China-Linked Hackers Wield AI-Powered Tool in Taiwan Government Network Breach

Taiwan's government networks were compromised by an autonomous hacking tool created using open-source AI agents, according to Israeli AI firm Dream. The attack, which lasted four days in early July, broke into at least 85 accounts and stole more than 2,500 personnel records.

The hackers, suspected to be linked to China, used the Hermes and OpenClaw agent frameworks to carry out tasks on their own. When one path was blocked, the system would re-score its options and hand a fresh agent the job of finding new information and building another route.

Dream's chief strategy officer, Amir Becker, stated that he had not seen a government hit by an 'end-to-end autonomous attack' before, and warned that every government should treat itself as permanently under assault. The firm also found that the model's security measures had been sidestepped by disguising the intrusion as a sanctioned security test.