Hackers used AI to attack Taiwan's energy and government systems

🛡 Hackers used AI to attack Taiwan's energy and government systems

From July 1 to 4, 2026, a "near-autonomous" attack was carried out against Taiwan, as described by researchers at Dream: an AI tool based on the open-source frameworks Hermes and OpenClaw deployed up to eight sub-agents in 12 "waves" over four days.

🌍 The first documented end-to-end offensive cycle conducted by AI agents against government systems. The tool is built from public frameworks — a low barrier to entry for other actors.

👤 The breakthrough was not through a zero-day, but through typical vulnerabilities: open APIs, exported employee data, password patterns. Checklist: inventory public endpoints and check access to SSO secrets.

Source 1: https://www.theregister.com/security/2026-08-12/near-autonomous-ai-agents-attack-taiwans-nuclear-safety-agency/5287055 Source 2: https://www.theguardian.com/technology/2026/aug/13/taiwan-ai-assisted-cyber-attacks-overseas