ConnectWise Halts Services After AI-Driven Supply Chain Intrusion

ConnectWise has suspended several of its core cloud services as a precautionary measure following a sophisticated supply chain attack that leveraged autonomous AI agents to breach thousands of enterprise networks. The incident, detailed in a recent Microsoft Security blog post, highlights a significant shift in the operational tempo of cyber threats, with automated systems executing complex penetration tests at a speed that traditional defensive measures struggle to match.

Microsoft identified the perpetrator as Storm-0940, a China-state-aligned actor active since 2021. The group utilized ARTEX, an open-source agentic AI penetration testing framework, to automate scanning, reconnaissance, and credential abuse. By exploiting stolen ConnectWise application credentials, the attackers authenticated as applications rather than individual users, allowing them to bypass interactive sign-in controls. This method enabled the group to attempt sign-ins across more than 2,250 Microsoft 365 tenants, with some privileged access roles secured in under three hours of initial targeting.

The scale of the automated password spraying campaign was substantial, targeting 327 accounts across eight organizations with over 151,000 failed logon attempts. At its peak on October 11, the system generated 2,900 attempts per minute, resulting in more than 111,000 failures that day. While 98% of these attempts were locked out, the successful exploitation of application credentials allowed the AI agents to sign in to customer tenants from specific IP ranges. In one instance, more than 200 sign-ins were recorded in under four minutes using just four stolen application credentials.

The targeted industries included aerospace and defense, semiconductor, government, and think tanks, with 14 specific sectors listed in the report. Security firm Huntress observed advanced persistence techniques, such as an application adding itself to the top administrator role in one tenant on October 15. This behavior triggered specific alerts for self-assigned Global Administrator roles. By October 16, Huntress confirmed unauthorized sign-ins for 162 tenants, a number that jumped to nearly 1,000 Entra tenants by the following day, with over 300 successful logins from the attacking IP addresses.

The use of ARTEX represents a notable evolution in attack methodology, as the framework uses LLM-driven agents to run autonomous tests with minimal human interaction. Microsoft noted that this pace leaves defenders with little time to detect and respond. The framework’s flexibility, which allows it to use any AI model, has drawn attention from other threat actors in different regions. Public reporting this month has linked similar ARTEX-based techniques to breaches at South Korean financial institutions, where suspects allegedly used the framework in conjunction with Anthropic’s Claude Code.

In response to the growing threat landscape, the developer of ARTEX, known by the handle Autumn-27, announced on GitHub that the project would cease updates and transition to a closed-source model. Meanwhile, ConnectWise has begun a phased restoration of its Cloud Backup, SaaS Security, and Cloud Management services, with full recovery expected to take several days. The company stated it has no current indicators that partner backup data was compromised, though it has advised customers to keep affected app connections disabled until formal restoration instructions are issued.

Microsoft clarified that the attack did not stem from a vulnerability in its own systems or products. However, the incident underscores the financial and operational risks associated with third-party application identities. For enterprise clients, the primary takeaway is the need to review activity logs from this month and monitor for anomalous behavior, as a successful sign-in does not automatically equate to data exfiltration but signals a significant breach of the perimeter.