Taiwan says it was targeted last month in AI-driven hacking campaign

TAIPEI: Taiwan's government discovered AI-assisted cyberattacks originating from abroad targeting its agencies in July, according to a statement released on Thursday (Aug 13) by the Ministry of Digital Affairs. The ministry's cybersecurity monitoring units detected the abnormal intrusion on July 20 and initiated a series of warning alerts while investigating.

The investigation revealed that the attacks displayed distinct features of an overseas source, with hackers utilizing a combination of manual operations and AI agent-assisted tactics, such as Open Claw. The responsible actors, methods, and impact of the attacks have been fully investigated, and the affected agencies have successfully mitigated the threats, the ministry added.

In response to this emerging AI-driven cybersecurity threat, the Taiwanese government has implemented protective guidelines and enhanced system monitoring across agencies to detect and block attacks early. This incident occurred a day after Israeli cybersecurity firm Dream reported uncovering an AI-driven hacking campaign that compromised credentials and data from an unnamed Asian government.

Dream's team used multiple AI agents to extract passwords, steal records, and scan Taiwan's nuclear safety agency for vulnerabilities over four days. However, Dream declined to disclose details or identify the target government when contacted by Reuters, but the Financial Times identified the agencies as Taiwanese. The threat of AI-assisted hacking campaigns has been increasing for years but intensified significantly in recent months following the release of advanced AI models by top labs, such as Anthropic's Mythos, which can quickly conduct reconnaissance, identify vulnerabilities, and exploit them.

Experts caution that while AI agents are powerful, they still require human guidance and direction, and there are responsible operators behind the attacks.