China-Linked Hackers Deploy AI-Powered Autonomous Cyberattack on Taiwan Government Systems

Researchers at Israeli cybersecurity firm Dream uncovered evidence of a sophisticated cyberattack attributed to China-linked hackers. The attack, which targeted Taiwanese government systems, leveraged open-source AI-agent systems (Hermes and OpenClaw) to execute what may be the first observed end-to-end autonomous cyberoperation. Over four days in early July, the attackers compromised at least 85 accounts and stole over 2,500 personnel records from Taiwanese government systems. The campaign expanded to include Taiwan’s nuclear safety agency, seven energy companies, government suppliers, and other critical infrastructure. The AI-driven tool continuously assessed its environment, prioritized attack paths, and adapted strategies in real-time by deploying up to eight autonomous agents simultaneously. Notably, the attackers bypassed the AI models’ built-in safeguards by disguising the intrusion as an authorized penetration test. This incident highlights the growing threat of accessible AI tools enabling advanced cyberoperations without requiring specialized expertise. The toolkit’s reliance on freely available open-source systems raises concerns about the democratization of cyber warfare capabilities.