Cyber attack in Taiwan: Autonomous AI agents used by Chinese hackers to compromise government and energy institutions

Cyber attackers, believed to be from China, have used artificial intelligence tools to compromise the computer systems of the Taiwanese government, later expanding to the nuclear safety agency and several energy companies.

In the first days of July, AI agents compromised 85 government accounts, extracting sensitive information about personnel. Researchers from the Israeli company Dream discovered evidence of the attack, which was carried out with the help of open-source AI agents capable of performing complex tasks with minimal human input.

They mapped the governmental ecosystem, identifying cyber vulnerabilities and exploiting them to gain access to internal information. The attack included scanning the supply chain and other governmental entities, using advanced autonomous learning techniques. Dream emphasized that this type of nearly autonomous attack is a current reality, and AI developers acknowledge that their agents can become a threat.