At the Responsible AI in the Military Domain (REAIM) Summit (4–5 February) in Coruña, Spain, the message was clear: it is time to move from principles to practice. Convened by Spain’s Ministry of Foreign Affairs, Ministry of Defence, and the A Coruña City Council, the Summit brought together a diverse mix of States, industry, academia, civil society, NGOs and INGOs, with particular value found in cross-sector dialogue and informal exchanges beyond the panels.
A strong theme was the growing urgency to operationalize REAIM outcomes into concrete requirements, standards, and shared understandings of what responsible military AI means in practice. While this work must be State-led, success depends on close collaboration with industry and academia, who develop much of the technology now entering military use.
RAW members Jessica Dorsey, Marijn Hoijtink, and Martine Jaarsma each contributed to these discussions through panels, expert engagements, and ongoing research initiatives.
From Dr. Damian Copeland, Director of Article 36 Legal in his recap of the Summit, “Outstanding work underway was noted within university research programmes examining the technological, military, ethical, legal, and social dimensions of military AI. Notable examples include Utrecht University’s Realities of Algorithmic Warfare programme, the Center for War Studies, University of Southern Denmark’s AutoNorms project, and the Adelaide University’s Research Unit on Military Law and Ethics (RUMLAE). The Responsible by Design Institute Ambassadors, Dr. Jessica Dorsey, Dr. Zena Assaad and Dr. Ingvild Bode have each made a strong impact on the REAIM26 discussions, reinforcing their status as truly world-leading and influential experts in their respective fields.”
Contributions and reflections from RAW members
We note that states are investing heavily in military AI, making initiatives that translate agreed REAIM principles into practical measures increasingly essential. Efforts such as the Responsible by Design Institute’s Expert Meetings on the Legal Review of Autonomous Weapons and the newly established Independent Advisory Board on Legal Reviews represent important steps toward implementation, including through Article 36 legal reviews. Realities of Algorithmic Warfare co-Director Jessica Dorsey is an ambassador and an advisory board member to this initiative.
During the Summit, Dorsey participated in the panel discussion at REAIM entitled Responsible Use of AI in Armed Conflict: Upholding Respect for International Law, co-organized by Harvard Law School’s Program on International Law and Armed Conflict and the Government of Switzerland. Building on the REAIM Summits in The Hague (2023) and Seoul (2024) and the Blueprint for Action, the discussion examined how international humanitarian law applies to AI-enabled military applications, from targeting and weapons functions to cyber operations, detention processes, de-mining, and humanitarian action, with particular attention to protecting civilians, civilian objects, and fighters hors de combat.
Additionally, before the Summit, Dorsey was an invited expert to a NATO-led tabletop exercise with state representatives and NATO’s Responsible AI team designed to test NATO’s Responsible AI Principles against a escalatory use-of-force scenario as well as a Centre for International Governance and Innovation-led workshop on their Responsible AI matrix development.
Moreover, RAW member Marijn Hoijtink participated in the panel ‘Safeguards to Ensure Reliability and Trustworthiness,’ alongside Dr. Damian Copeland (Article 36 Legal), Dr. Scott Mongeau (Defense lead at Google), moderated by Dr. Kerstin Vignard (Johns Hopkins Institute for Assured Autonomy). The panel discussed the need for a continuous certification and verification model to keep pace with non-deterministic systems, and the importance of involving a diverse set of stakeholders throughout the AI system lifecycle. In her contributions, Hoijtink emphasized the importance of incorporating responsibility and legal obligations into the design phase but also warned that not all challenges can be engineered away: “There is a danger that a focus on technical fixes narrows our attention to what is measurable or optimizable, while obscuring the much bigger questions we need to ask: What do we actually want these systems to do? What should they not do? And in which contexts should we decide not to use AI at all?”
Finally, RAW member Martine Jaarsma moderated the panel ‘AI in Cyber Operations, Electronic Warfare and Information Operations. Decision-making and Decision-Support Systems’, which included Sushil Shah (AWS), Neeti Pokhriyal (RAND), Nicolas Gastón Rozado (Indra) & Alfonso Bello Bocio (Barcelona Supercomputing Center). The panel discussed designing systems for responsible use, focusing on challenges related to data, model collapse, and accountability.
The contributions of Dorsey, Hoijtink, and Jaarsma highlight RAW’s ongoing commitment to advancing responsible, accountable, and legally grounded approaches to military AI. As REAIM 2026 emphasized, translating principles into practice will require continued collaboration among states, industry, academia, and civil society, a process to which RAW researchers remain actively and substantially engaged.
