Key Highlights

  • Researchers identified an unprecedented autonomous AI Cyber Attack that compromised at least 85 government accounts in Taiwan over four days.
  • The attack extracted more than 2,500 personnel records before expanding to Taiwan's nuclear safety agency and at least seven energy companies.
  • Attackers used open-source AI agent systems that deployed up to eight autonomous agents simultaneously to map systems and adapt tactics.
  • Taiwan faced an average of 2.6 million Chinese cyber attacks per day in 2025, according to the island's National Security Bureau.

Autonomous AI Tool Coordinated Multi-Stage Breach

Suspected Chinese hackers used publicly available AI tools to compromise government websites in Taiwan in what researchers describe as a first-of-its-kind breach. Researchers at Israeli cyber security company Dream said the attackers built an autonomous hacking tool that behaved like a coordinated cyber team, deploying up to eight AI agents simultaneously over four days at the start of July. The tool mapped 21 government systems, researched vulnerabilities, and changed tactics when blocked, ultimately compromising at least 85 government accounts and extracting more than 2,500 personnel records before expanding to Taiwan's nuclear safety agency and at least seven energy companies.

Researchers Point to Chinese Origins, Though Attribution Remains Unconfirmed

Dream has not formally attributed the attack to a specific group, though researchers noted the use of Simplified Chinese in internal attacker communications suggests a high probability of a China-linked operator. Data recovered from the target was written in Traditional Chinese, a script commonly used only in Taiwan, Hong Kong, and Macau government systems. The hacking tool used two open-source AI agent systems that can be downloaded and used to carry out autonomous tasks, though researchers could not identify which underlying AI model powered the agents, noting that the model's safety restrictions had been bypassed by framing the activity as an authorized security exercise.

Officials Warn of Growing AI-Driven Cyber Threats

A spokesperson for Taiwan's Ministry of Digital Affairs said the government handles all incidents affecting critical infrastructure according to established response procedures, noting that AI integration has transformed the nature of cyber security threats by making attacks more automated. Dream's chief strategy officer said the incident represented the first fully autonomous end-to-end cyber attack he had observed against a government target, adding that governments must now assume they face permanent cyber threats. Taiwan recorded an average of 2.6 million Chinese cyber attacks per day in 2025, up 6% from the previous year, according to the island's National Security Bureau.