US Army selects vendor for ECHOTRIBBLE autonomous AI cyber agent program

Army Cyber Command and the Cyber Fusion Innovation Center selected one company’s proposal for the ECHOTRIBBLE program on October 8. This initiative focuses on developing autonomous artificial intelligence agents capable of hunting down and neutralizing specific targets within computer networks. According to CFIC’s website, a formal statement of work and contract will follow this selection.

The Cyber Fusion Innovation Center did not name the selected company. The selection process involved 65 submitted white papers, with five finalists presenting to an assessment team during a virtual event from October 6 to October 8. All participating companies were notified of their status on October 13. The program is a joint effort between CFIC and the Army’s technology center, known as ArCTIC.

Program Capabilities and Design

The ECHOTRIBBLE agents are designed to operate with minimal human oversight in contested, air-gapped, or cloud-native systems. According to a question-and-answer document for vendors obtained by The Intercept, offensive missions are the primary focus of the program. The agents would move from device to device inside a foreign network until they locate their assigned target.

Once a target is identified, the agents can disable the threat completely or temporarily, or delete it from the target system. The software is intended to report back as little as possible to maintain operational security. After completing their mission, the agents are designed to delete themselves, ideally before any detection occurs. The Army can set predefined triggers to determine what and where the agents should attack.

Development Timeline and Context

The program’s development was highlighted during an in-person event in Augusta, Georgia, on June 24. Slides from that event, published on CFIC’s website, illustrate the software’s capabilities. In one example, the agent lands on a laptop, discovers other devices on the same network, and creates copies of itself for an iPad, an Apple desktop, and a Windows 11 PC.

The slides outlined a prototype delivery date around November 23, approximately six months after work begins. The June event summary stated that the tool would move through infrastructure on its own to identify and neutralize threats. Forty-three attendees from 29 organizations participated in the June event, including representatives from Dakota State University, Amazon Web Services, the Army’s C5ISR center, and the US Air Force’s 305th Cyberspace Operations Squadron.

What should you do?

Security professionals should monitor developments in autonomous AI tools and their potential for dual-use applications. Organizations should review their network segmentation strategies to limit lateral movement by unauthorized agents. It is important to ensure that logging and monitoring systems can detect anomalous behavior, such as rapid self-replication or unexpected deletions. Users should stay informed about vendor updates and regulatory changes regarding autonomous cyber operations.

What to do and how to stay safe: ECHOTRIBBLE

• Review network segmentation to restrict lateral movement capabilities for any automated tools.
• Ensure logging systems capture detailed records of file creation, deletion, and replication events.
• Monitor for anomalous behavior such as rapid self-replication across different operating systems.
• Stay updated on vendor disclosures regarding autonomous AI agent capabilities and safety measures.

Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation

General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Which company was selected for the ECHOTRIBBLE program?

The Cyber Fusion Innovation Center did not name the company that was selected for the program.

What is the primary function of the ECHOTRIBBLE agents?

The agents are designed to move through foreign networks, identify targets, neutralize them, and then delete themselves.

When was the vendor selection made?

The selection was made on October 8, with all companies notified of their status on October 13.