IncidentAI-enabled attacks
Taiwan: a team of AI agents breaks into government systems in four days
CriticalTaiwan Ministry of Digital Affairs
Taiwan confirmed that a July 2026 hacking campaign against its government combined human operators with a team of AI agents, including the OpenClaw agent. Over about four days the attackers stole passwords and personnel records, including from the justice ministry, and scanned the nuclear safety agency for weaknesses.
Replay the attack step by step in the Learning Center
How it happened
- Operators set objectives and targets; AI agents then mapped connected government systems and hunted for weak points.
- The agents harvested account credentials and moved between connected systems at machine speed.
- Personnel records were exfiltrated; other agencies, including the nuclear safety regulator, were scanned.
Impact
Government account passwords and personnel records were stolen across several agencies within about four days.
The fix
Taiwan's Ministry of Digital Affairs confirmed the attack after the Israeli security firm Dream reported it.
Lessons
- Agent teams compress weeks of intrusion work into days; response plans must assume that speed.
- Exposed portals, reused passwords and flat networks are what agents exploit first.
- There is still a human choosing the targets, so classic threat intelligence still applies.