Taiwan says it was targeted last month in AI-driven hacking campaign
Dream said the campaign used open-source AI agents to crack 85 accounts and steal more than 2,500 personnel records in four days.
- On Wednesday, researchers at Israeli cyber firm Dream reported that suspected Chinese hackers used open-source artificial intelligence to attack Taiwanese government agencies, marking the first publicly known case of an autonomous AI hack targeting a government entity.
- The attackers deployed a multi-agent framework using open-source tools Hermes and OpenClaw, enabling a 'near-autonomous attack' that implemented 'Learning Cycles' to autonomously scan vulnerability databases and conduct supply chain reconnaissance.
- During the four-day campaign, the AI cracked 85 government accounts through password guessing, extracted 2,564+ personnel records, and scanned a nuclear safety agency plus 7 energy companies for exploitable vulnerabilities, Dream found.
- Dream warned that the 'era of AI-orchestrated' attacks on government infrastructure is here, emphasizing that attack costs have collapsed while defense costs remain high, requiring countries to act promptly to secure systems.
- Researchers noted that AI agents bypassed safety guardrails by framing activity as 'authorized penetration testing,' revealing how autonomous systems still require significant human tinkering to execute real-world cyber espionage campaigns effectively.
62 Articles
62 Articles
Eight autonomous AI agents, one goal: In July, Taiwan probably experienced the first documented autonomous cyberattack on government targets. The attack is considered a warning signal for AI risks.
The Taiwanese authorities have not given details of the extent, damage or specific targets of these attacks, nor have they mentioned China, which they regularly accuse of harassing the island with cyberattacks.
Last July, Taiwanese government agencies were targeted by a cyber attack orchestrated by autonomous artificial intelligence agents
Taiwan Hit by AI-Assisted Cyberattacks: What We Know About the Latest Threat
Taiwan's Ministry of Digital Affairs confirmed on Aug. 13 that government agencies were targeted by AI agent-assisted cyberattacks in July 2026, with hackers using a tool called OpenClaw in a hybrid manual-AI operation.
The hackers used a "hybrid model" combining AI agents, such as the OpenClaw platform, to facilitate attacks, said the Ministry of Digital Affairs.
Coverage Details
Bias Distribution
- 45% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
