An investigation by GMO Cybersecurity by Ierae confirmed the misuse of the AI-powered penetration-testing tool ARTEX in unauthorized access targeting Japanese companies. The attackers targeted approximately 20 companies, of which 8 disclosed information leaks in September and October.
GMO Cybersecurity by Ierae announced that it had confirmed the misuse of the AI-powered penetration-testing tool ARTEX in unauthorized access targeting Japanese companies. ARTEX is an open-source tool originating in China that is designed to detect vulnerabilities by connecting to external LLMs, but it was allegedly misused in attacks on Japanese companies.
The company’s investigation found that the attackers were operating with a focused target set, mainly in Japan, and that approximately 20 Japanese companies were affected. The industries targeted varied by attacker. The main sectors identified included digital gifts, retail, education, insurance, finance, e-commerce and logistics, spanning multiple industries.
Based partly on the results of CrowdStrike’s investigation into the South Korean incident, the company believes that the attack group prepared different attack hosts for each campaign. No specific AI model was identified in the Japanese case on the blog, but methods for using AI to search for vulnerabilities and evade defensive equipment were described.
The investigation noted that the expansion of the damage was attributable not only to advances in AI, but also to an increasing number of cases in which skilled attackers use AI to exploit mismatches between offensive and defensive capabilities. The company said it had shared the information with law enforcement authorities.