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

The landscape of digital defense is shifting as threat actors integrate autonomous artificial intelligence into large-scale intrusion campaigns, reducing the time available for human responders to detect and contain breaches. This tactical evolution was starkly demonstrated in a recent supply chain attack that targeted thousands of organizations via a compromised multi-tenant application.

Microsoft identified the culprit as Storm-0940, a China-state-aligned actor active since 2021, which utilized ARTEX, an open-source framework designed for agentic penetration testing. By leveraging stolen credentials from ConnectWise applications, the group executed unauthorized sign-ins across more than 2,250 Microsoft 365 tenants. The use of automated agents allowed the attackers to authenticate as applications without interactive user sign-ins, streamlining the process of scanning, reconnaissance, and OAuth abuse.

The speed of the operation is the most critical metric for defenders. In several instances, the AI-driven agents achieved privileged access within three hours of the initial targeting. This rapid progression contrasts sharply with traditional attack timelines, leaving security teams with minimal windows for detection. The automation enabled the agents to work continuously with little human interaction, a pace that significantly compresses the response cycle for defenders.

Alongside the targeted application abuse, the group conducted extensive password spraying campaigns. These automated guessing attempts targeted 327 accounts across eight organizations, resulting in over 151,000 failed logons with no successful logins. At its peak on October 11, the campaign generated 2,900 attempts per minute, with nearly 98% of the activity resulting in account lockouts. Despite the high failure rate, the volume of attempts highlighted the scale of the automated assault.

The efficiency of the AI agents was evident in their ability to navigate tenant environments. Using four specific application credentials, the attackers executed more than 200 sign-ins in under four minutes from the same IP range. Within these tenants, the agents performed directory operations, changed app credentials, and modified identities and privileged roles. This systematic approach indicated a script working from a pre-defined target list, sweeping through tenants in sorted order at intervals of just a few seconds.

The incident highlights a broader trend in threat actor methodology. While the ConnectWise breach did not stem from a vulnerability in Microsoft’s systems, it underscores the growing reliance on open-source AI tools that can be adapted by various groups. Public reporting this month linked similar AI-driven techniques to breaches at South Korean financial institutions, suggesting a regional preference for this technology. The developer of ARTEX, known as Autumn-27, announced the project would become closed source, potentially signaling a maturation or consolidation of the tooling available to attackers.

ConnectWise responded by suspending several cloud services as a precautionary measure. The company has initiated a phased restoration process, with recovery expected to take several days, and currently reports no indicators that partner backup data was compromised. However, the uncertainty surrounding the extent of data access remains a key concern for affected organizations.

For the Southeast and beyond, this event serves as a warning that the integration of AI in cyberattacks is no longer theoretical. Security firms like Huntress confirmed unauthorized sign-ins for nearly 1,000 tenants, with over 300 successful connections. Microsoft advises customers to keep affected app connections disabled until formal restoration instructions are issued, emphasizing that a successful sign-in does not guarantee data exfiltration, but the potential for privilege escalation requires immediate attention.