The Department of War is integrating artificial intelligence (AI) at a pace that outstrips institutional capacity to validate the technology and develop the force to wield it safely. The link between AI capability and operational employment is not technical expertise, but data literacy. Operators and commanders must understand the origin and fidelity of information underpinning their models. Without this literacy, recognizing when foundational data has been corrupted or injected by an adversary becomes exceptionally difficult. Using the Air Force’s wargaming enterprise and the Collaborative Combat Aircraft program as case studies, this article exposes three interlocking vulnerabilities: data integrity concerns, a trust gap as simulations transition to physical platforms, and institutional shortfalls in professional military education. To bridge this gap, this article proposes a five-tier Acceptable Level of Risk schema, adapted from Air Force risk management doctrine, equipping commanders with a shared vernacular to assess, communicate, and manage AI integration risk.
Strategic Horizons (ISSN 3066-7801) is a peer-reviewed, open-access journal. This article appears in Volume 2, Number 2 (2026).
The full text is available on the journal’s site: Strategic Horizons article page ↗, or as the journal issue PDF ↗.