AI Weapons: Faster War, No Blame
ICC Warrants for Shahed-136 Drones
In March-June 2024, the International Criminal Court (ICC) issued arrest warrants against Russian high command for war crimes involving Shahed-136 drones autonomously destroying Ukrainian power plants. The Lieber Institute for Law and Land Warfare at West Point, the International Committee of the Red Cross, and the UN Office for Disarmament Affairs explain that an accountability gap arises because no universally agreed-upon precise threshold for "meaningful human control" (MHC) exists, though it generally requires humans to retain authority over critical moments of force application. Frontiers in Big Data, the Lieber Institute for Law and Land Warfare at West Point, and the Chicago Journal of International Law find that when machines select and attack targets independently, the opacity of algorithmic decision-making makes it highly problematic to prove intent (mens rea) or negligence among developers, commanders, or operators. This diffusion of responsibility spreads across states, military hierarchies, and private technology contractors, as Frontiers in Big Data and the European Journal of International Security found.
Distributed Responsibility Model Emerges
Frontiers in Big Data documents that scholars also propose a "distributed responsibility" model, which disaggregates accountability across the entire defense ecosystem, including civilian leadership, technology contractors, and military commanders. Command responsibility frameworks can still hold commanders liable if they knew or should have known a system under their control was likely to commit violations but failed to take reasonable preventive steps, as the Lieber Institute for Law and Land Warfare at West Point, Temple University Beasley School of Law, and the UN Office for Disarmament Affairs explain. Despite these challenges, ultimate accountability does not transfer entirely to the machine.
US 2023 DoD Directive 3000.09 Rejects MHC
Congress.gov and the United Nations Office of Legal Affairs report that the United States formally rejects MHC as ambiguous, instead focusing on "appropriate levels of human judgment" and replacing "human operator" with "operator" in its 2023 DoD Directive 3000.09, potentially broadening the distance between human action and weapon deployment. In contrast, the European Union maintains a shared institutional stance requiring MHC, with Germany legally mandating it in both design and use phases, and France prohibiting fully autonomous lethal weapons, Congress.gov documents. Russia asserts that existing international law is sufficient and opposes new limitations, the Lieber Institute for Law and Land Warfare at West Point states. These divergent national doctrines on meaningful human control (MHC) further complicate liability attribution in multinational coalition operations. These differences create legal friction and ambiguity within alliances, making it difficult to attribute mens rea for individual criminal liability and potentially undermining command responsibility principles in joint missions, as the Chicago Journal of International Law, the UN Office for Disarmament Affairs, the Georgetown Security Studies Review, and the United Nations Office of Legal Affairs highlight.
India-Pakistan Last Crisis in May 2025
The Asia-Pacific Leadership Network reports that the India-Pakistan Last Crisis in May 2025 showed how quickly crises can intensify, with AI-enabled systems shortening the decision fuse. arXiv, the UN Office for Disarmament Affairs, and the Georgetown Security Studies Review argue that the delegation of lethal decision-making to autonomous weapons systems significantly alters strategic stability by lowering domestic political casualty thresholds, thereby increasing conflict frequency and intensity. By removing operators from the immediate physical casualty chain and compressing decision timelines, states face fewer domestic political barriers to initiating military action, arXiv found. Routine cross-border drone incidents lower retaliation costs, while attribution ambiguity creates risks of deadly misunderstandings and automated air-defense escalations that outpace political leaders' ability to intervene, the Asia-Pacific Leadership Network highlights. The Asia-Pacific Leadership Network also documented Exercise Golden Eagle, conducted by the Pakistan Air Force in February 2026, which focused on AI-enabled, net-centric operations, integrating cyber, space, and electromagnetic spectrum capabilities with manned-unmanned teaming to demonstrate AI-driven intelligence, surveillance, and reconnaissance (ISR)-to-strike loops in contested environments.
Rapid, Algorithmically-Driven Escalations
The evidence makes clear that the unchecked proliferation and deployment of autonomous weapons systems will inevitably lead to more frequent and intense conflicts, while simultaneously making it harder to hold anyone truly responsible for the resulting devastation.
Comments ()