AI Agents Hit Supply Chain Targets

Corporate IT security leaders faced a new layer of automated threat as a China-aligned group used artificial intelligence to accelerate a supply chain attack. The incident highlights the growing financial and operational risks associated with agentic AI frameworks in modern cyber operations.

Microsoft identified the actor, designated Storm-0940, as utilizing ARTEX, an open-source penetration testing system driven by large language models. This tool allowed the group to conduct autonomous reconnaissance and password guessing with minimal human oversight. The speed of these operations left defenders with limited windows to detect and respond to the intrusion attempts.

The campaign targeted the multi-tenant platform ConnectWise, exploiting unauthorized access to the application. Using stolen credentials that authenticated without interactive user sign-ins, the attackers attempted logins at more than 2,250 Microsoft 365 tenants. In some cases, privileged access was achieved within three hours of initial targeting. The attack focused on high-value sectors including aerospace, defense, semiconductors, and government entities.

Automated password spraying resulted in over 151,000 failed logons across 327 accounts in eight organizations. At its peak on October 11, the system generated 2,900 attempts per minute. Although nearly 98 percent of these attempts were locked out, the use of valid application credentials allowed the agents to sign in from specific IP ranges. More than 200 successful sign-ins occurred in under four minutes using just four sets of application credentials.

Security firm Huntress confirmed unauthorized sign-ins for 162 tenants by October 16. By the following day, almost 1,000 Entra tenants showed sign-in activity from the attacking IP, with over 300 succeeding. In one instance, an application added itself to the top administrator role and planted new secrets set to expire in 2028. The sequential nature of the attacks suggested a script operating from a predefined target list.

ConnectWise suspended several services as a precaution and began a phased restoration process expected to take several days. The company reported no indicators that partner backup data was compromised. Microsoft advised customers to keep affected app connections disabled and to review recent activity logs, noting that successful sign-ins do not automatically mean data was accessed or exfiltrated.

The ARTEX framework, which can operate with any AI model, is being used by threat actors in multiple regions. Recent reports linked similar technology to breaches at South Korean financial institutions. The developer of ARTEX, known as Autumn-27, announced on GitHub that the project will become closed source and will no longer receive updates.