NATO’s Crossed Swords Exercise Brings AI Agents Into Cyber Operations Training

Executive Summary

NATO introduced AI agents and OpenAI Daybreak into cyber and kinetic operations training during the Crossed Swords exercise. Teams using AI agents for security or operational workflows should assess their autonomy and control boundaries.

What Happened

Affected Products

• OpenAI Daybreak

Vulnerability / Attack Technique

AI agent use in cyber operations

Technical Mechanism

The summary states that AI agents and OpenAI Daybreak were incorporated into joint cyber and kinetic operations training. It does not describe the agents’ architecture, permissions, attack steps, or exploitation mechanism.

AI Security Category

AI-assisted attack

Part of the AI-Assisted Attacks topic.

Relevant Frameworks

RealGround Analysis

AI agents used in cyber operations can create material risk if they execute actions beyond intended authorization or if their outputs are trusted without review. Teams running agentic workflows should test decision boundaries, tool permissions, escalation paths, and human approval requirements before deployment.

Recommended Controls

• Require explicit human approval for high-impact or irreversible agent actions.
• Apply least-privilege access and narrowly scoped permissions to agent tools and environments.
• Log and review agent prompts, tool calls, decisions, and resulting actions.
• Conduct adversarial red-team exercises against agent authorization and operational safety controls.
• Define emergency stop, rollback, and incident-response procedures for agent activity.

Related RealGround Service

AI Agent Business Logic Audit · or start with the free AI Agent Security Scanner.

Sources

securityweek.com: https://www.securityweek.com/natos-crossed-swords-exercise-brings-ai-agents-into-cyber-operations-training/

Summary and analysis written by RealGround from the source above; last updated 2026-11-03.