☰ SectionsWhat’s happening across America.
Tech · AI

2,250: The scale of the AI-driven Microsoft breach

Hackers aligned with China used an AI agent framework to target thousands of Microsoft 365 tenants via a third-party application.

By Sofia Ramirez ·

The figure that defines this incident is 2,250. That is the number of Microsoft 365 tenants for which sign-in attempts were made in a large-scale supply chain attack. The operation, attributed to a China-state-aligned actor named Storm-0940, did not target Microsoft's own systems. Instead, it exploited stolen credentials from a ConnectWise application to access customer environments. In some instances, attackers moved from initial targeting to privileged access in under three hours.

The campaign unfolded over several days in mid-October. According to Microsoft's security blog, the group first gained unauthorized access to a multi-tenant application. It then attempted sign-ins at more than 2,250 Microsoft 365 tenants using credentials that authenticated as the applications themselves, without requiring a user to log in. Microsoft's systems were not compromised. The activity focused on industries including aerospace and defense, semiconductors, government, and think tanks among 14 total sectors.

A notable feature of the attack was its use of ARTEX, an open-source agentic AI penetration testing framework. Microsoft described ARTEX as a system of LLM-driven agents that run an autonomous penetration test. This technology was used to automate reconnaissance, scanning, and password spraying. The agents can operate continuously with little human interaction, a pace that leaves defenders with little time to detect and respond. It remains uncertain how the initial ConnectWise credentials were stolen or the full number of tenants actually compromised.

The password spraying component of the campaign was extensive. It targeted 327 accounts across eight organizations, generating more than 151,000 failed logons with no successful breaches. At its peak on October 11, the attack reached 2,900 attempts per minute, all of which failed. Almost 98% of this activity was subsequently locked out.

Other security firms also observed malicious activity. On October 15, Huntress saw an application in one tenant add itself to the top administrator role. By October 17, the firm had observed sign-in activity from the attacking IP in almost 1,000 Entra tenants, with more than 300 succeeding. The pattern of access, moving through tenants in sorted order a few seconds apart, indicated a script working from a target list.

ConnectWise has suspended several services, including Cloud Backup, as a precaution. A phased restoration is underway and is expected to take several days. The company has 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 logs from this month where available.

The use of this technology in attacks is a growing concern. Public reporting this month tied the framework to breaches at South Korean financial institutions. ARTEX's developer, who goes by the handle “Autumn-27,” said on GitHub that the project will no longer be updated and would become closed source, Reuters reported.

Source: Tom's Hardware

More from AI