A sophisticated, China-linked threat actor has been caught leveraging an open-source artificial intelligence penetration testing framework known as Artex to automate a sweeping series of cyberattacks targeting organizations worldwide. According to recent threat intelligence reports published by Microsoft and other major cybersecurity firms, the campaign heavily incorporates automated, machine-speed tactics to compromise ConnectWise environments and exploit hundreds of downstream organizations across multiple critical sectors.
The threat group, actively tracked by intelligence analysts under the moniker Storm-0940, has maintained an active operational footprint since at least 2021. Historically, this advanced persistent threat (APT) collective has relied on traditional, labor-intensive techniques such as password spraying and the exploitation of vulnerable, internet-facing applications to infiltrate enterprises spanning a wide variety of industries. However, recent telemetry indicates a significant evolution in their operational capabilities through the integration of agentic AI frameworks.
Beginning in early September 2026, security researchers observed the threat actor abusing the Artex framework to dramatically scale up and automate its offensive operations. Prior investigative reporting by prominent cybersecurity organizations including CrowdStrike, Zscaler, and Recorded Future had already successfully linked this emerging AI-powered malicious activity to a wave of sophisticated intrusions targeting financial institutions in South Korea.
The threat actor’s methodology took a broader and more disruptive turn in mid-October 2026, when Microsoft reported that Storm-0940 deployed Artex to orchestrate automated intrusions into the environments of ConnectWise customers. The scale and speed of the breach prompted swift defensive measures, leading ConnectWise to temporarily suspend several core cloud services, including Cloud Backup, SaaS Security, and Cloud Management, in an effort to contain the fallout and protect its broader customer base.
During the height of the mid-October surge, cybersecurity firm Huntress observed threat actors successfully utilizing compromised ConnectWise application credentials to gain unauthorized access to hundreds of Microsoft 365 customer tenants, including more than 300 Entra tenants. Subsequent analyses by Microsoft revealed that the threat group weaponized Artex to dramatically accelerate foundational attack phases, including target reconnaissance, password spraying, and malicious OAuth application abuse. At its peak intensity, the automated infrastructure hammered more than 300 target accounts across eight distinct organizations at an astonishing pace exceeding 2,900 distinct login attempts per minute.
Following the initial unauthorized access, Microsoft telemetry captured evidence of Storm-0940 using Artex to attempt sign-ins across more than 2,250 separate corporate tenants. While the manipulation and abuse of OAuth applications has long been a standard component of the Storm-0940 operational toolset, cybersecurity analysts emphasize that the integration of the Artex framework elevated this behavior to unprecedented operational speeds. Instead of relying on manual operator inputs, the threat group allowed automated agents to drive the campaign at machine speed.
In total, Microsoft captured over 151,000 failed logon events directly associated with the aggressive automated campaign. Alongside these thousands of rejections, however, the threat actor secured successful authentication into multiple customer environments by hijacking the legitimate identities of several ConnectWise cloud applications.
Moving beyond simple credential theft, the automated AI agents operating on behalf of the threat actor proceeded to interact directly with Microsoft Graph and execute low-level directory operations. According to Microsoft’s technical breakdown, this post-exploitation activity included extensive environmental reconnaissance, the modification of application credentials to establish long-term persistence, and unauthorized changes involving core identities, security groups, and highly privileged administrative roles.
By systematically altering application credentials within compromised environments, the threat actors ensured they could maintain persistent, backdoor access to victim networks even after administrators managed to remediate the initially affected applications. In light of these findings, Microsoft has issued urgent technical advisories urging enterprise security teams to closely examine service principal sign-ins. Organizations are advised to thoroughly cross-reference these sign-ins with subsequent Microsoft Graph activity, directory alterations, and resource access patterns, carefully correlating any suspicious anomalies with authorized administrative actions.
The geographic and industrial scope of the campaign is exceptionally wide. Microsoft reports that since early September, Storm-0940 has utilized the Artex framework to execute automated attacks against hundreds of organizations operating across a diverse array of industries. Impacted sectors include aerospace, automotive, chemicals, defense, energy, financial services, government, legal, manufacturing, media, military, semiconductors, software development, and general technology. Non-profit organizations and policy think tanks have also been caught in the crosshairs of the automated campaigns.
Providing further technical clarity on the threat group’s operations, Zscaler published a comprehensive deep-dive analysis into Storm-0940’s AI-assisted hacking methodologies. The report outlines specific operational details regarding how the threat actor configures and deploys the Artex framework, including synchronization scripts, historical Claude Code prompt data, and various auxiliary network routing tools used to obscure their digital footprint.

