Tech Meridian ← LIVE FEED
RU

NEWS · RESEARCH · #369

Position paper: language models not ready for strategic wargames

This arXiv position paper argues that open-ended, LM-based strategic wargames are high-stakes simulations whose outputs can directly shape what actors attempt and what becomes simulated reality. It warns that no LM-enabled wargame should inform planning, doctrine, policy, or crisis response without an auditable safety case, identifies five failure modes (decision laundering, adjudication opacity, role collapse, escalation-through-adjudication, and failure of strategic imagination), and recommends using such wargames only as stress tests rather than safety proofs.

KEY POINTS

  1. This arXiv position paper argues that open-ended, LM-based strategic wargames are high-stakes simulations whose outputs can directly shape what actors attempt and what becomes simulated reality.
  2. It warns that no LM-enabled wargame should inform planning, doctrine, policy, or crisis response without an auditable safety case, identifies five failure modes (decision laundering, adjudication opacity, role collapse, escalation-through-adjudication, and failure of strategic imagination), and recommends using such wargames only as stress tests rather than safety proofs.
  3. Because LM-driven wargames can mislead decision-makers or produce opaque adjudications that alter real-world choices, meaning they must be subject to auditable safety cases before influencing policy or crisis response.

WHY IT MATTERS

Because LM-driven wargames can mislead decision-makers or produce opaque adjudications that alter real-world choices, meaning they must be subject to auditable safety cases before influencing policy or crisis response.

SOURCES & TIMELINE

1