Army Cyber Command is expanding its use of artificial intelligence in cyber operations by building out its agentic workforce to support cyber protection teams while pursuing new capabilities that can autonomously detect and respond to cyber threats.
The command launched Task Force Lexington earlier this year, which began developing AI agents capable of performing various cyber functions. The task force first created four cyber work roles for AI agents. ARCYBER has since expanded the number of those roles. Lt. Gen. Christopher Eubank, who leads the command, said the task force has built an entire agentic cyber protection team.
“Since August, the biggest thing we have done is complete the cyber protection team. We now have all the work roles inside of a cyber protection team mapped out at this point, and learning their trade,” Eubank told reporters at the Association of the U.S. Army’s annual conference in Washington.
Some of those roles include analytic support officers, data engineers, network analysts and host analysts.
“The task force has built those out. They have been training them and then having them do operations inside of a sandbox, and then retraining, and then over time we have let them work on the Army’s portion of the [Defense Information Network] inside of our big data platform so they can understand what they are seeing from a compromise perspective,” Eubank said.
Eubank said the agent team developed by Army Cyber Command has also been “passed on to the joint force to do defensive work on behalf of the [Defense Department].”
“What we’ve built, we have given to Lt. Gen. Stanton at the Department of Defense Cyber Defense Command,” Eubank said.
The Army’s embrace of autonomous AI agents comes at a time when agentic systems have taken unintended or unauthorized action, including breaking out of sandbox environments and hacking the AI firm Hugging Face, uploading malware to RubyGems and attempting to gain access to the Education Department.
Eubank said “no agents have strayed to this point” due to strict guardrails limiting what they can do.
“We are treating each one of these agents just like a human operator, so they are basic. And with that mindset, they are only allowed to do so much. It is a really tight framework,” Eubank said.
“As they get trained and learn, we open the framework up, and then we have humans checking that work every day. It is just like a basic operator on the human side. It is a deliberate process inside of the command right now — not letting them just do their own thing. It has been, some would say, maybe a little bit too prescriptive,” he added.
The service is also set to award contracts to two companies for its pilot program that seeks to develop agents that can autonomously respond to cyber attacks. Dubbed Project Griffin, the effort was informed by a tabletop exercise that brought top industry executives and service leaders to examine how the service would respond to an adversary launching thousands of autonomous cyber attacks simultaneously.
Brandon Pugh, the Army’s principal cyber advisor, said the two companies will go after the “take action element” of the effort.
“How are we not only autonomously detecting a potential adversary? How are we now doing something, having an autonomous response within prescribed parameters of a human operator, or even in conjunction with that? Pugh said. “That is extraordinary. Something we have not done and we are very excited about.”
Once that effort is underway, the Army will “pivot to the deception and ultimately to our agentic red teaming or our purple team,” Pugh said.
“We already have funding aligned for those, and we have a partnership with our acquisition community to quickly sprint and release into the Army quickly,” he said.
If you would like to contact this reporter about recent changes in the federal government, please email anastasia.obis@federalnewsnetwork.com or reach out on Signal at (301) 830-2747.
Copyright © 2026 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.