Structural

Mitigation strategy

1. Establish and ratify binding international agreements, or formal unilateral declarations, to prohibit the integration of Artificial Intelligence/Machine Learning systems into nuclear launch authority and critical command, control, and communications (NC3) decision-making loops without a time-assured, human-in-the-loop override mechanism. 2. Mandate the use of Explainable AI (XAI) principles for all strategically relevant military decision-support systems to ensure human operators can comprehend model rationale and provenance, thereby mitigating opaque recommendations that could compress decision timelines and bias escalatory actions. 3. Implement rigorous, continuous red-teaming and scenario-based training for operators of strategic systems that requires critical function performance without AI assistance, thereby counteracting agency decay and ensuring cognitive autonomy in high-stakes, uncertain environments.