Hackers Use AI Agents To Break Into Microsoft Accounts

Security researchers are tracking a large-scale digital break-in that relies on artificial intelligence to speed up the process of getting inside protected systems.

Microsoft reported that a China-aligned hacking group, tracked as Storm-0940, is using an open-source tool called ARTEX to conduct a supply chain attack involving ConnectWise software. This attack allowed the group to attempt sign-ins to more than 2,250 Microsoft 365 tenants using stolen application credentials. According to the report, some of the attacks achieved privileged access in less than three hours from the initial targeting.

The group has been active since 2021 and is classified as a China-state aligned actor. The targets in this campaign include government agencies, think tanks, and companies in the aerospace, defense, and semiconductor sectors. Microsoft clarified that the activity did not result from a compromise of or a vulnerability in its own systems, products, or services.

The ARTEX framework is described as a system of LLM-driven agents that run an autonomous penetration test. Because the agents can operate around the clock with little human interaction, they can accelerate scanning, reconnaissance, and password spraying. Microsoft noted that this pace leaves defenders with little time to detect and respond. It is not known how the ConnectWise credentials were stolen or exactly how many tenants were compromised.

The campaign involved automated password guessing, also known as password spraying, across 327 accounts at eight different organizations. This effort generated more than 151,000 failed logins but no successes. On October 11, the number of attempts reached 2,900 per minute, totaling over 111,000 for that day. Almost 98% of this activity was locked out.

Despite the failed passwords, the attackers used credentials belonging to ConnectWise cloud apps to sign in to customer tenants. The agents were able to perform operations inside some tenants, including reconnaissance, changes to application credentials, and modifications to identities, groups, and privileged roles. In one case, a security firm observed an application adding itself to a top administrator role and planting new secrets set to expire in 2028.

By October 16, a security firm confirmed unauthorized sign-ins for 162 tenants. A day later, activity from the attacking IP address was seen across almost 1,000 Entra tenants, with more than 300 succeeding. The pattern of accessing tenants in sorted order, a few seconds apart, suggested a script working from a target list.

This use of an open-source pentest framework differs from many traditional attacks. Threat actors in multiple regions are using similar technology, and public reporting this month tied it to breaches at South Korean financial institutions. The developer of ARTEX has stated that the project will no longer be updated and will become closed source.

In response to the unauthorized access, ConnectWise suspended Cloud Backup, SaaS Security, and Cloud Management as a precaution. The company has started a phased restoration with recovery expected to take several days. ConnectWise reported no indicators to date that partner backup data was compromised. Microsoft advises customers to keep affected app connections disabled until formal restoration instructions are issued.