Army hosts industry leaders for AI tabletop drill on cyber defense

The Army brought 14 senior cybersecurity executives to the Pentagon on April 27 for its second artificial intelligence tabletop exercise, an event aimed at speeding the use of agentic AI in cyber defense.

C-suite leaders from Amazon Web Services, Google, Microsoft, OpenAI, CrowdStrike, Palo Alto Networks and other firms joined Army leaders for the half-day session. The Office of the Principal Cyber Advisor organized the exercise, with design and moderation from the Special Competitive Studies Project and support from U.S. Cyber Command, U.S. Army Cyber Command and the Army Cyber Institute at West Point.

“We have to move faster to deliver the best capabilities to the warfighter,” said Secretary of the Army Dan Driscoll. “One of the hacks to do that is through better, more effective collaboration with industry.”

Set against a hypothetical Indo-Pacific crisis, the exercise explored how an adversary might use AI to launch continual, adaptive cyberattacks faster than human operators could counter. Participants were asked to spotlight scalable, existing AI-enabled tools that could give Army cyber defenders an edge.

The discussion focused on two core challenges: developing agentic AI capabilities across the Army’s digital terrain and mitigating weaknesses stemming from varied networks, legacy systems and uneven modernization. Brandon Pugh, the principal cyber advisor to the secretary of the Army, said the format reflects a new approach to working with the private sector. “We are not here to develop new requirements from scratch,” Pugh said. “We are here to identify scalable, adaptable and existing AI-driven capabilities that can give our cyber defenders a decisive advantage today.”

Lt. Gen. Christopher Eubank, who leads Army Cyber Command, said the exchanges underscored how to balance people and machines on the network. “Speed wins, scale decides, and you have to determine the difference in speed — human speed, machine speed and organizational speed — and then leverage AI to do the things that it should be doing at speed,” Eubank said. He added that he left the session with 19 areas to refine or improve, none tied to specific products.

The event builds on the inaugural AI TTX held in September 2025, which assembled roughly 15 CEOs representing more than $15 trillion in enterprise value and led to Project ARIA — the Army Rapid Implementation of Artificial Intelligence. That initiative set three lines of effort: establishing a “model armory” to deliver AI to the tactical edge, deploying agentic tools to automate the planning, programming, budgeting and execution process, and modernizing supply chains with AI.

Unlike the first meeting, AI TTX 2.0 concentrated on cyber defense and included a review of policy gaps that could slow enterprise-wide AI adoption. Army officials plan to use rapid prototyping authorities at Army Cyber Command and acquisition initiatives such as FUZE to pilot promising capabilities within 30 to 90 days, with the goal of getting solutions to operational units soon after.

“The Army that masters the integration of data, AI compute and human judgment into every warfighting function will have a decisive advantage,” Driscoll said. “The Army that fails to do so will be outpaced, outmaneuvered and unable to achieve its objectives.”