AI Agents in Supply Chain Attack: Rapid Breaches and New Threats

Businesses relying on third-party management tools now face a faster and more automated threat. A recent large-scale attack leveraged artificial intelligence to probe for vulnerabilities in a widely used platform, allowing intruders to test credentials across thousands of organizations. For IT teams, this means that traditional detection windows are shrinking, and the pace of compromise can outstrip manual response capabilities.

The first key takeaway is the scale of the initial reconnaissance. The group, identified as Storm-0940, used a framework called ARTEX to attempt sign-ins at more than 2,250 Microsoft 365 tenants. They utilized stolen credentials from the ConnectWise application to authenticate without requiring interactive user logins. This allowed the AI agents to scan and operate with minimal human oversight, significantly accelerating the process of identifying weak points in security configurations.

Second, the speed of the intrusion was notable. In some instances, the attackers reached privileged access levels in less than three hours after initially targeting a tenant. The automated password spraying campaign targeted 327 accounts across eight organizations, resulting in over 151,000 failed logon attempts. While almost 98% of these attempts were eventually locked out, the volume of traffic created a dense fog that made detection difficult for defenders.

Third, the impact extended to critical sectors. The attack targeted industries including aerospace, defense, semiconductors, government, and think tanks. In one specific case on October 15, a security firm observed an application adding itself to the top administrator role. The attackers also planted new secrets set to expire in 2028, suggesting an intent to maintain long-term access. By October 16, unauthorized sign-ins had been confirmed for 162 tenants, rising to nearly 1,000 by the following day.

Contextually, this incident highlights a shift in threat actor methodology. Storm-0940 has been active since 2021 and is aligned with a Chinese state. The use of ARTEX, an open-source framework that drives large language models to perform autonomous penetration tests, marks a departure from traditional manual attacks. The developers of ARTEX have since announced the project will become closed source, potentially altering how other groups access this technology.

Looking ahead, the outlook requires vigilance from both vendors and customers. ConnectWise has suspended several of its services as a precaution and is working on a phased restoration over the coming days. Microsoft advises customers to keep affected app connections disabled until formal restoration instructions are issued. While not every sign-in indicates data theft, the incident underscores the need to review activity logs from this month and monitor for unusual application behavior.

The event also connects to broader trends in cyber espionage. Public reports this month have linked similar AI-driven tools to breaches at South Korean financial institutions. As autonomous agents become more common, the ability to work around the clock with little human interaction poses a persistent challenge for security teams who must detect and respond to threats at machine speed.