AI-Driven Intrusion: Over 2,250 Tenants Targeted in Supply Chain Breach

More than 2,250 Microsoft 365 tenants faced unauthorized sign-in attempts in a sophisticated supply chain attack that leveraged artificial intelligence to accelerate intrusion. The incident, attributed to a China-state-aligned actor known as Storm-0940, utilized stolen credentials from ConnectWise applications to authenticate directly to customer environments without interactive user logins. This scale of automated probing suggests a shift in how threat actors exploit multi-tenant software vulnerabilities, raising questions about the resilience of current identity management frameworks against machine-speed attacks.

The core of the operation relied on ARTEX, an open-source agentic AI penetration testing framework. By deploying LLM-driven agents that operate autonomously, the group significantly expedited scanning, reconnaissance, and password spraying. Microsoft noted that these agents function with minimal human intervention, creating a tempo that leaves defenders with little time to detect anomalies. The use of such autonomous tools marks a departure from traditional manual exploitation, where human speed often serves as a natural limit on the rate of attack.

Despite the heavy automated pressure, the initial password spraying campaign resulted in no successful logins. The attackers targeted 327 accounts across eight organizations, generating over 151,000 failed attempts. At its peak on October 11, the campaign registered 2,900 attempts per minute, with 48 accounts suffering more than 1,000 failures each. Nearly 98% of this activity triggered lockouts, indicating that while the volume was high, the brute-force approach to individual account passwords largely failed to gain a foothold.

Success came through the abuse of OAuth application credentials rather than individual user passwords. With control of multiple ConnectWise cloud app identities, the AI agents signed into customer tenants from a specific IP range. In one rapid sequence, over 200 sign-ins were executed in under four minutes using just four application credentials. Within some tenants, the attackers performed Graph and directory operations, modified app credentials, and altered privileged roles, demonstrating a swift escalation from initial access to significant control.

The speed of the compromise was stark in specific cases. Some attacks reached privileged access levels less than three hours after initial targeting. On October 15, security firm Huntress observed an application adding itself to the top administrator role in one tenant, triggering Microsoft Defender alerts. The attackers also planted new secrets set to expire in 2028, a tactic designed to maintain long-term persistence. By October 16, Huntress confirmed unauthorized sign-ins for 162 tenants, a number that grew to nearly 1,000 Entra tenants with activity from the attacking IP by the following day.

Contextual analysis suggests the attackers moved through tenants in a sorted order, with actions separated by only seconds, indicating a script working from a predefined target list. The group, active since 2021, targeted sectors including aerospace, defense, semiconductors, government, and think tanks. Microsoft clarified that the breach did not stem from a vulnerability in its own systems or services, but rather from the compromise of third-party application credentials. It remains uncertain how the initial ConnectWise credentials were stolen or the full extent of the actual data compromised.

The incident highlights the growing reliance on AI in cyber operations. ARTEX’s developer, identified as “Autumn-27,” announced on GitHub that the project would cease updates and become closed source following public reporting linking it to breaches at South Korean financial institutions. CrowdStrike indicated that the suspect in those separate attacks was likely based in China and used ARTEX with Anthropic’s Claude Code. As ConnectWise begins a phased restoration of suspended services, Microsoft advises customers to keep affected app connections disabled and review recent activity logs to assess potential exposure.