AI-Driven Intrusion: Storm-0940 Exploits ConnectWise for Rapid Tenant Compromise

China-state-aligned hackers have deployed autonomous artificial intelligence agents to execute a massive supply chain attack, attempting sign-ins to over 2,250 Microsoft 365 tenants through compromised ConnectWise credentials. According to Microsoft’s recent security analysis, the group, identified as Storm-0940, utilized a stolen application identity to authenticate into customer environments without requiring interactive user logins. This method allowed the attackers to move with unprecedented speed, with some operations reaching privileged access levels less than three hours after the initial targeting began.

The attack relied on ARTEX, an open-source framework that drives large language model agents to perform autonomous penetration testing. This technology accelerated critical phases of the intrusion, including reconnaissance, password spraying, and the abuse of OAuth applications. By operating continuously with minimal human intervention, the AI agents created a tempo that left defenders with little time to detect anomalies or respond effectively. While the exact method of credential theft remains unclear, the automated nature of the assault significantly outpaced traditional manual hacking techniques.

In addition to the application-based intrusion, the group conducted extensive automated password guessing campaigns. These efforts targeted 327 accounts across eight organizations, resulting in more than 151,000 failed login attempts with no successes. The peak of this activity occurred on October 11, when the system executed 2,900 attempts per minute. Although nearly 98% of these activities were locked out, the sheer volume demonstrated the scalability of the AI-driven approach. The attackers focused on high-value sectors, including aerospace, defense, semiconductors, government, and think tanks.

Once inside the tenants, the AI agents performed a series of sophisticated actions. These included directory operations, reconnaissance, and modifications to application credentials, identities, and groups. In one specific case on October 15, security firm Huntress observed an application granting itself the top administrator role. The attackers also planted new secrets with expiration dates set for 2028, indicating an intent to maintain long-term access. The sequential nature of the sign-ins, occurring just seconds apart from the same IP range, suggested the use of a script working from a pre-defined target list.

The use of ARTEX marks a shift in threat landscape, as the framework can leverage various AI models. Other regional actors have reportedly used similar technology, with recent breaches at South Korean financial institutions linked to the use of ARTEX. The developer of the framework, known as Autumn-27, announced on GitHub that the project will cease updates and transition to a closed-source model. Meanwhile, ConnectWise has suspended several cloud services as a precaution and is in the process of a phased restoration, which is expected to take several days.

Microsoft advises affected customers to keep compromised application connections disabled until formal restoration instructions are issued. The company emphasizes that a successful sign-in does not automatically guarantee that data was accessed or exfiltrated. Organizations are encouraged to review their activity logs from this month to identify any unauthorized changes. As recovery efforts continue, the incident highlights the growing speed and complexity of AI-assisted cyberattacks against enterprise supply chains.

The attack did not stem from a vulnerability in Microsoft’s own systems or products, but rather from the compromise of the third-party ConnectWise application. This distinction is crucial for understanding the vector of intrusion. As the phased restoration of ConnectWise services progresses, experts are monitoring for further signs of compromise. The incident serves as a clear signal to IT leaders that traditional defense mechanisms may need to adapt to the rapid, autonomous pace of modern AI-driven threats.