On November 6, 2026, Microsoft Threat Intelligence reported that Storm-3358, which it describes as a Cambodia-based intrusion cluster that mainly targets financial organizations in Southeast Asia, began using the ARTEX agentic AI offensive framework in October 2026. A smaller number of large financial organizations elsewhere were also targeted.
Microsoft says the group combined ARTEX with conventional tooling to scan internet-facing systems, exploit web application flaws, install remote access tools and harvest credentials, and that its indicators were still active on the day of publication. It notes that agentic scanning has a recognizable rhythm - many varied requests in a short time, repeated retries of one technique, and machine-speed moves to the next stage - and that deception such as honeypots, decoys and “fake wins” reduced attack traffic reaching real systems in its testing.
Why it matters: it is the third major vendor in two weeks to document ARTEX in live attacks, after CrowdStrike’s espionage case and Check Point’s cybercrime survey, showing one open-source agent framework spreading across both state-linked and criminal groups.
What it does not show: Microsoft gives no victim count or confirmed losses, and the post doubles as product guidance for its own security tools. Its indicator list also contains one IP address written two different ways.