Taiwan’s Ministry of Digital Affairs said an “unusual” cyberattack using artificial intelligence targeted multiple government agencies from a foreign source beginning on July 20. The country’s cybersecurity monitoring unit detected the operation, while the National Institute of Cyber Security issued several alerts during the investigation. Israeli AI company Dream said attackers used open-source AI agents to build an automated hacking tool that operated like a coordinated cyber team.

Dream said the attackers accessed at least 85 government user accounts and collected information on more than 2,500 employees. The operation was later expanded to target Taiwan’s nuclear safety agency and at least seven energy companies. Dream described the case as unprecedented, though cybersecurity researcher Chris Thomas cautioned that AI should not be viewed as a fully autonomous hacking weapon because humans still determine targets, objectives and instructions.

The ministry said its investigation into the attack’s source, methods and impact had concluded, and affected agencies had taken necessary measures. Taiwan has prepared defensive guidance and strengthened monitoring to detect and stop attacks earlier. Authorities did not directly blame China, while reports cited suspicions of China-linked hackers and Dream noted simplified Chinese in internal communications connected to the attack.