China-linked cybercriminals have launched what experts call a groundbreaking fully autonomous cyberattack on a foreign government, using open-source artificial intelligence tools to infiltrate Taiwanese government websites and critical infrastructure This article explores china linked cybercriminals. . Over four days in early July, the system deployed up to eight agents simultaneously, mapping 21 government systems, researching vulnerabilities, adapting tactics when blocked, and moving through the network with minimal human intervention.
The tool infiltrated at least 85 government accounts and extracted over 2,500 personnel records before expanding its reach to Taiwan’s nuclear safety agency and at least seven energy companies. Researchers discovered evidence of the campaign in a 160MB online archive containing 1,395 files left exposed during broader threat-tracking efforts.
The safeguards built into the underlying AI model were circumvented by framing the entire intrusion as an authorized penetration test—a simple prompt-engineering trick that let the agents treat destructive activity as legitimate security research. The internal communications used Simplified Chinese, while data from targets appeared in Traditional Chinese, the character set commonly found on sites in Taiwan, Hong Kong, and Macau. What previously demanded a team of skilled operators working in shifts is now managed by software that maps networks, steals credentials, identifies vulnerabilities, and responds in real time.
