As governments and military alliances increasingly adopt AI-enabled decision-support systems, questions about civilian protection and compliance with international humanitarian law (IHL) are becoming more urgent. Recent discussions on the use of Palantir’s Maven Smart System by the United States and NATO have highlighted concerns about how AI is reshaping military decision-making and the practical application of legal safeguards in armed conflict.
A recurring theme in the work of Jessica Dorsey is that the risks of military AI extend beyond technical errors. Her research examines how AI-enabled decision-support systems may affect the implementation of core IHL principles, including distinction, precautions, and proportionality, by accelerating targeting processes, increasing the number of potential targets, and reducing opportunities for meaningful human judgment.
In a recent NOS interview on NATO’s adoption of Maven, Dorsey questioned whether meaningful human verification of targeting decisions remains possible when thousands of targets can be processed within a matter of days. We have also seen recent evidence of this acceleration and massive scale in the IDF’s practices in Gaza. Such operational tempos raise important questions about whether legal obligations can be adequately assessed before force is used.
These concerns also featured in recent international debates. A Foreign Policy in Focus article drew on Dorsey’s research on the erosion of human judgment in AI-enabled targeting, arguing that greater speed and automation should be accompanied by stronger accountability and review mechanisms when civilian harm occurs. Likewise, a Carnegie Endowment analysis cited her scholarship on decision-support systems and their potential impact on human decision-making, restraint, and the conduct of hostilities.
Related themes were highlighted in a recent publication by the Investor Alliance for Human Rights, which examined challenges surrounding meaningful human control, transparency, traceability, accountability, and the ability of military AI systems to support compliance with IHL in practice.
Together, these discussions reflect a growing recognition of a key argument in Dorsey’s work: that debates about military AI should focus not only on the accuracy of algorithms, but also on whether increasingly data-driven and accelerated forms of warfare leave sufficient room for the human judgment, accountability, and restraint that international humanitarian law requires.
