As a China-linked hacking group utilized artificial intelligence-based penetration tools to attack government agencies and corporations worldwide, South Korean diplomatic and policy-related organizations were also found to have been targeted.
Adam Meyers, Head of Counter Adversary Operations at U.S. cybersecurity firm CrowdStrike, stated during an online briefing for domestic reporters on the 17th that multiple South Korean targets were observed in connection with the attack activities of the China-linked hacking group “Vapor Panda.”
Enlarge image
Meyers explained that these included think tanks and organizations involved in South Korea’s diplomacy and foreign policy-related operations.
Regarding whether the attacks were actually successful, he said he would not go into detail about specifically what happened, but confirmed that the attackers were successful.
However, he did not disclose the targets whose attacks succeeded or the specific scope of the compromise, adding that the relevant facts had been notified to the concerned institutions and others.
Previously, CrowdStrike revealed in research findings published on the 4th of this month that Vapor Panda had attacked more than 200 organizations worldwide using “Artext,” an AI agent-based penetration testing tool.
The attacks, carried out between October 17 and 29, targeted diplomatic and security agencies in Latin America, Europe, Asia, and the Middle East, as well as U.S. companies in the pharmaceutical, semiconductor, defense, nuclear, and financial sectors.
It was found that Vapor Panda operated Artext centered around the Chinese AI model “DeepSeek,” and was also able to utilize paid subscription services of other cutting-edge AI models as auxiliary means.
Artext is a tool that repeatedly scans for vulnerabilities in web applications and application programming interfaces (APIs), and utilizes the data and results secured at each stage to execute subsequent attacks.
Meyers explained that by automating tasks previously performed manually by humans, the barrier to entry for attacks has been lowered, adding that Vapor Panda is a highly sophisticated hacking group and that they caught an opportunity to utilize this on a large scale targeting hundreds of targets.
CrowdStrike has been tracking Vapor Panda since 2021 and has authored more than 50 related reports.
It has identified the group’s activities by comprehensively analyzing attack infrastructure, malware, web shells, and targets.
This attack is separate from the attacks observed last month targeting the domestic financial sector using Artext.
Meyers reported that over the past month and a half, cases of different hacking groups using Artext have been appearing frequently.
The possibility of reduced attack costs following the adoption of AI was also raised.
An analysis suggests that about 1.4 billion input tokens were consumed in a single penetration process by Vapor Panda, and they likely significantly reduced attack costs by running language models in their own inference environment.
Meyers emphasized that AI-based hacking is not an entirely new attack, but rather an issue of speed and scale, and that such attacks can be detected and defended against if appropriate security technology and visibility are secured along with 24/7 threat hunting.
(Photo: Yonhap News)
※ Please note: This article was translated by AI and may contain errors.
Click Here For The Original Source.
——————————————————–
