2,250 Tenants Targeted: AI Agents Accelerate Supply Chain Intrusion

Microsoft disclosed this week that a China-aligned threat actor named Storm-0940 deployed an open-source AI framework to execute a sophisticated supply chain attack against ConnectWise. The operation resulted in unauthorized sign-in attempts across more than 2,250 Microsoft 365 tenants. The most alarming aspect of the incident is the speed with which the attackers operated. In some cases, the group secured privileged access to customer environments in under three hours after initial targeting, a timeframe that significantly outpaces traditional human-led intrusion methods.

The scale of the automated assault is reflected in the sheer volume of attempts. The attackers targeted 327 accounts across eight organizations, generating more than 151,000 failed logons through password spraying. Peak activity occurred on October 11, when the system executed 2,900 attempts per minute, totaling over 111,000 for that day. Despite 48 accounts logging more than 1,000 failures each, nearly 98% of this activity was locked out. However, the attackers pivoted to abusing stolen ConnectWise application credentials, achieving more than 200 successful sign-ins in under four minutes using just four application identities.

The group responsible, Storm-0940, has been active since 2021 and is described as a developing actor with no known industry aliases. Their targets included high-value sectors such as aerospace and defense, semiconductors, government bodies, and think tanks, with 14 industries listed in total. Microsoft clarified that the attack did not stem from a vulnerability in their own systems or products, but rather from the unauthorized use of third-party application credentials within the ConnectWise ecosystem.

The operational efficiency of the attack is attributed to the use of ARTEX, an agentic AI penetration testing framework. This system utilizes large language model driven agents to perform autonomous scanning, reconnaissance, and credential abuse with minimal human interaction. The continuous nature of these agents leaves defenders with very little time to detect and respond. Security firm Huntress observed specific malicious behavior, including an application adding itself to the top administrator role in one tenant on October 15. By October 17, almost 1,000 Entra tenants showed sign-in activity from the attacking IP range, with over 300 of those attempts succeeding.

This incident marks a shift in threat landscape dynamics. While ARTEX is open-source and can utilize various AI models, its application in high-profile breaches, including those at South Korean financial institutions, highlights its growing utility for threat actors. The developer of ARTEX, known by the handle Autumn-27, announced on GitHub that the project would stop receiving updates and transition to closed source. Meanwhile, ConnectWise has suspended several services as a precaution and begun a phased restoration process, expecting recovery to take several days. Customers are advised to keep affected app connections disabled until formal instructions are issued and to review their activity logs for signs of compromise from this month.