AI Speed: Hackers Breach 2,250 Tenants via ConnectWise

More than 2,250 Microsoft 365 tenants faced attempted sign-ins from a large-scale cyberattack this week. The operation targeted the ConnectWise multi-tenant application, a critical piece of infrastructure for many businesses. Microsoft Security detailed the incident in a recent blog post. The attackers used stolen credentials to authenticate as the apps themselves. This bypassed the need for interactive user logins.

The group behind the intrusion is known as Storm-0940. Microsoft describes the actor as aligned with the Chinese state. The group has been active since 2021 but has no other known aliases. Its name follows the convention for a developing threat actor. The attack relied on ARTEX, an open-source framework for agentic AI penetration testing. This system uses large language model-driven agents to run autonomous tests. It accelerated scanning, reconnaissance, and the abuse of OAuth applications.

The use of AI allowed the agents to work around the clock with minimal human interaction. This pace left defenders with little time to detect and respond. At its peak on October 11, the automated password spraying campaign hit 2,900 attempts per minute. That day saw over 111,000 failed logons. Almost 98% of this activity resulted in account lockouts. However, the attackers also targeted 327 accounts across eight organizations with over 151,000 failed logons, with no successes recorded in that specific spray.

Despite the high failure rate in password guessing, the attackers gained significant ground through application credentials. Using four specific application credentials, they executed more than 200 sign-ins in under four minutes. These sign-ins came from the same IP range. In some cases, privileged access was reached less than three hours after the initial targeting began. The sweep through tenants occurred in sorted order, just seconds apart. This pattern suggested a script was working from a predefined target list.

The affected industries include aerospace and defense, semiconductors, government, and think tanks. Fourteen industries were listed in total. Microsoft clarified that the activity did not stem from a compromise or vulnerability in its own systems. Instead, it exploited unauthorized access to ConnectWise. In one tenant, Huntress, a security firm, observed an app adding itself to the top administrator role. This behavior triggered a Microsoft Defender alert for self-assigned Global Administrator roles. The attackers also planted new secrets set to expire in 2028.

By October 16, Huntress confirmed unauthorized sign-ins for 162 tenants. The following day, nearly 1,000 Entra tenants showed sign-in activity from the attacking IP. More than 300 of these sign-ins succeeded. The use of ARTEX marks a shift from traditional attack methods. Threat actors in multiple regions are now using this technology. Public reporting this month linked it to breaches at South Korean financial institutions. CrowdStrike noted that the suspect in those separate attacks was likely based in China and used ARTEX with Anthropic’s Claude Code.

ARTEX’s developer, known online as “Autumn-27,” announced the project would no longer be updated. The framework will become closed source, according to Reuters. ConnectWise acknowledged the unauthorized access and suspended several services as a precaution. These include Cloud Backup, SaaS Security, and Cloud Management. A phased restoration has begun, with recovery expected to take several days. There are no current indicators that partner backup data was compromised.

Microsoft advises customers to keep affected app connections disabled until formal restoration instructions are issued. A successful sign-in or affected app does not automatically mean data was accessed or taken. Organizations should review activity from this month where logs allow. The incident highlights the speed and scale possible when AI agents are integrated into supply chain attacks. Defenders now face a new challenge in monitoring for these rapid, automated intrusions.