Armadin and TENEX.ai Execute Largest Controlled AI Cyberattack

The engagement pitted Armadin’s autonomous swarm—comprising 26,000 agents—against the institution’s external perimeter and internal network. Operating under zero-knowledge conditions, the swarm mapped complex dependency graphs to identify 38 validated attack paths and 238 distinct security findings. While the attack unfolded, TENEX.ai’s platform ingested and triaged over 101,000 alerts, reconstructing the entire offensive sequence across 231 billion raw events.

This forensic feat would have required a five-person analyst team approximately 2,400 hours to perform manually. Instead, the AI-driven defense platform correlated 2,164 source addresses and mapped findings to industry frameworks like MITRE ATT&CK in real time. The exercise highlights a widening chasm in corporate security: while attackers increasingly leverage agentic AI to exploit networks at machine speed, many organizations remain tethered to traditional, slower-paced detection models. According to Armadin CEO Kevin Mandia, this adversarial loop is becoming the new baseline for enterprise security, forcing defenders to adopt autonomous systems to survive.