Warfare at AI Speed and the Extended Meaningful Human Control Problem

As militaries increasingly incorporate Artificial Intelligence (AI) into their decision-making architecture, legal and ethical debates have primarily focused on Autonomous Weapon Systems (AWS). This paper argues that equally urgent challenges concern AI-enabled Decision-Support Systems (DSS-AI), even though these systems formally preserve human authority over the use of force. We develop the concept of extended meaningful human control (MHC) to address normative risks posed by DSS-AI in operational environments where time pressure, cognitive constraints, and system opacity converge to marginalize human agency in the decision-making loop. Our argument draws on two scenarios in which the integration of AI in military decision-making is linked to the scarcity of temporal resources—whether inherent (as in nuclear early warning systems) or induced (as in the alleged use of DSS-AI by the Israeli Defense Forces in Gaza). Both cases illustrate how taking critical military decisions at AI speed raises profound ethical and legal concerns. To resist the erosion of MHC, we propose three implementation pathways: human-centered system design, clearly distributed responsibility within command structures, and training programs to counter cognitive and institutional sources of automation bias. These measures should be codified in a multilateral treaty or, failing that, embedded in national doctrines and military manuals to resist the systemic drift toward “warfare at AI speed”.