ConnectWise — China-Linked Threat Actors Use ARTEX AI Agents in ConnectWise Supply Chain Attack

Publication date: October 22, 2026
Category: Supply Chain Attack / Cloud Security

Introduction

Recent cybersecurity investigations have revealed that a China state-aligned threat group tracked as Storm-0940 executed a sophisticated supply chain compromise targeting the ConnectWise Cloud Backup application. During this campaign, the adversaries integrated ARTEX, a Go-based agentic AI penetration testing framework, to automate and accelerate reconnaissance, mass scanning, password spraying, and OAuth application abuse. The employment of AI capabilities at “machine speed” allowed the threat actors to breach enterprise and government environments globally without requiring constant human interaction.

What is Storm-0940 and the ARTEX Framework? (General Analysis)

Storm-0940 is an advanced persistent threat (APT) group active since at least 2021, previously associated with leveraging botnets like Quad7 for evasive campaigns and mass authentication. The core element of this new incident is ARTEX, an offensive agentic AI framework that originally operated as closed-source software until its malicious abuse against South Korean financial institutions prompted its restriction.

By deploying autonomous agents, the framework removes standard human operational limitations (such as fatigue or time constraints), executing multiple breach paths simultaneously. From a threat taxonomy standpoint, this incident combines logical vulnerabilities in cloud identity management with advanced supply chain abuse techniques. As no formal CVE identifier has been assigned at this specific time, it is reasonably estimated that the associated vectors correspond to access control flaws and OAuth credential management issues, carrying a Critical severity impact (Estimated CVSS v3.1: 9.0 / Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H and CWE-284: Improper Access Control).

How Does It Work? (Technical Analysis)

The technical mechanism of the attack is structured into several automated phases operating at high speed:

  • Initial password spraying and scanning phase: Operating from malicious infrastructure (such as IP ranges like 144.172.64[.]99), ARTEX’s AI agents executed automated brute-force and spraying attacks at massive scale. Microsoft recorded over 151,000 failed logon events over a four-day window, peaking at 2,900 failures per minute.
  • OAuth application compromise and identity spoofing: After bypassing or compromising traditional authentication mechanisms, the agents gained access to the identities of multiple ConnectWise cloud applications. This allowed the threat actors to authenticate to customer environments using legitimate application credentials that already possessed an established trust relationship within the tenant.
  • Lateral movement and cloud-based persistence: Once inside the target environments (including Microsoft 365 and Exchange Online), the autonomous agents interacted directly with Microsoft Graph and directory operations. This included modifying credentials on existing OAuth applications, manipulating identities, and reassigning privileged roles, establishing persistent backdoors that remain active even after remediating the original compromised application.

Affected Systems / Environments

The attack directly impacted the ConnectWise Cloud Backup infrastructure and thousands of downstream customer tenants globally. Impacted sectors span a broad range of critical industries, including:

  • Aerospace and defense, automotive, energy, and financial services.
  • Information technology, software development, and Managed Service Providers (MSPs).
  • Governments, military entities, legal firms, think tanks, and non-profit organizations.
  • Corporate infrastructures integrating centralized authentication via Microsoft 365 and Graph API.

Mitigation and Detection

Remediation

  • OAuth token and credential revocation: Immediately audit all integrated third-party applications and revoke OAuth credentials or permissions exhibiting anomalous activity or unauthorized modifications.
  • Access restriction and multi-factor authentication (MFA): Implement strict phishing-resistant MFA policies across all accounts with administrative and directory privileges.
  • Vendor updates and patching: Promptly apply bulletins and restoration guidelines issued by ConnectWise and affected cloud service providers.

Detection

  • Mass authentication log monitoring: Configure alerts for unusual spikes in failed login attempts (password spraying) originating from suspicious IP addresses or unusual ranges.
  • Directory and Graph API change auditing: Track modifications to OAuth application credentials and changes to privileged roles performed outside authorized maintenance windows.

“Agents are not constrained by factors affecting humans – they can work around the clock, on weekends, and on holidays and attempt multiple breach paths simultaneously, enabling significantly reduced time from selecting a target to achieving objectives.”

Wrapping Up

The ConnectWise incident highlights a dangerous evolution in cybercrime and cyberespionage tactics: the integration of agentic AI into offensive supply chain workflows. By exploiting the intrinsic trust granted to integrated third-party applications at “machine speed,” groups like Storm-0940 can compromise complex infrastructures within hours. This scenario demands that both Managed Service Providers (MSPs) and consumer organizations strengthen their defensive posture through strict identity visibility, proactive API monitoring, and rigorous cloud privilege management.

References