Chinese Hackers Deploy AI Agents in ConnectWise Breach, Target 2,250 Organizations in Hours

Reviewed byNidhi Govil

9 Sources

Share

A China-aligned hacking group breached ConnectWise's cloud backup service and deployed AI agents to attack over 2,250 organizations at machine speed. Microsoft says Storm-0940 used the ARTEX AI penetration testing framework to conduct unauthorized sign-in attempts at 2,900 per minute, gaining privileged access in under three hours at multiple firms.

News article

Chinese Hackers Exploit ConnectWise Breach to Target Thousands

A ConnectWise breach has exposed thousands of organizations to cyber espionage after Chinese hackers infiltrated the IT management giant's cloud backup service and weaponized stolen credentials against its customers. Microsoft revealed that Storm-0940, a China-aligned hacking group active since 2021, compromised ConnectWise's multi-tenant application and used the access to launch a massive AI cyberattack targeting more than 2,250 Microsoft 365 tenants

1

2

. The ConnectWise supply chain attack demonstrates how threat actors are leveraging AI to compress attack timelines and overwhelm security defenses.

ARTEX AI Agents Accelerate Attack at Machine Speed

The Chinese hackers deployed ARTEX AI agents, an open-source AI penetration testing framework, to automate and accelerate their operations at what Microsoft describes as machine speed. During the attack window between October 10 and October 15, 2026, the ARTEX-powered campaign conducted unauthorized sign-in attempts at a staggering rate of 2,900 attempts per minute

1

3

. Microsoft recorded over 151,000 failed login events during password spraying operations alone, with the activity peaking on October 11 at more than 111,000 failures in a single day

2

4

. The AI agents operated around the clock with minimal human interaction, scanning through tenants in sorted order just seconds apart, indicating a script working from a target list.

Privileged Access Achieved in Under Three Hours

The speed and efficiency of the AI cyberattack allowed Storm-0940 to progress from initial targeting to privileged access in less than three hours at multiple organizations

2

3

. After gaining control of ConnectWise application identities, the hackers authenticated to customer environments using stolen app credentials that required no interactive user sign-in. On October 15, security firm Huntress observed an app add itself to the top administrator role within one tenant, planting new secrets set to expire in 2028

2

. Microsoft confirmed unauthorized sign-ins for 162 tenants by October 16, which jumped to almost 1,000 Entra tenants with sign-in activity the following day, with more than 300 succeeding

2

. The pace of the attack gave defenders minimal time to detect and respond to the intrusion.

Aerospace, Defense, and Semiconductors Among Targeted Sectors

The ConnectWise supply chain attack targeted organizations across 14 industries, with a focus on aerospace and defense contractors, semiconductor manufacturers, government agencies, financial services, energy firms, and think tanks

4

5

. Storm-0940 conducted reconnaissance and attempted to modify privileges through Microsoft Graph and directory operations, including application credential changes that could enable persistent access even after the original compromised application was remediated

2

4

. The AI agents progressed beyond authentication to execute changes involving identities, groups, and privileged roles, establishing backdoors that could remain active even after cleanup efforts. ConnectWise CEO Manny Rivelo confirmed the company suspended its Cloud Backup, SaaS Security, and Cloud Management services and began a phased restoration expected to take several days

3

.

ARTEX Framework Closes After Abuse in Multiple Attacks

The ARTEX AI penetration testing framework, developed by a user known as "Autumn-27," was previously linked to breaches at South Korean financial institutions where it was allegedly paired with Anthropic's Claude Code

2

. Following public reporting of its abuse, the developer announced on GitHub that the ARTEX project would no longer be updated and would become closed source

2

. Microsoft warns that threat actors across multiple geographic regions have weaponized the framework, which can be configured with any AI model to automate reconnaissance, password spraying, OAuth abuse, and data exfiltration

4

5

. The Chinese Embassy in Washington did not respond to requests for comment, while Beijing routinely denies conducting cyber-enabled data theft

5

. Organizations should review activity logs from October, keep affected app connections disabled until ConnectWise issues formal restoration instructions, and watch for signs of persistent backdoors that could outlast immediate remediation efforts.

© 2026 TheOutpost.AI All rights reserved