Army Cyber Command (ARCYBER) is expanding the role of artificial intelligence (AI) agents in its cyber operations, assigning them new responsibilities across defensive and offensive operations and network security while ensuring human operators maintain oversight of their work.
Speaking on Oct. 12 at the 2026 Association of the United States Army (AUSA) conference in Washington, D.C., Lt. Gen. Christopher Eubank, ARCYBER commanding general, said the command is building a workforce of AI agents that can operate alongside humans on the Army’s portion of the Department of Defense Information Network.
“We are in the process of creating an agent workforce that works hand in hand with the human workforce, so human-machine teaming each and every day,” Eubank said.
The effort began in April under Task Force Lexington, which initially focused on developing AI agents for a handful of cyber roles like exploitation analysts and data engineers.
Since then, ARCYBER has expanded the effort to map an entire Cyber Protection Team of AI agents, including roles in analytic support and network analysis, Eubank said.
The command is applying those capabilities across missions to “operate and secure the network, defend the network, or attack the network,” he said.
As the agents take on more responsibilities, ARCYBER is evaluating which tasks they can perform independently, and which still require human review.
He said the task force asks: “Do we let it do that on its own, or is it still something a human should check?” Eubank said the command is taking an incremental approach, monitoring the agents’ performance, retraining them as needed, and gradually expanding their autonomy as they demonstrate their capabilities.
“This is an iterative process. It’s happening daily and weekly inside headquarters as we watch agents learn and grow, then decide what we should let them do on their own versus what needs a human in the loop,” Eubank said. “All work right now is safeguarded with a human in the loop as we continue to learn across this very important emergent technology.”
Eubank emphasized that the command views AI as a way to strengthen its workforce rather than replace personnel. He said human operators will remain essential to managing, evaluating, and retraining the AI agents as the technology takes on more routine tasks.
“Every machine team we have checks in with the human team daily… The humans are checking their work. If there’s retraining to be done, then the retraining occurs,” he said.
This initiative comes as the military weighs how to expand AI capabilities while managing the risks of giving autonomous systems greater responsibility in cyber operations.
Brandon Pugh, the Army’s principal cyber advisor, acknowledged those concerns on the same panel but cautioned against letting them slow adoption of the technology.
“We can’t shy away from fully embracing the technology because of potential fears,” Pugh said.
“No matter what we do from a policy or regulatory standpoint, bad actors might seek to do harm leveraging SI. It is on us to make sure that our defenders and our operators have an advantage of using the same technology and are staying ahead of those individuals,” he said.
Pugh used the term “super intelligence,” or “SI,” to refer to AI, reflecting terminology the Trump administration recently adopted for official executive branch documents and communications.
“SI is real. It’s here. It’s here to stay, and I think we should all just embrace it,” Eubank said.
Army begins Project Griffin awards
Pugh also shared that the Army is preparing to announce its first Project Griffin awards, an initiative aimed at developing AI agents that can autonomously detect malicious activity on Army networks and respond to cyber threats. Pugh said the service expects to announce awards to two companies on Friday, Oct. 16. He did not identify the companies, but said the Army received more than 100 industry submissions. They narrowed the field to 10 companies and ultimately selected two.
According to solicitation documents, Project Griffin aims to develop an autonomous, AI-driven capability called the Intelligent Response and Orchestration Node. The system would serve as a central decision-making component for automating cyber defense actions on the Army’s portion of the Department of Defense Information Network.
Pugh said the initial awards will focus on Project Griffin’s response capabilities, moving beyond detecting potential threats to taking action against them within parameters established by human operators.
“The one that we plan to award on Friday is our take action element. So essentially, 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 haven’t done, and we’re very excited about,” he added.