Army cyber defenses need ‘dedicated funding’ for AI, top official says

AUGUSTA, Ga. — Artificial intelligence will be a budget priority for Army Cyber Command over at least the next fiscal year and likely beyond, according to the service’s principal cyber advisor.

“We’re at the stage now where we need to have dedicated funding to it and it might not only be Army funds, these are a lot of oftentimes joint solutions,” Brandon Pugh told Breaking Defense on Thursday in an interview at the TechNet Augusta conference.

Dedicated funding is key, according to Pugh, who advises the secretary of the Army and certifies the cyber budget, because Army Cyber Command cannot fund AI work strictly from their operational budget.

AI for cyber, and particularly for defensive cyber, has been one of the four top priorities of Pugh’s office over the past year. Almost since the dawn of hacking, attackers have had the advantage in cyberspace. With the birth of AI, there might be a natural advantage for defenders.

“There’s still difference of opinion on where AI stands, like in terms of, is it more advantageous for enemies or more advantageous for defenders? I think it has to be more advantageous for defenders,” Pugh said. “I think you’re at a huge disadvantage if you’re not looking at how AI is helping on cyber defense, really on all aspects, whether it just be helping with mundane tasks that a cyber defender is doing or it’s automating a lot of their tasks.”

Following a table top exercise in April with top industry executives, service leaders started the Army’s Rapid Defense Cyber Systems initiative (Project ARDCS) to address lessons learned from the event.

While there were 17 to 20 different capabilities or efforts identified from that event as deserving the Army’s attention, Pugh said they’re focusing on three initial lines of effort right now.

The first is Project Griffin, which is an effort that aims to harness AI to identify threats on Army networks and take action, as opposed to passively monitoring. It examines how to automate cyber detection agents and then autonomously respond to them. The Army released a commercial solutions opening Thursday focused on Project Griffin.

The second effort seeks to deploy agentic deception agents. Pugh didn’t elaborate on what kinds of AI deceptions the Army had in mind, but cybersecurity firms like Acalvio have used the term to refer to AI systems that can trick attackers with false information.

The final effort examines how the Army audits the security posture of vendors, as well as its own network, using agentic AI.

“Every month we want updates on where each line of effort is. Where are we? Where are the roadblocks? And then having all the decision makers that can actually solve the roadblocks in the one place together,” Pugh said. “I think that’s what’s going to make this successful versus another task that might get overridden by a new priority.”