration coincides with ongoing diplomatic efforts to regulate lethal autonomous weapons. In September, 128 states party to the Convention on Certain Conventional Weapons agreed on a non-binding text regarding these systems. Critics argue that the final document was diluted in the final hours of negotiations, particularly regarding provisions for human judgment and control.
As autonomous machines become more prevalent in conflicts like the war in Ukraine, the gap between technological capability and international law widens. The central question remains how such systems will be governed when they operate in contested environments without immediate human supervision.
