Storm-0940 weaponises Artex AI tool to breach cloud accounts

Artex, an open-source, multi-agent AI penetration-testing framework first published to GitHub in July and written in Go, has been repurposed by threat actors to automate credential attacks and cloud breaches at machine speed. While intended as a legitimate tool, forks and variants of Artex—along with about 45 similar agentic testing frameworks—are being used to brute-force passwords, dump databases, enumerate and abuse APIs, and conduct end-to-end attacks across multiple industries. The framework’s AI agent capability allows attackers to chain disparate steps with minimal hacking skill, significantly accelerating operation speed.

The best-documented incidents involve Storm-0940, a Chinese-leaning actor, which adapted Artex into its offence. At the end of October the group launched a large password-spraying and credential-stuffing campaign against more than 300 user accounts at eight organisations, generating about 151,000 authentication requests (averaging around 2,900 attempts per minute) before the effort largely failed.

When that faltered, the group pivoted to target ConnectWise’s internal infrastructure, exfiltrating trusted OAuth credentials for ConnectWise Cloud Backup and compromising more than 2,250 downstream Microsoft 365 and Azure environments within three hours of initial targeting. Microsoft and others have noted continued use of Artex by diverse actors to scan, exploit web apps, abuse credentials, and leverage cloud identities for persistence and data exfiltration.

Crowdsourced telemetry from Censys showed more than 1,400 internet-exposed Artex consoles as of early November, with researchers cautioning that actual global use likely exceeds that figure due to non-exposed deployments.