Storm-0940 Exploits AI Agents in ConnectWise Breach Targeting 2,250 Tenants
A China-aligned threat group used an open-source AI framework to rapidly probe thousands of Microsoft 365 environments via a supply chain compromise in ConnectWise.
By Adrian Romano ·

The speed at which hackers can now move through corporate networks has crossed a new threshold, leaving security teams with barely enough time to even notice an intrusion. This week, Microsoft detailed a massive supply chain attack where a group of China-state-aligned hackers leveraged artificial intelligence agents to sweep through more than 2,250 Microsoft 365 tenants. The operation was fast, automated, and largely undetected until it was too late, highlighting a growing tension between traditional defense mechanisms and autonomous offensive tools.
The primary culprit behind this disruption is a threat actor known as Storm-0940. According to Microsoft’s security blog, this group has been active since 2021 and carries no other known aliases in the industry. They utilized a tool called ARTEX, an open-source framework that drives large language model agents to conduct autonomous penetration testing. By using ARTEX, the attackers accelerated the entire process of scanning, reconnaissance, and credential abuse, allowing them to operate around the clock with minimal human intervention.
The attack vector centered on unauthorized access to ConnectWise, a multi-tenant application. Once inside, the group used stolen credentials to authenticate as applications rather than individual users, bypassing standard interactive sign-in checks. This allowed the AI agents to sign into customer tenants from a specific IP range, executing over 200 sign-ins in less than four minutes using just four sets of application credentials. In some cases, the attackers reached privileged access levels in under three hours from the initial targeting.
Beyond the direct application abuse, the group engaged in aggressive password spraying. They targeted 327 accounts across eight organizations, generating more than 151,000 failed logon attempts. On the peak day of the operation, October 11, the spray reached 2,900 attempts per minute. While nearly 98 percent of these attempts were locked out, the sheer volume demonstrated the capability of the AI agents to overwhelm systems with low-effort, high-volume guessing.
The targets for this campaign were specific, focusing on high-value sectors such as aerospace and defense, semiconductors, government agencies, and think tanks. Microsoft clarified that the attack did not stem from a vulnerability in its own systems or products, but rather from the compromised third-party application. The sorted order of the tenant sweeps, occurring just seconds apart, suggested the attackers were working from a pre-defined target list, further emphasizing the automated nature of the operation.
Security firm Huntress observed specific malicious behavior in one tenant on October 15, where an application added itself to the top administrator role. This action triggered a Microsoft Defender alert regarding an application self-assigning a Global Administrator role. The attackers also planted new secrets set to expire in 2028, a tactic designed to maintain long-term access. By October 16, Huntress confirmed unauthorized sign-ins for 162 tenants, a number that climbed to nearly 1,000 Entra tenants by the following day, with more than 300 successful logins recorded.
The use of ARTEX marks a shift in how threat actors operate. The framework is not limited to a single AI model, and public reporting this month has linked similar tools to breaches at South Korean financial institutions. In those separate incidents, the suspect was likely based in China and used ARTEX alongside Anthropic’s Claude Code. In response to the increased scrutiny, the developer of ARTEX, known by the handle Autumn-27, announced on GitHub that the project would cease updates and transition to closed-source status.
ConnectWise has taken precautionary steps to manage the fallout, suspending its Cloud Backup, SaaS Security, and Cloud Management services. A phased restoration process has begun, with recovery expected to take several days. The company has not yet found indicators that partner backup data was compromised. Microsoft advises customers to keep affected app connections disabled until formal instructions are issued and to review their logs for activity from this month to assess potential impact.
While an affected app or a successful sign-in does not automatically mean data was stolen, the incident underscores the urgency of monitoring application identities. As AI agents continue to evolve, the window for detection shrinks, forcing organizations to rethink how they secure their supply chains against autonomous threats.






