Hackers use AI agent framework ARTEX to attack security company Barracuda, breaching 10 cloud accounts in 44 hours and harvesting credentials
駭客利用AI代理框架ARTEX攻擊資安公司Barracuda,44小時入侵10個雲端帳戶搜刮憑證
Security company Barracuda disclosed that, according to its investigation, attackers used the AI penetration testing framework ARTEX to breach 10 of its cloud accounts within 44 hours, collecting credentials and attempting to establish backdoors. The entry point was a cloud access key Barracuda had provided to an AI company for a project and had not yet revoked; the AI company's development environment was breached in October 2026, and attackers are suspected of obtaining the key from it, first using it in the early hours of November 1 UTC, reading stored AWS access keys and Azure service principal secrets, then penetrating data analytics systems and Kubernetes clusters, and pushing short-lived branches to code repositories to trigger build processes to steal secrets; they attempted to create new cloud administrator identities but were unsuccessful. Barracuda discovered anomalous activity on November 4; although it found no indication that XDR customer data was directly accessed, incomplete event records meant it could not rule out the possibility, so it rotated all potentially affected credentials and notified some customers to revoke and rebuild API integration credentials.