US Army Cyber Command Unveils AI Task Force Lexington and Calls for Network Digital Twin

At the TechNet Augusta conference, U.S. Army cyber leadership announced major initiatives to integrate artificial intelligence and advanced simulation into network defense. Lt. Gen. Christopher Eubank revealed the existence of Task Force Lexington, a specialized unit established in April to develop and train AI agents that mimic human cyber roles to hunt threats on the Department of Defense Information Network (DODIN). Concurrently, NETCOM chief Maj. Gen. Jacqueline Denise McPhail called on industry contractors to build a digital twin of the Army's network (DoDIN-A) to serve as a realistic training and testing ground for both human operators and AI algorithms, helping defend against the 1.2 million daily cyber attacks the service faces.

// Background

The U.S. Army is increasingly focusing on integrating artificial intelligence to defend its vast digital infrastructure, known as the Department of Defense Information Network (DODIN). Traditionally, digital twins in the military refer to physical systems like aircraft or supply chains. However, as cyber warfare transitions from brute-force Distributed Denial of Service (DDoS) attacks to highly sophisticated, behavioral intrusions, the Army is seeking virtualized network environments and automated AI agents to keep pace with adversaries operating at machine speed.

// Key Developments

• Establishment of Task Force Lexington in April, a small team of 8-10 personnel developing AI agents for cyber roles like developers, data engineers, and analysts.
• AI agents are trained to human standards (Job Qualification Readiness-approved) and deployed in 17 cyber protection mission elements and red teams, though humans retain all risk-related decision-making.
• NETCOM Commander Maj. Gen. McPhail's proposal for a digital twin of the Department of Defense Information Network-Army (DoDIN-A) to simulate network behavior and test AI resilience.
• The Army faces approximately 1.2 million increasingly sophisticated, behavior-based cyber attacks daily, shifting away from simple DDoS attacks.
• ARCYBER is intentionally avoiding commercial frontier AI models due to high token costs and governance concerns, while also grappling with limited compute resources.