Security researchers detailed attacks in Japan where threat actors utilized AI agent workflows, open-source penetration testing frameworks, and LLMs to coordinate intrusions, generate localized phishing pages, and assist server targeting.

In one of the cases analyzed, researchers identified three Internet hotspots used for malicious activities in Japan. The first of these attacks relied on a special tool that leverages dozens of different artificial intelligence programs to automate the entire process of an attack, from identifying security vulnerabilities to intercepting data and creating fake login pages in Japanese. This activity affected dozens of organizations in a short period of time.

In a second case, AI helped people design an attack on a geospatial data server, which resulted in the theft of files and login credentials.…