Tech and AI

Microsoft says a state-aligned group used AI agents to try more than 2,250 Microsoft 365 tenants after the ConnectWise breach

By Staff Writer  |  23 October 2026

The group, which Microsoft tracks as Storm-0940, used an open-source AI penetration testing tool and in some cases went from first targeting to privileged cloud access in under three hours.

Microsoft said on Wednesday that a China-state aligned group it tracks as Storm-0940 was responsible for the supply chain attack on ConnectWise, the IT management software company, and that the group used AI agents to speed up its attempts to break into the Microsoft 365 environments of ConnectWise's customers.

Using stolen credentials belonging to applications connected to ConnectWise's Cloud Backup service, the attackers tried to sign in to more than 2,250 tenants, the separate cloud environments that each customer organisation runs, at up to 2,900 attempts a minute. Those applications had already been granted access by customers, so no person had to log in. In several cases the time from first targeting to privileged access to an organisation's cloud systems was less than three hours, Microsoft said.

An open-source tool put to work

The tool was ARTEX, an open-source penetration testing system that coordinates a team of AI agents and can run on any AI model. Microsoft described it as a set of agents that "run an autonomous penetration test from reconnaissance through data exfiltration with little needed human interaction". Software of this kind is written so that security teams can attack their own systems with permission. Microsoft said it had seen threat actors from several regions using it, and that Storm-0940 had been using it since as early as September.

Before turning to ConnectWise, the group ran a password spraying campaign between 10 and 14 October against 327 accounts at eight organisations. Microsoft recorded more than 151,000 failed sign-ins, more than 111,000 of them on 11 October alone, and no successful log-in. Infrastructure used in that campaign matched addresses later published by ConnectWise.

Since early October, a wave of AI-driven cyberattacks has been ongoing,

Igor Tsyganskiy, Executive Vice President, Microsoft

In the environments the agents did reach, Microsoft saw them move on to reconnaissance through Microsoft Graph, change application credentials and alter identities, groups and privileged roles. Changes of that kind can leave a back door open after the original stolen credentials have been revoked. Microsoft has not said how many customers were compromised, and it said that a successful sign-in does not by itself show that data was accessed or taken.

Services coming back in stages

ConnectWise took its Cloud Backup, SaaS Security and Cloud Management services offline at the weekend and began restoring them in phases on Wednesday, telling customers that recovery would take several days. Microsoft's advice is to keep the affected connections disabled until ConnectWise issues restoration instructions, to preserve sign-in and audit records before deleting anything, and to review activity from at least 15 October.

Storm-0940 has been active since 2021, and Microsoft lists government, defence, finance, energy, manufacturing, legal and technology organisations among those it targeted from September.

None of the methods was new: stolen credentials, password spraying and the misuse of applications that customers already trust. What the agents changed was the pace, and with it the time defenders had to notice.