In this brief, scholars explain how they designed a wargame simulation to evaluate the escalation risks of large language models (LLMs) in high-stakes military and diplomatic decision-making.
