Hackers Used AI Tools to Target Thousands of Business Accounts

A coordinated cyber campaign has exposed the growing speed of modern intrusions, with automated tools allowing attackers to move through corporate networks in mere hours.

The scale of the operation was significant, involving attempts to sign in to more than 2,250 Microsoft 365 tenants. In a separate effort to crack passwords, the group generated over 151,000 failed logon attempts across 327 accounts in eight different organizations. At the height of the activity on October 11, the password spraying reached a peak of 2,900 attempts per minute, resulting in more than 111,000 failures that day. While almost 98% of these automated guesses were locked out, the attackers successfully used stolen credentials from four specific applications to make over 200 sign-ins in less than four minutes.

The entity behind the intrusion is identified as Storm-0940, a group active since 2021 and described as a China-state aligned actor. The group has no known industry aliases and follows a naming convention typical of developing threat actors. Their targets were specific, focusing on industries such as aerospace and defense, semiconductors, government agencies, and think tanks, with 14 sectors listed in total. The activity did not stem from a vulnerability in Microsoft's own systems but rather from unauthorized access to the ConnectWise multi-tenant application.

The attackers utilized a tool called ARTEX, an open-source framework that uses large language model driven agents to run autonomous penetration tests. This technology allowed the group to accelerate scanning, reconnaissance, and the abuse of OAuth applications. The agents worked around the clock with minimal human interaction, creating a pace that left defenders with little time to detect and respond. It remains uncertain how the initial ConnectWise credentials were stolen or the exact number of tenants that were fully compromised.

The speed of the attack was a defining characteristic. Some of the group's operations reached privileged access in under three hours after initial targeting. Within specific tenants, the agents performed directory operations, conducted reconnaissance, and altered application credentials. They also changed identities, groups, and privileged roles. In one instance on October 15, a security firm named Huntress observed an application adding itself to the top administrator role. The attackers also planted new secrets set to expire in 2028, indicating a long-term presence.

The pattern of the sweep suggested a scripted approach, as the agents moved through tenants in sorted order at intervals of only a few seconds. By October 16, Huntress confirmed unauthorized sign-ins for 162 tenants. The following day, the count rose to nearly 1,000 Entra tenants with sign-in activity from the attacking IP, with more than 300 of those attempts succeeding. This rapid progression highlighted the efficiency of the automated framework.

The use of ARTEX distinguishes this campaign from many traditional attacks. Microsoft noted that the framework can utilize any AI model, and threat actors in various regions are adopting the technology. Public reports this month linked the tool to breaches at South Korean financial institutions. CrowdStrike suggested that the suspect in those separate attacks was likely based in China and may have used ARTEX with Anthropic's Claude Code. The developer of ARTEX, known by the handle Autumn-27, announced on GitHub that the project would no longer be updated and would become closed source.

ConnectWise responded to the unauthorized access by suspending its Cloud Backup, SaaS Security, and Cloud Management services as a precaution. The company has begun a phased restoration process, with full recovery expected to take several days. To date, there are no indicators that partner backup data was compromised. Microsoft advises customers to keep affected app connections disabled until formal restoration instructions are issued and to review activity from this month where logs permit.

The incident underscores the shifting landscape of cyber threats, where automation reduces the time between intrusion and privilege escalation. For organizations, the stakes are high, as the ability of AI agents to operate autonomously demands faster detection and response strategies.