Taiwan Reports AI-Assisted Cyberattack on Government Agencies
Taiwan says an overseas cyberattack that began on July 20 combined human operators with AI agents and targeted government systems; officials did not identify a responsible country.
In short
- Taiwan says an overseas cyberattack that began on July 20 combined human operators with AI agents and targeted government systems; officials did not identify a responsible country.
Taiwan’s government said on August 13 that it had investigated an AI-assisted cyberattack targeting public agencies, describing the incident as an overseas operation that combined manual hacking with artificial-intelligence tools.
The Ministry of Digital Affairs said its monitoring systems detected abnormal activity beginning on July 20. Taiwan’s National Institute of Cyber Security issued alerts while authorities investigated the source, methods and impact. The ministry said affected bodies had since completed response measures, and that the government had strengthened monitoring and issued protective guidance for this emerging form of attack.
Taiwanese officials did not accuse China or name any other country. That distinction is important because separate reporting has linked the operation to suspected China-connected actors, but neither Taiwan’s official statement nor the cybersecurity company that detected the intrusion publicly attributed it to a specific group.
The Guardian, citing the Financial Times and Israeli cybersecurity company Dream, reported that the attackers used open-source AI agents to assemble an automated tool that operated like a coordinated hacking team. Dream reportedly said at least 85 government accounts were compromised and more than 2,500 personnel records were extracted before the operation expanded toward Taiwan’s nuclear-safety agency and at least seven energy companies. Those figures were reported by the company and were not included in the Taiwanese ministry’s public account cited by the Guardian.
The ministry said investigators found a hybrid operation in which human operators worked alongside AI agents. It also said the activity showed clear signs of originating overseas, without providing a public attribution.
The case underscores a growing cybersecurity concern: AI systems can accelerate reconnaissance, vulnerability discovery and exploitation while remaining directed by human operators. A security researcher quoted in the source cautioned that such campaigns should not be treated as fully autonomous, because people still select targets and define objectives.
Taiwan has repeatedly warned that its institutions face sustained cyber pressure. Its National Security Bureau said in January that attacks against key infrastructure averaged 2.63 million a day in 2025, six percent more than a year earlier. Officials have previously described some cyber activity, disinformation and military pressure as elements of broader hybrid threats.
China’s Taiwan Affairs Office did not immediately respond to requests for comment reported by the source. The absence of confirmed attribution means responsibility for the July attack remains unresolved.
