An international investigation led by Japan has identified around 1,000 servers that were likely used to prepare and carry out cyberattacks using artificial intelligence. According to experts, these systems may have been operating online since at least October.
This is reported by Kyodo, citing the Japan Computer Emergency Response Team Coordination Center.
The investigation began in early October. It is being conducted in cooperation with foreign cybersecurity companies. The servers that were identified contained AI-based tools that may have been used to search for vulnerabilities in systems and to attack websites in various countries. Specialists are currently tracking between 600 and 700 such servers.
According to JPCERT/CC, most of them hosted artificial intelligence tools, including the Artex penetration-testing tool. They were likely used to look for weaknesses in systems and to attempt to steal configuration files.
One of the servers also showed traces of a programming support tool created using AI by the US company Anthropic PBC.
Some publicly accessible servers contained software for analysing the internal structure of mobile applications and checking systems for vulnerabilities. Experts suggest that attackers may have analysed applications and tried to gain unauthorised access via data-exchange interfaces with external systems.
JPCERT/CC notes that advances in AI have made tasks easier for hackers that previously required significant time even for experienced specialists. At the same time, the centre’s analyst Ken Sajo warned that the use of artificial intelligence in cyberattacks is still at an early stage, but that the number of more sophisticated attacks will increase in the future.