Taiwan's government agencies faced an AI-powered cyberattack, compromising extensive sensitive data and highlighting vulnerabilities.
On August 13, Taiwan's Ministry of Digital Development disclosed a cyberattack on various governmental bodies involving artificial intelligence tools. The ministry indicated that this attack originated internationally, with the assailants employing both manual and AI-assisted methods, specifically the OpenClaw tool.
Anomalies in digital security were first detected in July, leading the National Cyber Security Institute to alert government agencies and commence an investigation. The ministry assured that all affected organizations have implemented measures to bolster their security.
In reaction to the breach, authorities have rolled out recommendations to counteract future threats and have intensified the surveillance of government systems. Notably, cyberattacks against Taiwan’s critical infrastructure surged 6% in 2025, with the National Security Bureau reporting an alarming average of 2.63 million attacks per day, often linked to hybrid threats from China.
Details of the Incident
On July 29, Israeli firm Dream reported an attack on an unnamed Asian government body, which sources believe was related to Taiwan. Researchers reconstructed the hacking setup, which executed 12 simultaneous attacks over four days using publicly accessible tools.
The breach led to the compromise of 85 employee accounts, with the majority unwittingly granting attackers access to internal networks. The stolen data included over 2,500 personnel documents, login credentials for internal databases, and blueprints for network infrastructure.
The breach eventually reached contractors involved in government IT, the nuclear safety authority, government email systems, and seven energy companies. Dream pointed out that the lack of technical sophistication was overshadowed by the operational autonomy enabled by AI tools, which allowed for rapid vulnerability scanning and exploitation.
Experts have raised concerns about tools like OpenClaw, which CertiK analysts warned could lead to significant threats, including data leakage and system hijacking. The Cyber Center of China has also echoed these warnings.
