Listen to the article
The open-source framework has been used in attacks across multiple sectors, with investigators reporting exposed personal data and attempts to reach thousands of downstream organisations.
The FBI says attackers have been using ARTEX, an open-source artificial intelligence framework, to automate stages of cyber intrusions, from reconnaissance to data theft. In an advisory issued on 13 November 2026, the bureau said the tool had featured in campaigns against organisations in multiple sectors worldwide since at least August. In some cases, attackers moved from choosing a target to gaining privileged access in less than three hours. (No citations in body! that parenthetical is not okay. Remove.)
ARTEX uses specialised software agents, guided by large language models, to carry out and link technical tasks. Google’s Threat Intelligence Group and Mandiant say they have observed it being used for activities including reconnaissance, privilege escalation and movement through compromised networks. In some investigations, attackers supplied stolen organisational credentials and directed the framework to continue the intrusion.
The FBI advisory also describes attacks on South Korean financial institutions in which personal information belonging to about 67,000 people was exposed. CrowdStrike’s investigation independently documents suspected ARTEX-related activity against financial organisations in South Korea from late September into early October 2026. The FBI further cites a ConnectWise-related supply-chain incident involving attempts to reach more than 2,250 downstream tenants. Microsoft’s account cautions that attempted access does not, by itself, establish that customer data was reached or stolen.
Although ARTEX’s original developer removed the project’s GitHub repository in October, copies and modified versions remain in circulation, according to the FBI and Google’s threat researchers. Palo Alto Networks’ Unit 42 says its October 2026 scan found hundreds of ARTEX hosts exposed to the internet, illustrating how widely the tool had spread beyond its original repository. The end of development has therefore not eliminated the risk; related tools and new versions may continue to support similar operations.
The FBI advises organisations to investigate the indicators of compromise in its advisory and strengthen protection around internet-facing services, cloud infrastructure, self-service portals and partner systems. Its recommendations include reviewing unusual account and permission changes, finding and replacing exposed secrets in code and deployment workflows, and watching for excessive or automated queries. Microsoft also urges organisations to examine application permissions and sign-in activity, investigate unusual Microsoft Graph activity, and preserve identity and audit records. Google’s researchers recommend limiting privileges, segmenting networks and using short-lived credentials to reduce the opportunities available to attackers.
Source Reference Map
Inspired by headline at: [1]
Sources by paragraph:
Source: Fuse Wire Services


