On November 13, 2026, the FBI released a FLASH advisory (FLASH-20261113-001) warning that attackers are using ARTEX, an AI-driven penetration testing framework, to automate intrusions against organizations worldwide. Microsoft observed several Storm-0940 intrusions in which the group gained privileged access to a target’s cloud environment in under three hours. For defenders, those cases raise a practical question. Which applications and vendor connections could give an attacker access to sensitive data?
What frameworks mean in a pentest
Security teams use testing guides and software tools to plan and carry out a penetration test. The OWASP Web Security Testing Guide, for example, helps testers choose checks that fit the application. Tools such as Burp Suite help testers examine how a web application handles requests and verify potential flaws.
The same capabilities that make these tools useful to testers make them useful to attackers. Criminal groups have also used Cobalt Strike, a platform built for red teams. ARTEX adds another capability. Its agents can use the results of one test to decide what to check next.
What ARTEX is
According to the FBI, ARTEX is an open-source, multi-agent platform written in Go. An operator describes an objective in natural language, and specialized AI agents handle each stage of the attack, from mapping the target and finding weaknesses to exploiting them, gaining higher privileges, and pulling out data, with little need for a human operator. ARTEX connects to an external language model chosen by the operator.
A developer using the name Autumn-27 published ARTEX on GitHub in July 2026 and took the original repository offline on October 8 after reports of abuse. Copies continued to circulate. Check Point Research counted 1,608 publicly accessible ARTEX instances during its November study.
Where it has been used
The FBI says attackers have used ARTEX across multiple industries since at least August 2026. Three cases show the different kinds of access those attackers found.
- ConnectWise Cloud Backup (October 2026): Microsoft assesses that Storm-0940, a group aligned with the Chinese state, used ARTEX and compromised ConnectWise application credentials to sign in to customers’ Microsoft 365 environments. Microsoft recorded sign-in attempts across more than 2,250 tenants. By October 16, Huntress had confirmed unauthorized sign-ins in 162 tenants it monitored.
- South Korean financial institutions: Zscaler reported that attackers targeted customer inquiry portals and employee-facing mobile backends at South Korean financial firms, stealing personal information belonging to about 67,000 people.
- Barracuda (disclosed November 10): In its November 11 technical account, Barracuda said an attacker used a cloud key that remained valid after a vendor project ended. The company believes the attacker obtained the key from the vendor’s compromised development environment. Barracuda says the activity matches ARTEX but has not identified who was behind it.
A pentest limited to the main website could miss both the customer inquiry portal and the access a former vendor’s key still provides.
What to do now
Use the following checks to review your exposure.
- Inventory internet-facing applications, including partner portals, mobile backends, and legacy services. Record who owns each one and when it was last tested.
- Verify that portal and API users can access only the records their role permits. Check that unusually high volumes of lookups trigger an alert.
- Restrict cloud administration and publishing interfaces to trusted networks or dedicated administrative access paths.
- Review application permissions and consent grants, and remove access that the application no longer needs.
- Scan code, pipelines, and configurations for embedded secrets.
- Rotate exposed credentials and revoke vendor keys when a project ends.
If your organization uses or previously authorized an affected ConnectWise application, identify it by application ID and review its sign-in and audit history using the ConnectWise guidance.
How Thoropass tests your attack surface
At Thoropass, we use AI-assisted tools similar to ARTEX to help you find and fix vulnerabilities before an attacker can exploit them. Our pentesters use automation to identify potential weaknesses in the applications, APIs, and integrations included in your test. We manually verify what access those weaknesses allow, explain the impact, and give your team clear steps to fix them.
If your last pentest focused on your main application, talk to the Thoropass Pentest Team about bringing partner portals, APIs, and vendor integrations into scope.










.png)



