How Autonomous Weapon Systems Threaten International Humanitarian Law

Autonomous weapon systems (AWS) operate in various crises today, yet no binding international law governs their use. This regulatory vacuum allows machines to make life-or-death decisions while accountability is murky, spread across an array of programmers, commanders, and manufacturers, threatening the core principles of international humanitarian law. AWS deployed in Ukraine, Gaza, and Sudan cannot reliably distinguish civilians from combatants in complex urban environments, contravening Geneva Convention Articles 48 and 51 of Additional Protocol I, which prohibit indiscriminate attacks. Dual-use technology renders traditional arms control obsolete. Unlike nuclear weapons, AWS emerge from civilian AI frameworks, making proliferation impossible to monitor. Iran's drone programme exemplifies how actors exploit commercial technology to bypass export controls. Machine-learning systems evolve unpredictably during deployment, rendering pre-deployment legal reviews meaningless. Article 36 of the Geneva Conventions requires states assess whether weapons can be used lawfully; autonomously evolving systems cannot fulfil this obligation. Distributed responsibility across commanders, operators, programmers, and manufacturers creates an accountability void. When AWS commit unlawful acts, international humanitarian law frameworks cannot assign responsibility, as no single actor sufficiently understands or controls the system’s decision-making to bear legal culpability.