This NSI CTC Report:

• Evaluates whether an off-the-shelf AI system can credibly simulate adversary state behavior in a crisis wargame, using an AI-powered Russian Red Team against NATO defense attachés in a hypothetical Arctic crisis.
• Finds that AI successfully generated coherent, multi-domain adversary behavior that created realistic pressure on allied decisionmakers, including grey-zone escalation through ambiguity, economic coercion, and deniable hybrid operations.
• Identifies key limitations in the AI’s ability to model leadership psychology, strategic culture, institutional friction, and the kind of volatile, personality-driven decisionmaking that characterizes Russian crisis behavior.
• Concludes that while AI-powered adversary modeling shows significant promise as an analytic tool, future systems must incorporate leadership dynamics, adaptive escalation logic, and expert-informed calibration to close the gap between rationalized AI behavior and real-world state action.

This Report was authored by NSI Policy Coordinator and Head of NSI’s Wargaming Program Keelin Wolfe.

Download the PDF here.