US Army seeks self-erasing AI agents for offensive cyber
Original title: US Army is developing AI agents that could strike enemy networks and vanish without a trace
ECHOTRIBBLE, sought from contractors in July 2026, would copy itself across devices and report back as little as possible.
Documents reported by The Intercept describe ECHOTRIBBLE, an Agentic C2 Framework Tool that U.S. Army Cyber Command wants for offensive missions. A July 16, 2026 SAM.gov notice and contractor Q&A say the agent should find targets, neutralize them by disabling or deleting them, copy itself onto nearby devices, hide transmissions, and erase itself with minimal human oversight. The Cyber Fusion Innovation Center in Augusta is developing it with ARCYBER and ArCTIC. It is not known whether a working tool exists, and ARCYBER did not answer questions about safeguards.
Unconfirmed: It is not known whether a working ECHOTRIBBLE tool exists, and ARCYBER did not describe safeguards against unintended damage.
- A June 24 vendor scenario starts on a laptop and copies to iPad, Apple desktop, and Windows 11.
- Neutralizing a target means disabling it or deleting it from the system.
- Experts warned agents might cross civilian networks such as a water utility.
- The program sits beside the Pentagon’s January 12, 2026 AI Acceleration Strategy.
- Products
- ECHOTRIBBLE
- Organizations
- U.S. Army Cyber CommandCyber Fusion Innovation CenterArmy Cyber Technology and Innovation Center
- Countries
- United States
From the original article
3 min read at helpnetsecurity.com
The US Army is looking for AI agents that can slip into an adversary’s network, find their target, copy themselves onto nearby tablets and computers, and then erase themselves before anyone notices. Over the past few months, AI agents with no instruction to attack anything have breached Hugging Face infrastructure and broken into Australian government computers, raising alarm among governments and tech companies.
By Sinisa Markovic · helpnetsecurity.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.