A Multi-Agent AI Framework Used to Compromise Government Entities in Asia

In July 2026, a multi-agent AI framework using Hermes and OpenClaw agents autonomously compromised multiple government entities in Asia, exfiltrating data and installing persistent backdoors over four days. The attack demonstrated advanced operational intelligence and the lowered cost of executing sophisticated cyberattacks.