AI-driven hackers breach thousands of corporate accounts via software flaw

A significant cyber intrusion is currently unfolding as a China-aligned hacking group deploys autonomous AI agents to compromise corporate networks through a supply chain vulnerability. The operation, identified by Microsoft as originating from the actor Storm-0940, utilized an open-source framework to execute a massive attack on ConnectWise, a multi-tenant application service. This incident marks a notable shift in attack methodology, relying heavily on automated systems to bypass traditional security defenses at a speed that challenges human response times.

The core of the operation involved the use of ARTEX, an agentic AI penetration testing framework designed to run autonomous tests with minimal human oversight. According to Microsoft, the attackers gained unauthorized access to ConnectWise and used stolen application credentials to authenticate as apps rather than individual users. This allowed them to attempt sign-ins at more than 2,250 Microsoft 365 tenants. In several cases, the AI agents reached privileged access levels within less than three hours of initial targeting, a timeline that leaves defenders with very little time to detect and respond to the intrusion.

The campaign also included a high-volume password spraying effort aimed at 327 accounts across eight organizations. This resulted in more than 151,000 failed logons with no successful breaches via that specific method. At its peak on October 11, the attack generated 2,900 attempts per minute, with over 111,000 failures recorded that day. While almost 98% of these attempts were locked out, the AI agents successfully signed into customer tenants using four application credentials, executing over 200 sign-ins in under four minutes from the same IP range.

The targeted industries included aerospace and defense, semiconductor, government, and think tanks, with 14 specific sectors listed. The activity did not stem from a vulnerability in Microsoft’s own systems but rather from the compromise of the third-party application. Security firm Huntress confirmed unauthorized sign-ins for 162 tenants by October 16, a number that rose to nearly 1,000 Entra tenants by October 17, with more than 300 succeeding. In one instance, an app added itself to the top administrator role and planted new secrets set to expire in 2028.

This attack differs from traditional methods due to the use of ARTEX, which can utilize various AI models. Recent reports have linked similar technology to breaches at South Korean financial institutions, with the suspect allegedly using Anthropic’s Claude Code. The developer of ARTEX, known as Autumn-27, has stated the project will no longer be updated and will become closed source. It remains uncertain how the initial ConnectWise credentials were stolen or exactly how many tenants were fully compromised.

ConnectWise has suspended its Cloud Backup, SaaS Security, and Cloud Management services as a precautionary measure. A phased restoration process has begun, with recovery expected to take several days. The company has not yet found 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 possible.

The rapid pace of the attack, sweeping through tenants in sorted order with seconds between actions, suggests a script working from a pre-defined target list. As AI-driven threats become more prevalent, this incident highlights the growing challenge for organizations to maintain security in an environment where automated agents can operate around the clock with little need for human interaction.