Artificial intelligence is moving from the margins of military decision-making towards the point at which lethal force is applied.
In loitering munitions, that transition is especially consequential. These weapons already combine surveillance, target acquisition and strike within one platform. AI can now support several distinct functions: object classification, electronic-emitter detection, resilient navigation, collaborative search, target prioritization and the in-flight allocation of weapons across a networked salvo.
The operational advantage is clear. These capabilities can expand the area searched, accelerate attacks against mobile targets, reduce operator workload and sustain missions through degraded communications. Against an integrated air-defense system, cooperating munitions could locate emitters, exchange detections and coordinate attacks before the enemy relocates or responds.
The policy risk doesn’t arise from every autonomous function equally. Navigating around a threat is fundamentally different from determining that an object satisfies the conditions for attack. The critical boundary is crossed when a system moves from supporting a human decision to constructing the engagement decision itself (The kill order).
