The drive to dominate the battlefield through artificial intelligence is producing lethal errors as militaries embed mistake-prone algorithms into the kill chain, according to a joint report by the AI Now Institute and the conflict-monitoring group Airwars. The findings raise hard questions about prioritizing rapid deployment over the safety of American personnel and the protection of civilians.
The Automated Kill Chain
The report describes a system of interconnected AI modules, each introducing distinct flaws that compound as data passes from one algorithm to the next. Heidy Khlaaf, chief AI scientist at AI Now, stated that these chained systems build upon each other's errors, with the final output often diverging significantly from ground truth. The risk is not merely theoretical. The technology is now operational, feeding targeting recommendations to forces engaged in Ukraine and across active theaters in the Middle East.
Humans are prone to just accepting recommendations of AI algorithms without corroborating that evidence.
This reliance on machine output creates a dangerous dynamic where operators rubber-stamp computational suggestions rather than exercising rigorous human judgment. For American strategic interests, which depend on a stable global order and the avoidance of unwinnable entanglements, the embrace of unaccountable targeting systems represents a step toward perpetual, error-driven conflict.
Cost in Blood and Treasure
The report documents an uptick in civilian casualties directly tied to algorithmic failures. Malfunctioning pattern recognition has mistaken humanitarian convoys for military formations. Database mismatches have designated neutral parties as hostile. These are not system bugs; they are the predictable result of technology being pushed into service before reliability thresholds are met. The defense contractors profiting from these systems—an industry that spent over $115 million on lobbying in 2023 alone—have a vested interest in downplaying the operational hazards.
For the American worker, already shouldering the trillion-dollar price of the national security state, the expansion of unreliable AI warfare promises only a heavier burden and a less secure homeland. Wars driven by algorithmic accident are not wars of national necessity; they are acts of machines gone awry. The U.S. must recalibrate its approach to warfighting technology, ensuring that ironclad verification standards and strict human control are non-negotiable prerequisites, not afterthoughts.