Executive Summary: International Cooperation and Capacity-Building as Practical Risk Reduction
In this policy brief, Irma Arguello builds on her remarks at the UNODA Informal Exchanges on AI in the Military Domain, Geneva, June 2026, and approaches the issue from the perspective of international cooperation and practical capacity-building for States.
The focus is placed upon procurement due diligence, auditability, incident communication, red-line analysis and secure decommissioning, with particular attention to the risks of importing dependency through AI procurement. It also includes a short operational annex with five governance questions before signing a military AI contract.
This policy brief is particularly relevant for States developing, acquiring or integrating artificial intelligence capabilities for military purposes.
Key Messages
- Artificial intelligence in the military domain creates systemic risks that no State can manage alone. These risks are distributed, less visible than traditional threats, and may emerge through interactions among systems, doctrines, suppliers, data, human operators and crises.
- International cooperation on military AI should be understood as practical risk reduction. Its immediate purpose is not to force trust-building where trust does not exist, but to reduce strategic opacity, improve predictability and build governance capacities.
- Capacity-buildingshould be treated as risk-reduction infrastructure: it should go beyond generic training to establish a set of practical capacities for judgment, evaluation, supervision, acquisition, auditability, response, accountability and decommissioning.
- For most Statesthe urgent challenge will not be developing their own advanced military AI systems, but buying, licensing, integrating and operating such systems without importing dependency or eroding their sovereignty. In this context, the global AI divide is both a development and a security concern.