In January 2026, a BAE Systems Bofors ALMA drone fitted with Scaleout Systems AI flew over a snow-covered Swedish test range, detected several objects, ranked an armored engineering vehicle as the highest-value target, and struck it autonomously.
The entire engagement—from detection through selection to neutralization—occurred without real-time human control, representing a markedly more autonomous approach than current battlefield systems. Human operators started the mission and kept a pilot in reserve, but did not intervene during the test.
The demonstration has drawn fresh attention as AI safety dominates discussions among industry and government stakeholders. An autonomous drone severed from communications is theoretically immune to jamming, yet it also becomes dangerous if it malfunctions or misidentifies a target without oversight.
BAE Systems Bofors highlighted the three-day Karlskoga Winter Demo 2026 around roughly 20 participating startups and Swedish Defense Minister Pål Jonson’s call for more entrepreneurial involvement in defense, omitting the autonomous strike from its official release. Nonetheless, the test required human-set parameters and a manual start command.
This is not the first instance of machine target selection: Israel’s Harpy loitering munition has autonomously attacked radar emitters since the 1990s. However, the proliferation of small AI models capable of end-to-end targeting on lightweight airframes intensifies regulatory pressure. On September 5, 128 states party to the Convention on Certain Conventional Weapons agreed on a non-binding text on lethal autonomous weapons, though advocates say late changes by the United States and Russia substantially weakened human-judgment provisions.
Scaleout Systems showed that detection, ranking, and targeting can run on a small airframe without a data link, within parameters established by people. Whether this capability reaches the battlefield appears increasingly likely given the Russia-Ukraine conflict’s reliance on autonomous machines. The pressing question is how such a drone makes decisions when jammed from its source and cut off from human oversight.

Also Read
- Multi‑Generation Reproductive Decline Observed in Mice Exposed to Spaceflight
- Treasury Secretary Warns AI Executives Will Be Held Accountable for Model Misdeeds
- Researchers Develop Swallowable Paper Battery to Power Temporary Internal Medical Devices
- Samsung C&T Backs Kairos Power With Up to $100 Million Investment for Google Nuclear Reactor Project

