June 2 Cybersecurity Report: Chinese Cyberespionage and AI Misuse Threats

Taiwan has long been a frontline target of Chinese cyber-espionage, with government agencies and critical infrastructure under sustained pressure. Israeli cybersecurity company Dream’s findings mark a potential step change: rather than merely helping human operators write code or analyze data, open-source AI agents allegedly mapped networks, selected vulnerabilities, attempted intrusions and changed tactics when blocked. The case matters because autonomous, parallel agents could let attackers operate faster and at far greater scale while lowering the expertise and cost needed for sophisticated campaigns.

Dream said on Aug. 12 that the suspected China-linked campaign ran for four days in early July, deploying as many as eight AI agents at once. The agents mapped 21 government systems, compromised at least 85 accounts and stole more than 2,500 personnel records, while also probing Taiwan’s nuclear safety agency and at least seven energy companies. Taiwan’s Administration for Cyber Security said on Aug. 13 that it had issued alerts beginning July 20 and helped affected agencies complete remediation, though the government did not publicly attribute the operation to China.